1 CHANGES - changes for libpng
   2 
   3 version 0.1 [March 29, 1995]
   4   initial work-in-progress release
   5 
   6 version 0.2 [April 1, 1995]
   7   added reader into png.h
   8   fixed small problems in stub file
   9 
  10 version 0.3 [April 8, 1995]
  11   added pull reader
  12   split up pngwrite.c to several files
  13   added pnglib.txt
  14   added example.c
  15   cleaned up writer, adding a few new transformations
  16   fixed some bugs in writer
  17   interfaced with zlib 0.5
  18   added K&R support
  19   added check for 64 KB blocks for 16 bit machines
  20 
  21 version 0.4 [April 26, 1995]
  22   cleaned up code and commented code
  23   simplified time handling into png_time
  24   created png_color_16 and png_color_8 to handle color needs
  25   cleaned up color type defines
  26   fixed various bugs
  27   made various names more consistent
  28   interfaced with zlib 0.71
  29   cleaned up zTXt reader and writer (using zlib's Reset functions)
  30   split transformations into pngrtran.c and pngwtran.c
  31 
  32 version 0.5 [April 30, 1995]
  33   interfaced with zlib 0.8
  34   fixed many reading and writing bugs
  35   saved using 3 spaces instead of tabs
  36 
  37 version 0.6 [May 1, 1995]
  38   first beta release
  39   added png_large_malloc() and png_large_free()
  40   added png_size_t
  41   cleaned up some compiler warnings
  42   added png_start_read_image()
  43 
  44 version 0.7 [June 24, 1995]
  45   cleaned up lots of bugs
  46   finished dithering and other stuff
  47   added test program
  48   changed name from pnglib to libpng
  49 
  50 version 0.71 [June 26, 1995]
  51   changed pngtest.png for zlib 0.93
  52   fixed error in libpng.txt and example.c
  53 
  54 version 0.8 [August 20, 1995]
  55   cleaned up some bugs
  56   added png_set_filler()
  57   split up pngstub.c into pngmem.c, pngio.c, and pngerror.c
  58   added #define's to remove unwanted code
  59   moved png_info_init() to png.c
  60   added old_size into png_realloc()
  61   added functions to manually set filtering and compression info
  62   changed compression parameters based on image type
  63   optimized filter selection code
  64   added version info
  65   changed external functions passing floats to doubles (k&r problems?)
  66   put all the configurable stuff in pngconf.h
  67   enabled png_set_shift to work with paletted images on read
  68   added png_read_update_info() - updates info structure with transformations
  69 
  70 Version 0.81 [August, 1995]
  71   incorporated Tim Wegner's medium model code (thanks, Tim)
  72 
  73 Version 0.82 [September, 1995]
  74   [unspecified changes]
  75 
  76 Version 0.85 [December, 1995]
  77   added more medium model code (almost everything's a far)
  78   added i/o, error, and memory callback functions
  79   fixed some bugs (16-bit, 4-bit interlaced, etc.)
  80   added first run progressive reader (barely tested)
  81 
  82 Version 0.86 [January, 1996]
  83   fixed bugs
  84   improved documentation
  85 
  86 Version 0.87 [January, 1996]
  87   fixed medium model bugs
  88   fixed other bugs introduced in 0.85 and 0.86
  89   added some minor documentation
  90 
  91 Version 0.88 [January, 1996]
  92   fixed progressive bugs
  93   replaced tabs with spaces
  94   cleaned up documentation
  95   added callbacks for read/write and warning/error functions
  96 
  97 Version 0.89 [June 5, 1996]
  98   Added new initialization API to make libpng work better with shared libs
  99     we now have png_create_read_struct(), png_create_write_struct(),
 100     png_create_info_struct(), png_destroy_read_struct(), and
 101     png_destroy_write_struct() instead of the separate calls to
 102     malloc and png_read_init(), png_info_init(), and png_write_init()
 103   Changed warning/error callback functions to fix bug - this means you
 104     should use the new initialization API if you were using the old
 105     png_set_message_fn() calls, and that the old API no longer exists
 106     so that people are aware that they need to change their code
 107   Changed filter selection API to allow selection of multiple filters
 108     since it didn't work in previous versions of libpng anyways
 109   Optimized filter selection code
 110   Fixed png_set_background() to allow using an arbitrary RGB color for
 111     paletted images
 112   Fixed gamma and background correction for paletted images, so
 113     png_correct_palette is not needed unless you are correcting an
 114     external palette (you will need to #define PNG_CORRECT_PALETTE_SUPPORTED
 115     in pngconf.h) - if nobody uses this, it may disappear in the future.
 116   Fixed bug with Borland 64K memory allocation (Alexander Lehmann)
 117   Fixed bug in interlace handling (Smarasderagd, I think)
 118   Added more error checking for writing and image to reduce invalid files
 119   Separated read and write functions so that they won't both be linked
 120     into a binary when only reading or writing functionality is used
 121   New pngtest image also has interlacing and zTXt
 122   Updated documentation to reflect new API
 123 
 124 Version 0.89c [June 17, 1996]
 125   Bug fixes.
 126 
 127 Version 0.90 [January, 1997]
 128   Made CRC errors/warnings on critical and ancillary chunks configurable
 129   libpng will use the zlib CRC routines by (compile-time) default
 130   Changed DOS small/medium model memory support - needs zlib 1.04 (Tim Wegner)
 131   Added external C++ wrapper statements to png.h (Gilles Dauphin)
 132   Allow PNG file to be read when some or all of file signature has already
 133     been read from the beginning of the stream.  ****This affects the size
 134     of info_struct and invalidates all programs that use a shared libpng****
 135   Fixed png_filler() declarations
 136   Fixed? background color conversions
 137   Fixed order of error function pointers to match documentation
 138   Current chunk name is now available in png_struct to reduce the number
 139     of nearly identical error messages (will simplify multi-lingual
 140     support when available)
 141   Try to get ready for unknown-chunk callback functions:
 142     - previously read critical chunks are flagged, so the chunk handling
 143       routines can determine if the chunk is in the right place
 144     - all chunk handling routines have the same prototypes, so we will
 145       be able to handle all chunks via a callback mechanism
 146   Try to fix Linux "setjmp" buffer size problems
 147   Removed png_large_malloc, png_large_free, and png_realloc functions.
 148 
 149 Version 0.95 [March, 1997]
 150   Fixed bug in pngwutil.c allocating "up_row" twice and "avg_row" never
 151   Fixed bug in PNG file signature compares when start != 0
 152   Changed parameter type of png_set_filler(...filler...) from png_byte
 153     to png_uint_32
 154   Added test for MACOS to ensure that both math.h and fp.h are not #included
 155   Added macros for libpng to be compiled as a Windows DLL (Andreas Kupries)
 156   Added "packswap" transformation, which changes the endianness of
 157     packed-pixel bytes (Kevin Bracey)
 158   Added "strip_alpha" transformation, which removes the alpha channel of
 159     input images without using it (not necessarily a good idea)
 160   Added "swap_alpha" transformation, which puts the alpha channel in front
 161     of the color bytes instead of after
 162   Removed all implicit variable tests which assume NULL == 0 (I think)
 163   Changed several variables to "png_size_t" to show 16/32-bit limitations
 164   Added new pCAL chunk read/write support
 165   Added experimental filter selection weighting (Greg Roelofs)
 166   Removed old png_set_rgbx() and png_set_xrgb() functions that have been
 167     obsolete for about 2 years now (use png_set_filler() instead)
 168   Added macros to read 16- and 32-bit ints directly from buffer, to be
 169     used only on those systems that support it (namely PowerPC and 680x0)
 170     With some testing, this may become the default for MACOS/PPC systems.
 171   Only calculate CRC on data if we are going to use it
 172   Added macros for zTXt compression type PNG_zTXt_COMPRESSION_???
 173   Added macros for simple libpng debugging output selectable at compile time
 174   Removed PNG_READ_END_MODE in progressive reader (Smarasderagd)
 175   More description of info_struct in libpng.txt and png.h
 176   More instructions in example.c
 177   More chunk types tested in pngtest.c
 178   Renamed pngrcb.c to pngset.c, and all png_read_<chunk> functions to be
 179     png_set_<chunk>.  We now have corresponding png_get_<chunk>
 180     functions in pngget.c to get information in info_ptr.  This isolates
 181     the application from the internal organization of png_info_struct
 182     (good for shared library implementations).
 183 
 184 Version 0.96 [May, 1997]
 185   Fixed serious bug with < 8bpp images introduced in 0.95
 186   Fixed 256-color transparency bug (Greg Roelofs)
 187   Fixed up documentation (Greg Roelofs, Laszlo Nyul)
 188   Fixed "error" in pngconf.h for Linux setjmp() behavior
 189   Fixed DOS medium model support (Tim Wegner)
 190   Fixed png_check_keyword() for case with error in static string text
 191   Added read of CRC after IEND chunk for embedded PNGs (Laszlo Nyul)
 192   Added typecasts to quiet compiler errors
 193   Added more debugging info
 194 
 195 Version 0.97 [January, 1998]
 196   Removed PNG_USE_OWN_CRC capability
 197   Relocated png_set_crc_action from pngrutil.c to pngrtran.c
 198   Fixed typecasts of "new_key", etc. (Andreas Dilger)
 199   Added RFC 1152 [sic] date support
 200   Fixed bug in gamma handling of 4-bit grayscale
 201   Added 2-bit grayscale gamma handling (Glenn R-P)
 202   Added more typecasts. 65536L becomes (png_uint_32)65536L, etc. (Glenn R-P)
 203   Minor corrections in libpng.txt
 204   Added simple sRGB support (Glenn R-P)
 205   Easier conditional compiling, e.g.,
 206     define PNG_READ/WRITE_NOT_FULLY_SUPPORTED;
 207     all configurable options can be selected from command-line instead
 208     of having to edit pngconf.h (Glenn R-P)
 209   Fixed memory leak in pngwrite.c (free info_ptr->text) (Glenn R-P)
 210   Added more conditions for png_do_background, to avoid changing
 211     black pixels to background when a background is supplied and
 212     no pixels are transparent
 213   Repaired PNG_NO_STDIO behavior
 214   Tested NODIV support and made it default behavior (Greg Roelofs)
 215   Added "-m" option and PNGTEST_DEBUG_MEMORY to pngtest (John Bowler)
 216   Regularized version numbering scheme and bumped shared-library major
 217     version number to 2 to avoid problems with libpng 0.89 apps
 218     (Greg Roelofs)
 219 
 220 Version 0.98 [January, 1998]
 221   Cleaned up some typos in libpng.txt and in code documentation
 222   Fixed memory leaks in pCAL chunk processing (Glenn R-P and John Bowler)
 223   Cosmetic change "display_gamma" to "screen_gamma" in pngrtran.c
 224   Changed recommendation about file_gamma for PC images to .51 from .45,
 225     in example.c and libpng.txt, added comments to distinguish between
 226     screen_gamma, viewing_gamma, and display_gamma.
 227   Changed all references to RFC1152 to read RFC1123 and changed the
 228     PNG_TIME_RFC1152_SUPPORTED macro to PNG_TIME_RFC1123_SUPPORTED
 229   Added png_invert_alpha capability (Glenn R-P -- suggestion by Jon Vincent)
 230   Changed srgb_intent from png_byte to int to avoid compiler bugs
 231 
 232 Version 0.99 [January 30, 1998]
 233   Free info_ptr->text instead of end_info_ptr->text in pngread.c (John Bowler)
 234   Fixed a longstanding "packswap" bug in pngtrans.c
 235   Fixed some inconsistencies in pngconf.h that prevented compiling with
 236     PNG_READ_GAMMA_SUPPORTED and PNG_READ_hIST_SUPPORTED undefined
 237   Fixed some typos and made other minor rearrangement of libpng.txt (Andreas)
 238   Changed recommendation about file_gamma for PC images to .50 from .51 in
 239     example.c and libpng.txt, and changed file_gamma for sRGB images to .45
 240   Added a number of functions to access information from the png structure
 241     png_get_image_height(), etc. (Glenn R-P, suggestion by Brad Pettit)
 242   Added TARGET_MACOS similar to zlib-1.0.8
 243   Define PNG_ALWAYS_EXTERN when __MWERKS__ && WIN32 are defined
 244   Added type casting to all png_malloc() function calls
 245 
 246 Version 0.99a [January 31, 1998]
 247   Added type casts and parentheses to all returns that return a value.(Tim W.)
 248 
 249 Version 0.99b [February 4, 1998]
 250   Added type cast png_uint_32 on malloc function calls where needed.
 251   Changed type of num_hist from png_uint_32 to int (same as num_palette).
 252   Added checks for rowbytes overflow, in case png_size_t is less than 32 bits.
 253   Renamed makefile.elf to makefile.lnx.
 254 
 255 Version 0.99c [February 7, 1998]
 256   More type casting.  Removed erroneous overflow test in pngmem.c.
 257   Added png_buffered_memcpy() and png_buffered_memset(), apply them to rowbytes.
 258   Added UNIX manual pages libpng.3 (incorporating libpng.txt) and  png.5.
 259 
 260 Version 0.99d [February 11, 1998]
 261   Renamed "far_to_near()" "png_far_to_near()"
 262   Revised libpng.3
 263   Version 99c "buffered" operations didn't work as intended.  Replaced them
 264     with png_memcpy_check() and png_memset_check().
 265   Added many "if (png_ptr == NULL) return" to quell compiler warnings about
 266     unused png_ptr, mostly in pngget.c and pngset.c.
 267   Check for overlength tRNS chunk present when indexed-color PLTE is read.
 268   Cleaned up spelling errors in libpng.3/libpng.txt
 269   Corrected a problem with png_get_tRNS() which returned undefined trans array
 270 
 271 Version 0.99e [February 28, 1998]
 272   Corrected png_get_tRNS() again.
 273   Add parentheses for easier reading of pngget.c, fixed "||" should be "&&".
 274   Touched up example.c to make more of it compileable, although the entire
 275     file still can't be compiled (Willem van Schaik)
 276   Fixed a bug in png_do_shift() (Bryan Tsai)
 277   Added a space in png.h prototype for png_write_chunk_start()
 278   Replaced pngtest.png with one created with zlib 1.1.1
 279   Changed pngtest to report PASS even when file size is different (Jean-loup G.)
 280   Corrected some logic errors in png_do_invert_alpha() (Chris Patterson)
 281 
 282 Version 0.99f [March 5, 1998]
 283   Corrected a bug in pngpread() introduced in version 99c (Kevin Bracey)
 284   Moved makefiles into a "scripts" directory, and added INSTALL instruction file
 285   Added makefile.os2 and pngos2.def (A. Zabolotny) and makefile.s2x (W. Sebok)
 286   Added pointers to "note on libpng versions" in makefile.lnx and README
 287   Added row callback feature when reading and writing nonprogressive rows
 288     and added a test of this feature in pngtest.c
 289   Added user transform callbacks, with test of the feature in pngtest.c
 290 
 291 Version 0.99g [March 6, 1998, morning]
 292   Minor changes to pngtest.c to suppress compiler warnings.
 293   Removed "beta" language from documentation.
 294 
 295 Version 0.99h [March 6, 1998, evening]
 296   Minor changes to previous minor changes to pngtest.c
 297   Changed PNG_READ_NOT_FULLY_SUPPORTED to PNG_READ_TRANSFORMS_NOT_SUPPORTED
 298     and added PNG_PROGRESSIVE_READ_NOT_SUPPORTED macro
 299   Added user transform capability
 300 
 301 Version 1.00 [March 7, 1998]
 302   Changed several typedefs in pngrutil.c
 303   Added makefile.wat (Pawel Mrochen), updated makefile.tc3 (Willem van Schaik)
 304   Replaced "while(1)" with "for(;;)"
 305   Added PNGARG() to prototypes in pngtest.c and removed some prototypes
 306   Updated some of the makefiles (Tom Lane)
 307   Changed some typedefs (s_start, etc.) in pngrutil.c
 308   Fixed dimensions of "short_months" array in pngwrite.c
 309   Replaced ansi2knr.c with the one from jpeg-v6
 310 
 311 Version 1.0.0 [March 8, 1998]
 312   Changed name from 1.00 to 1.0.0 (Adam Costello)
 313   Added smakefile.ppc (with SCOPTIONS.ppc) for Amiga PPC (Andreas Kleinert)
 314 
 315 Version 1.0.0a [March 9, 1998]
 316   Fixed three bugs in pngrtran.c to make gamma+background handling consistent
 317     (Greg Roelofs)
 318   Changed format of the PNG_LIBPNG_VER integer to xyyzz instead of xyz
 319     for major, minor, and bugfix releases.  This is 10001. (Adam Costello,
 320     Tom Lane)
 321   Make months range from 1-12 in png_convert_to_rfc1123
 322 
 323 Version 1.0.0b [March 13, 1998]
 324   Quieted compiler complaints about two empty "for" loops in pngrutil.c
 325   Minor changes to makefile.s2x
 326   Removed #ifdef/#endif around a png_free() in pngread.c
 327 
 328 Version 1.0.1 [March 14, 1998]
 329   Changed makefile.s2x to reduce security risk of using a relative pathname
 330   Fixed some typos in the documentation (Greg).
 331   Fixed a problem with value of "channels" returned by png_read_update_info()
 332 
 333 Version 1.0.1a [April 21, 1998]
 334   Optimized Paeth calculations by replacing abs() function calls with intrinsics
 335   plus other loop optimizations. Improves avg decoding speed by about 20%.
 336   Commented out i386istic "align" compiler flags in makefile.lnx.
 337   Reduced the default warning level in some makefiles, to make them consistent.
 338   Removed references to IJG and JPEG in the ansi2knr.c copyright statement.
 339   Fixed a bug in png_do_strip_filler with XXRRGGBB => RRGGBB transformation.
 340   Added grayscale and 16-bit capability to png_do_read_filler().
 341   Fixed a bug in pngset.c, introduced in version 0.99c, that sets rowbytes
 342     too large when writing an image with bit_depth < 8 (Bob Dellaca).
 343   Corrected some bugs in the experimental weighted filtering heuristics.
 344   Moved a misplaced pngrutil code block that truncates tRNS if it has more
 345     than num_palette entries -- test was done before num_palette was defined.
 346   Fixed a png_convert_to_rfc1123() bug that converts day 31 to 0 (Steve Eddins).
 347   Changed compiler flags in makefile.wat for better optimization
 348     (Pawel Mrochen).
 349 
 350 Version 1.0.1b [May 2, 1998]
 351   Relocated png_do_gray_to_rgb() within png_do_read_transformations() (Greg).
 352   Relocated the png_composite macros from pngrtran.c to png.h (Greg).
 353   Added makefile.sco (contributed by Mike Hopkirk).
 354   Fixed two bugs (missing definitions of "istop") introduced in libpng-1.0.1a.
 355   Fixed a bug in pngrtran.c that would set channels=5 under some circumstances.
 356   More work on the Paeth-filtering, achieving imperceptible speedup
 357     (A Kleinert).
 358   More work on loop optimization which may help when compiled with C++
 359     compilers.
 360   Added warnings when people try to use transforms they've defined out.
 361   Collapsed 4 "i" and "c" loops into single "i" loops in pngrtran and pngwtran.
 362   Revised paragraph about png_set_expand() in libpng.txt and libpng.3 (Greg)
 363 
 364 Version 1.0.1c [May 11, 1998]
 365   Fixed a bug in pngrtran.c (introduced in libpng-1.0.1a) where the masks for
 366     filler bytes should have been 0xff instead of 0xf.
 367   Added max_pixel_depth=32 in pngrutil.c when using FILLER with palette images.
 368   Moved PNG_WRITE_WEIGHTED_FILTER_SUPPORTED and PNG_WRITE_FLUSH_SUPPORTED
 369     out of the PNG_WRITE_TRANSFORMS_NOT_SUPPORTED block of pngconf.h
 370   Added "PNG_NO_WRITE_TRANSFORMS" etc., as alternatives for *_NOT_SUPPORTED,
 371     for consistency, in pngconf.h
 372   Added individual "ifndef PNG_NO_[CAPABILITY]" in pngconf.h to make it easier
 373     to remove unwanted capabilities via the compile line
 374   Made some corrections to grammar (which, it's) in documentation (Greg).
 375   Corrected example.c, use of row_pointers in png_write_image().
 376 
 377 Version 1.0.1d [May 24, 1998]
 378   Corrected several statements that used side effects illegally in pngrutil.c
 379     and pngtrans.c, that were introduced in version 1.0.1b
 380   Revised png_read_rows() to avoid repeated if-testing for NULL (A Kleinert)
 381   More corrections to example.c, use of row_pointers in png_write_image()
 382     and png_read_rows().
 383   Added pngdll.mak and pngdef.pas to scripts directory, contributed by
 384     Bob Dellaca, to make a png32bd.dll with Borland C++ 4.5
 385   Fixed error in example.c with png_set_text: num_text is 3, not 2 (Guido V.)
 386   Changed several loops from count-down to count-up, for consistency.
 387 
 388 Version 1.0.1e [June 6, 1998]
 389   Revised libpng.txt and libpng.3 description of png_set_read|write_fn(), and
 390     added warnings when people try to set png_read_fn and png_write_fn in
 391     the same structure.
 392   Added a test such that png_do_gamma will be done when num_trans==0
 393     for truecolor images that have defined a background.  This corrects an
 394     error that was introduced in libpng-0.90 that can cause gamma processing
 395     to be skipped.
 396   Added tests in png.h to include "trans" and "trans_values" in structures
 397     when PNG_READ_BACKGROUND_SUPPORTED or PNG_READ_EXPAND_SUPPORTED is defined.
 398   Add png_free(png_ptr->time_buffer) in png_destroy_read_struct()
 399   Moved png_convert_to_rfc_1123() from pngwrite.c to png.c
 400   Added capability for user-provided malloc_fn() and free_fn() functions,
 401     and revised pngtest.c to demonstrate their use, replacing the
 402     PNGTEST_DEBUG_MEM feature.
 403   Added makefile.w32, for Microsoft C++ 4.0 and later (Tim Wegner).
 404 
 405 Version 1.0.2 [June 14, 1998]
 406   Fixed two bugs in makefile.bor .
 407 
 408 Version 1.0.2a [December 30, 1998]
 409   Replaced and extended code that was removed from png_set_filler() in 1.0.1a.
 410   Fixed a bug in png_do_filler() that made it fail to write filler bytes in
 411     the left-most pixel of each row (Kevin Bracey).
 412   Changed "static pngcharp tIME_string" to "static char tIME_string[30]"
 413     in pngtest.c (Duncan Simpson).
 414   Fixed a bug in pngtest.c that caused pngtest to try to write a tIME chunk
 415     even when no tIME chunk was present in the source file.
 416   Fixed a problem in pngrutil.c: gray_to_rgb didn't always work with 16-bit.
 417   Fixed a problem in png_read_push_finish_row(), which would not skip some
 418     passes that it should skip, for images that are less than 3 pixels high.
 419   Interchanged the order of calls to png_do_swap() and png_do_shift()
 420     in pngwtran.c (John Cromer).
 421   Added #ifdef PNG_DEBUG/#endif surrounding use of PNG_DEBUG in png.h .
 422   Changed "bad adaptive filter type" from error to warning in pngrutil.c .
 423   Fixed a documentation error about default filtering with 8-bit indexed-color.
 424   Separated the PNG_NO_STDIO macro into PNG_NO_STDIO and PNG_NO_CONSOLE_IO
 425     (L. Peter Deutsch).
 426   Added png_set_rgb_to_gray() and png_get_rgb_to_gray_status() functions.
 427   Added png_get_copyright() and png_get_header_version() functions.
 428   Revised comments on png_set_progressive_read_fn() in libpng.txt and example.c
 429   Added information about debugging in libpng.txt and libpng.3 .
 430   Changed "ln -sf" to "ln -s -f" in makefile.s2x, makefile.lnx, and
 431     makefile.sco.
 432   Removed lines after Dynamic Dependencies" in makefile.aco .
 433   Revised makefile.dec to make a shared library (Jeremie Petit).
 434   Removed trailing blanks from all files.
 435 
 436 Version 1.0.2a [January 6, 1999]
 437   Removed misplaced #endif and #ifdef PNG_NO_EXTERN near the end of png.h
 438   Added "if" tests to silence complaints about unused png_ptr in png.h and png.c
 439   Changed "check_if_png" function in example.c to return true (nonzero) if PNG.
 440   Changed libpng.txt to demonstrate png_sig_cmp() instead of png_check_sig()
 441     which is obsolete.
 442 
 443 Version 1.0.3 [January 14, 1999]
 444   Added makefile.hux, for Hewlett Packard HPUX 10.20 and 11.00 (Jim Rice)
 445   Added a statement of Y2K compliance in png.h, libpng.3, and Y2KINFO.
 446 
 447 Version 1.0.3a [August 12, 1999]
 448   Added check for PNG_READ_INTERLACE_SUPPORTED in pngread.c; issue a warning
 449     if an attempt is made to read an interlaced image when it's not supported.
 450   Added check if png_ptr->trans is defined before freeing it in pngread.c
 451   Modified the Y2K statement to include versions back to version 0.71
 452   Fixed a bug in the check for valid IHDR bit_depth/color_types in pngrutil.c
 453   Modified makefile.wat (added -zp8 flag, ".symbolic", changed some comments)
 454   Replaced leading blanks with tab characters in makefile.hux
 455   Changed "dworkin.wustl.edu" to "ccrc.wustl.edu" in various documents.
 456   Changed (float)red and (float)green to (double)red, (double)green
 457     in png_set_rgb_to_gray() to avoid "promotion" problems in AIX.
 458   Fixed a bug in pngconf.h that omitted <stdio.h> when PNG_DEBUG==0 (K Bracey).
 459   Reformatted libpng.3 and libpngpf.3 with proper fonts (script by J. vanZandt).
 460   Updated documentation to refer to the PNG-1.2 specification.
 461   Removed ansi2knr.c and left pointers to the latest source for ansi2knr.c
 462     in makefile.knr, INSTALL, and README (L. Peter Deutsch)
 463   Fixed bugs in calculation of the length of rowbytes when adding alpha
 464     channels to 16-bit images, in pngrtran.c (Chris Nokleberg)
 465   Added function png_set_user_transform_info() to store user_transform_ptr,
 466     user_depth, and user_channels into the png_struct, and a function
 467     png_get_user_transform_ptr() to retrieve the pointer (Chris Nokleberg)
 468   Added function png_set_empty_plte_permitted() to make libpng useable
 469     in MNG applications.
 470   Corrected the typedef for png_free_ptr in png.h (Jesse Jones).
 471   Correct gamma with srgb is 45455 instead of 45000 in pngrutil.c, to be
 472     consistent with PNG-1.2, and allow variance of 500 before complaining.
 473   Added assembler code contributed by Intel in file pngvcrd.c and modified
 474     makefile.w32 to use it (Nirav Chhatrapati, INTEL Corporation,
 475     Gilles Vollant)
 476   Changed "ln -s -f" to "ln -f -s" in the makefiles to make Solaris happy.
 477   Added some aliases for png_set_expand() in pngrtran.c, namely
 478     png_set_expand_PLTE(), png_set_expand_depth(), and png_set_expand_tRNS()
 479     (Greg Roelofs, in "PNG: The Definitive Guide").
 480   Added makefile.beo for BEOS on X86, contributed by Sander Stok.
 481 
 482 Version 1.0.3b [August 26, 1999]
 483   Replaced 2147483647L several places with PNG_MAX_UINT macro, defined in png.h
 484   Changed leading blanks to tabs in all makefiles.
 485   Define PNG_USE_PNGVCRD in makefile.w32, to get MMX assembler code.
 486   Made alternate versions of  png_set_expand() in pngrtran.c, namely
 487     png_set_gray_1_2_4_to_8, png_set_palette_to_rgb, and png_set_tRNS_to_alpha
 488     (Greg Roelofs, in "PNG: The Definitive Guide").  Deleted the 1.0.3a aliases.
 489   Relocated start of 'extern "C"' block in png.h so it doesn't include pngconf.h
 490   Revised calculation of num_blocks in pngmem.c to avoid a potentially
 491     negative shift distance, whose results are undefined in the C language.
 492   Added a check in pngset.c to prevent writing multiple tIME chunks.
 493   Added a check in pngwrite.c to detect invalid small window_bits sizes.
 494 
 495 Version 1.0.3d [September 4, 1999]
 496   Fixed type casting of igamma in pngrutil.c
 497   Added new png_expand functions to scripts/pngdef.pas and pngos2.def
 498   Added a demo read_user_transform_fn that examines the row filters in pngtest.c
 499 
 500 Version 1.0.4 [September 24, 1999, not distributed publicly]
 501   Define PNG_ALWAYS_EXTERN in pngconf.h if __STDC__ is defined
 502   Delete #define PNG_INTERNAL and include "png.h" from pngasmrd.h
 503   Made several minor corrections to pngtest.c
 504   Renamed the makefiles with longer but more user friendly extensions.
 505   Copied the PNG copyright and license to a separate LICENSE file.
 506   Revised documentation, png.h, and example.c to remove reference to
 507     "viewing_gamma" which no longer appears in the PNG specification.
 508   Revised pngvcrd.c to use MMX code for interlacing only on the final pass.
 509   Updated pngvcrd.c to use the faster C filter algorithms from libpng-1.0.1a
 510   Split makefile.win32vc into two versions, makefile.vcawin32 (uses MMX
 511     assembler code) and makefile.vcwin32 (doesn't).
 512   Added a CPU timing report to pngtest.c (enabled by defining PNGTEST_TIMING)
 513   Added a copy of pngnow.png to the distribution.
 514 
 515 Version 1.0.4a [September 25, 1999]
 516   Increase max_pixel_depth in pngrutil.c if a user transform needs it.
 517   Changed several division operations to right-shifts in pngvcrd.c
 518 
 519 Version 1.0.4b [September 30, 1999]
 520   Added parentheses in line 3732 of pngvcrd.c
 521   Added a comment in makefile.linux warning about buggy -O3 in pgcc 2.95.1
 522 
 523 Version 1.0.4c [October 1, 1999]
 524   Added a "png_check_version" function in png.c and pngtest.c that will generate
 525     a helpful compiler error if an old png.h is found in the search path.
 526   Changed type of png_user_transform_depth|channels from int to png_byte.
 527   Added "Libpng is OSI Certified Open Source Software" statement to png.h
 528 
 529 Version 1.0.4d [October 6, 1999]
 530   Changed 0.45 to 0.45455 in png_set_sRGB()
 531   Removed unused PLTE entries from pngnow.png
 532   Re-enabled some parts of pngvcrd.c (png_combine_row) that work properly.
 533 
 534 Version 1.0.4e [October 10, 1999]
 535   Fixed sign error in pngvcrd.c (Greg Roelofs)
 536   Replaced some instances of memcpy with simple assignments in pngvcrd (GR-P)
 537 
 538 Version 1.0.4f [October 15, 1999]
 539   Surrounded example.c code with #if 0 .. #endif to prevent people from
 540     inadvertently trying to compile it.
 541   Changed png_get_header_version() from a function to a macro in png.h
 542   Added type casting mostly in pngrtran.c and pngwtran.c
 543   Removed some pointless "ptr = NULL" in pngmem.c
 544   Added a "contrib" directory containing the source code from Greg's book.
 545 
 546 Version 1.0.5 [October 15, 1999]
 547   Minor editing of the INSTALL and README files.
 548 
 549 Version 1.0.5a [October 23, 1999]
 550   Added contrib/pngsuite and contrib/pngminus (Willem van Schaik)
 551   Fixed a typo in the png_set_sRGB() function call in example.c (Jan Nijtmans)
 552   Further optimization and bugfix of pngvcrd.c
 553   Revised pngset.c so that it does not allocate or free memory in the user's
 554     text_ptr structure.  Instead, it makes its own copy.
 555   Created separate write_end_info_struct in pngtest.c for a more severe test.
 556   Added code in pngwrite.c to free info_ptr->text[i].key to stop a memory leak.
 557 
 558 Version 1.0.5b [November 23, 1999]
 559   Moved PNG_FLAG_HAVE_CHUNK_HEADER, PNG_FLAG_BACKGROUND_IS_GRAY and
 560     PNG_FLAG_WROTE_tIME from flags to mode.
 561   Added png_write_info_before_PLTE() function.
 562   Fixed some typecasting in contrib/gregbook/*.c
 563   Updated scripts/makevms.com and added makevms.com to contrib/gregbook
 564     and contrib/pngminus (Martin Zinser)
 565 
 566 Version 1.0.5c [November 26, 1999]
 567   Moved png_get_header_version from png.h to png.c, to accommodate ansi2knr.
 568   Removed all global arrays (according to PNG_NO_GLOBAL_ARRAYS macro), to
 569     accommodate making DLL's: Moved usr_png_ver from global variable to function
 570     png_get_header_ver() in png.c.  Moved png_sig to png_sig_bytes in png.c and
 571     eliminated use of png_sig in pngwutil.c.  Moved the various png_CHNK arrays
 572     into pngtypes.h.  Eliminated use of global png_pass arrays.  Declared the
 573     png_CHNK and png_pass arrays to be "const".  Made the global arrays
 574     available to applications (although none are used in libpng itself) when
 575     PNG_NO_GLOBAL_ARRAYS is not defined or when PNG_GLOBAL_ARRAYS is defined.
 576   Removed some extraneous "-I" from contrib/pngminus/makefile.std
 577   Changed the PNG_sRGB_INTENT macros in png.h to be consistent with PNG-1.2.
 578   Change PNG_SRGB_INTENT to PNG_sRGB_INTENT in libpng.txt and libpng.3
 579 
 580 Version 1.0.5d [November 29, 1999]
 581   Add type cast (png_const_charp) two places in png.c
 582   Eliminated pngtypes.h; use macros instead to declare PNG_CHNK arrays.
 583   Renamed "PNG_GLOBAL_ARRAYS" to "PNG_USE_GLOBAL_ARRAYS" and made available
 584     to applications a macro "PNG_USE_LOCAL_ARRAYS".
 585   comment out (with #ifdef) all the new declarations when
 586     PNG_USE_GLOBAL_ARRAYS is defined.
 587   Added PNG_EXPORT_VAR macro to accommodate making DLL's.
 588 
 589 Version 1.0.5e [November 30, 1999]
 590   Added iCCP, iTXt, and sPLT support; added "lang" member to the png_text
 591     structure; refactored the inflate/deflate support to make adding new chunks
 592     with trailing compressed parts easier in the future, and added new functions
 593     png_free_iCCP, png_free_pCAL, png_free_sPLT, png_free_text, png_get_iCCP,
 594     png_get_spalettes, png_set_iCCP, png_set_spalettes (Eric S. Raymond).
 595     NOTE: Applications that write text chunks MUST define png_text->lang
 596     before calling png_set_text(). It must be set to NULL if you want to
 597     write tEXt or zTXt chunks.  If you want your application to be able to
 598     run with older versions of libpng, use
 599 
 600       #ifdef PNG_iTXt_SUPPORTED
 601          png_text[i].lang = NULL;
 602       #endif
 603 
 604   Changed png_get_oFFs() and png_set_oFFs() to use signed rather than unsigned
 605     offsets (Eric S. Raymond).
 606   Combined PNG_READ_cHNK_SUPPORTED and PNG_WRITE_cHNK_SUPPORTED macros into
 607     PNG_cHNK_SUPPORTED and combined the three types of PNG_text_SUPPORTED
 608     macros, leaving the separate macros also available.
 609   Removed comments on #endifs at the end of many short, non-nested #if-blocks.
 610 
 611 Version 1.0.5f [December 6, 1999]
 612   Changed makefile.solaris to issue a warning about potential problems when
 613     the ucb "ld" is in the path ahead of the ccs "ld".
 614   Removed "- [date]" from the "synopsis" line in libpng.3 and libpngpf.3.
 615   Added sCAL chunk support (Eric S. Raymond).
 616 
 617 Version 1.0.5g [December 7, 1999]
 618   Fixed "png_free_spallettes" typo in png.h
 619   Added code to handle new chunks in pngpread.c
 620   Moved PNG_CHNK string macro definitions outside of PNG_NO_EXTERN block
 621   Added "translated_key" to png_text structure and png_write_iTXt().
 622   Added code in pngwrite.c to work around a newly discovered zlib bug.
 623 
 624 Version 1.0.5h [December 10, 1999]
 625   NOTE: regarding the note for version 1.0.5e, the following must also
 626     be included in your code:
 627         png_text[i].translated_key = NULL;
 628   Unknown chunk handling is now supported.
 629   Option to eliminate all floating point support was added.  Some new
 630     fixed-point functions such as png_set_gAMA_fixed() were added.
 631   Expanded tabs and removed trailing blanks in source files.
 632 
 633 Version 1.0.5i [December 13, 1999]
 634   Added some type casts to silence compiler warnings.
 635   Renamed "png_free_spalette" to "png_free_spalettes" for consistency.
 636   Removed leading blanks from a #define in pngvcrd.c
 637   Added some parameters to the new png_set_keep_unknown_chunks() function.
 638   Added a test for up->location != 0 in the first instance of writing
 639     unknown chunks in pngwrite.c
 640   Changed "num" to "i" in png_free_spalettes() and png_free_unknowns() to
 641     prevent recursion.
 642   Added png_free_hIST() function.
 643   Various patches to fix bugs in the sCAL and integer cHRM processing,
 644     and to add some convenience macros for use with sCAL.
 645 
 646 Version 1.0.5j [December 21, 1999]
 647   Changed "unit" parameter of png_write_sCAL from png_byte to int, to work
 648     around buggy compilers.
 649   Added new type "png_fixed_point" for integers that hold float*100000 values
 650   Restored backward compatibility of tEXt/zTXt chunk processing:
 651     Restored the first four members of png_text to the same order as v.1.0.5d.
 652     Added members "lang_key" and "itxt_length" to png_text struct.  Set
 653     text_length=0 when "text" contains iTXt data.  Use the "compression"
 654     member to distinguish among tEXt/zTXt/iTXt types.  Added
 655     PNG_ITXT_COMPRESSION_NONE (1) and PNG_ITXT_COMPRESSION_zTXt(2) macros.
 656     The "Note" above, about backward incompatibility of libpng-1.0.5e, no
 657     longer applies.
 658   Fixed png_read|write_iTXt() to read|write parameters in the right order,
 659     and to write the iTXt chunk after IDAT if it appears in the end_ptr.
 660   Added pnggccrd.c, version of pngvcrd.c Intel assembler for gcc (Greg Roelofs)
 661   Reversed the order of trying to write floating-point and fixed-point gAMA.
 662 
 663 Version 1.0.5k [December 27, 1999]
 664   Added many parentheses, e.g., "if (a && b & c)" becomes "if (a && (b & c))"
 665   Added png_handle_as_unknown() function (Glenn)
 666   Added png_free_chunk_list() function and chunk_list and num_chunk_list members
 667     of png_ptr.
 668   Eliminated erroneous warnings about multiple sPLT chunks and sPLT-after-PLTE.
 669   Fixed a libpng-1.0.5h bug in pngrutil.c that was issuing erroneous warnings
 670     about ignoring incorrect gAMA with sRGB (gAMA was in fact not ignored)
 671   Added png_free_tRNS(); png_set_tRNS() now malloc's its own trans array (ESR).
 672   Define png_get_int_32 when oFFs chunk is supported as well as when pCAL is.
 673   Changed type of proflen from png_int_32 to png_uint_32 in png_get_iCCP().
 674 
 675 Version 1.0.5l [January 1, 2000]
 676   Added functions png_set_read_user_chunk_fn() and png_get_user_chunk_ptr()
 677     for setting a callback function to handle unknown chunks and for
 678     retrieving the associated user pointer (Glenn).
 679 
 680 Version 1.0.5m [January 7, 2000]
 681   Added high-level functions png_read_png(), png_write_png(), png_free_pixels().
 682 
 683 Version 1.0.5n [January 9, 2000]
 684   Added png_free_PLTE() function, and modified png_set_PLTE() to malloc its
 685     own memory for info_ptr->palette.  This makes it safe for the calling
 686     application to free its copy of the palette any time after it calls
 687     png_set_PLTE().
 688 
 689 Version 1.0.5o [January 20, 2000]
 690   Cosmetic changes only (removed some trailing blanks and TABs)
 691 
 692 Version 1.0.5p [January 31, 2000]
 693   Renamed pngdll.mak to makefile.bd32
 694   Cosmetic changes in pngtest.c
 695 
 696 Version 1.0.5q [February 5, 2000]
 697   Relocated the makefile.solaris warning about PATH problems.
 698   Fixed pngvcrd.c bug by pushing/popping registers in mmxsupport (Bruce Oberg)
 699   Revised makefile.gcmmx
 700   Added PNG_SETJMP_SUPPORTED, PNG_SETJMP_NOT_SUPPORTED, and PNG_ABORT() macros
 701 
 702 Version 1.0.5r [February 7, 2000]
 703   Removed superfluous prototype for png_get_itxt from png.h
 704   Fixed a bug in pngrtran.c that improperly expanded the background color.
 705   Return *num_text=0 from png_get_text() when appropriate, and fix documentation
 706     of png_get_text() in libpng.txt/libpng.3.
 707 
 708 Version 1.0.5s [February 18, 2000]
 709   Added "png_jmp_env()" macro to pngconf.h, to help people migrate to the
 710     new error handler that's planned for the next libpng release, and changed
 711     example.c, pngtest.c, and contrib programs to use this macro.
 712   Revised some of the DLL-export macros in pngconf.h (Greg Roelofs)
 713   Fixed a bug in png_read_png() that caused it to fail to expand some images
 714     that it should have expanded.
 715   Fixed some mistakes in the unused and undocumented INCH_CONVERSIONS functions
 716     in pngget.c
 717   Changed the allocation of palette, history, and trans arrays back to
 718     the version 1.0.5 method (linking instead of copying) which restores
 719     backward compatibility with version 1.0.5.  Added some remarks about
 720     that in example.c.  Added "free_me" member to info_ptr and png_ptr
 721     and added png_free_data() function.
 722   Updated makefile.linux and makefile.gccmmx to make directories conditionally.
 723   Made cosmetic changes to pngasmrd.h
 724   Added png_set_rows() and png_get_rows(), for use with png_read|write_png().
 725   Modified png_read_png() to allocate info_ptr->row_pointers only if it
 726     hasn't already been allocated.
 727 
 728 Version 1.0.5t [March 4, 2000]
 729   Changed png_jmp_env() migration aiding macro to png_jmpbuf().
 730   Fixed "interlace" typo (should be "interlaced") in contrib/gregbook/read2-x.c
 731   Fixed bug with use of PNG_BEFORE_IHDR bit in png_ptr->mode, introduced when
 732     PNG_FLAG_HAVE_CHUNK_HEADER was moved into png_ptr->mode in version 1.0.5b
 733   Files in contrib/gregbook were revised to use png_jmpbuf() and to select
 734     a 24-bit visual if one is available, and to allow abbreviated options.
 735   Files in contrib/pngminus were revised to use the png_jmpbuf() macro.
 736   Removed spaces in makefile.linux and makefile.gcmmx, introduced in 1.0.5s
 737 
 738 Version 1.0.5u [March 5, 2000]
 739   Simplified the code that detects old png.h in png.c and pngtest.c
 740   Renamed png_spalette (_p, _pp) to png_sPLT_t (_tp, _tpp)
 741   Increased precision of rgb_to_gray calculations from 8 to 15 bits and
 742     added png_set_rgb_to_gray_fixed() function.
 743   Added makefile.bc32 (32-bit Borland C++, C mode)
 744 
 745 Version 1.0.5v [March 11, 2000]
 746   Added some parentheses to the png_jmpbuf macro definition.
 747   Updated references to the zlib home page, which has moved to freesoftware.com.
 748   Corrected bugs in documentation regarding png_read_row() and png_write_row().
 749   Updated documentation of png_rgb_to_gray calculations in libpng.3/libpng.txt.
 750   Renamed makefile.borland,turboc3 back to makefile.bor,tc3 as in version 1.0.3,
 751     revised borland makefiles; added makefile.ibmvac3 and makefile.gcc (Cosmin)
 752 
 753 Version 1.0.6 [March 20, 2000]
 754   Minor revisions of makefile.bor, libpng.txt, and gregbook/rpng2-win.c
 755   Added makefile.sggcc (SGI IRIX with gcc)
 756 
 757 Version 1.0.6d [April 7, 2000]
 758   Changed sprintf() to strcpy() in png_write_sCAL_s() to work without STDIO
 759   Added data_length parameter to png_decompress_chunk() function
 760   Revised documentation to remove reference to abandoned png_free_chnk functions
 761   Fixed an error in png_rgb_to_gray_fixed()
 762   Revised example.c, usage of png_destroy_write_struct().
 763   Renamed makefile.ibmvac3 to makefile.ibmc, added libpng.icc IBM project file
 764   Added a check for info_ptr->free_me&PNG_FREE_TEXT when freeing text in png.c
 765   Simplify png_sig_bytes() function to remove use of non-ISO-C strdup().
 766 
 767 Version 1.0.6e [April 9, 2000]
 768   Added png_data_freer() function.
 769   In the code that checks for over-length tRNS chunks, added check of
 770     info_ptr->num_trans as well as png_ptr->num_trans (Matthias Benckmann)
 771   Minor revisions of libpng.txt/libpng.3.
 772   Check for existing data and free it if the free_me flag is set, in png_set_*()
 773     and png_handle_*().
 774   Only define PNG_WEIGHTED_FILTERS_SUPPORTED when PNG_FLOATING_POINT_SUPPORTED
 775     is defined.
 776   Changed several instances of PNG_NO_CONSOLE_ID to PNG_NO_STDIO in pngrutil.c
 777     and mentioned the purposes of the two macros in libpng.txt/libpng.3.
 778 
 779 Version 1.0.6f [April 14, 2000]
 780   Revised png_set_iCCP() and png_set_rows() to avoid prematurely freeing data.
 781   Add checks in png_set_text() for NULL members of the input text structure.
 782   Revised libpng.txt/libpng.3.
 783   Removed superfluous prototype for png_set_iTXt from png.h
 784   Removed "else" from pngread.c, after png_error(), and changed "0" to "length".
 785   Changed several png_errors about malformed ancillary chunks to png_warnings.
 786 
 787 Version 1.0.6g [April 24, 2000]
 788   Added png_pass-* arrays to pnggccrd.c when PNG_USE_LOCAL_ARRAYS is defined.
 789   Relocated paragraph about png_set_background() in libpng.3/libpng.txt
 790     and other revisions (Matthias Benckmann)
 791   Relocated info_ptr->free_me, png_ptr->free_me, and other info_ptr and
 792     png_ptr members to restore binary compatibility with libpng-1.0.5
 793     (breaks compatibility with libpng-1.0.6).
 794 
 795 Version 1.0.6h [April 24, 2000]
 796   Changed shared library so-number pattern from 2.x.y.z to xy.z (this builds
 797     libpng.so.10 & libpng.so.10.6h instead of libpng.so.2 & libpng.so.2.1.0.6h)
 798     This is a temporary change for test purposes.
 799 
 800 Version 1.0.6i [May 2, 2000]
 801   Rearranged some members at the end of png_info and png_struct, to put
 802     unknown_chunks_num and free_me within the original size of the png_structs
 803     and free_me, png_read_user_fn, and png_free_fn within the original png_info,
 804     because some old applications allocate the structs directly instead of
 805     using png_create_*().
 806   Added documentation of user memory functions in libpng.txt/libpng.3
 807   Modified png_read_png so that it will use user_allocated row_pointers
 808     if present, unless free_me directs that it be freed, and added description
 809     of the use of png_set_rows() and png_get_rows() in libpng.txt/libpng.3.
 810   Added PNG_LEGACY_SUPPORTED macro, and #ifdef out all new (since version
 811     1.00) members of png_struct and png_info, to regain binary compatibility
 812     when you define this macro.  Capabilities lost in this event
 813     are user transforms (new in version 1.0.0),the user transform pointer
 814     (new in version 1.0.2), rgb_to_gray (new in 1.0.5), iCCP, sCAL, sPLT,
 815     the high-level interface, and unknown chunks support (all new in 1.0.6).
 816     This was necessary because of old applications that allocate the structs
 817     directly as authors were instructed to do in libpng-0.88 and earlier,
 818     instead of using png_create_*().
 819   Added modes PNG_CREATED_READ_STRUCT and PNG_CREATED_WRITE_STRUCT which
 820     can be used to detect codes that directly allocate the structs, and
 821     code to check these modes in png_read_init() and png_write_init() and
 822     generate a libpng error if the modes aren't set and PNG_LEGACY_SUPPORTED
 823     was not defined.
 824   Added makefile.intel and updated makefile.watcom (Pawel Mrochen)
 825 
 826 Version 1.0.6j [May 3, 2000]
 827   Overloaded png_read_init() and png_write_init() with macros that convert
 828     calls to png_read_init_2() or png_write_init_2() that check the version
 829     and structure sizes.
 830 
 831 Version 1.0.7beta11 [May 7, 2000]
 832   Removed the new PNG_CREATED_READ_STRUCT and PNG_CREATED_WRITE_STRUCT modes
 833     which are no longer used.
 834   Eliminated the three new members of png_text when PNG_LEGACY_SUPPORTED is
 835     defined or when neither PNG_READ_iTXt_SUPPORTED nor PNG_WRITE_iTXt_SUPPORTED
 836     is defined.
 837   Made PNG_NO_READ|WRITE_iTXt the default setting, to avoid memory
 838     overrun when old applications fill the info_ptr->text structure directly.
 839   Added PNGAPI macro, and added it to the definitions of all exported functions.
 840   Relocated version macro definitions ahead of the includes of zlib.h and
 841     pngconf.h in png.h.
 842 
 843 Version 1.0.7beta12 [May 12, 2000]
 844   Revised pngset.c to avoid a problem with expanding the png_debug macro.
 845   Deleted some extraneous defines from pngconf.h
 846   Made PNG_NO_CONSOLE_IO the default condition when PNG_BUILD_DLL is defined.
 847   Use MSC _RPTn debugging instead of fprintf if _MSC_VER is defined.
 848   Added png_access_version_number() function.
 849   Check for mask&PNG_FREE_CHNK (for TEXT, SCAL, PCAL) in png_free_data().
 850   Expanded libpng.3/libpng.txt information about png_data_freer().
 851 
 852 Version 1.0.7beta14 [May 17, 2000] (beta13 was not published)
 853   Changed pnggccrd.c and pngvcrd.c to handle bad adaptive filter types as
 854     warnings instead of errors, as pngrutil.c does.
 855   Set the PNG_INFO_IDAT valid flag in png_set_rows() so png_write_png()
 856     will actually write IDATs.
 857   Made the default PNG_USE_LOCAL_ARRAYS depend on PNG_DLL instead of WIN32.
 858   Make png_free_data() ignore its final parameter except when freeing data
 859     that can have multiple instances (text, sPLT, unknowns).
 860   Fixed a new bug in png_set_rows().
 861   Removed info_ptr->valid tests from png_free_data(), as in version 1.0.5.
 862   Added png_set_invalid() function.
 863   Fixed incorrect illustrations of png_destroy_write_struct() in example.c.
 864 
 865 Version 1.0.7beta15 [May 30, 2000]
 866   Revised the deliberately erroneous Linux setjmp code in pngconf.h to produce
 867     fewer error messages.
 868   Rearranged checks for Z_OK to check the most likely path first in pngpread.c
 869     and pngwutil.c.
 870   Added checks in pngtest.c for png_create_*() returning NULL, and mentioned
 871     in libpng.txt/libpng.3 the need for applications to check this.
 872   Changed names of png_default_*() functions in pngtest to pngtest_*().
 873   Changed return type of png_get_x|y_offset_*() from png_uint_32 to png_int_32.
 874   Fixed some bugs in the unused PNG_INCH_CONVERSIONS functions in pngget.c
 875   Set each pointer to NULL after freeing it in png_free_data().
 876   Worked around a problem in pngconf.h; AIX's strings.h defines an "index"
 877     macro that conflicts with libpng's png_color_16.index. (Dimitri
 878     Papadapoulos)
 879   Added "msvc" directory with MSVC++ project files (Simon-Pierre Cadieux).
 880 
 881 Version 1.0.7beta16 [June 4, 2000]
 882   Revised the workaround of AIX string.h "index" bug.
 883   Added a check for overlength PLTE chunk in pngrutil.c.
 884   Added PNG_NO_POINTER_INDEXING macro to use array-indexing instead of pointer
 885     indexing in pngrutil.c and pngwutil.c to accommodate a buggy compiler.
 886   Added a warning in png_decompress_chunk() when it runs out of data, e.g.
 887     when it tries to read an erroneous PhotoShop iCCP chunk.
 888   Added PNG_USE_DLL macro.
 889   Revised the copyright/disclaimer/license notice.
 890   Added contrib/msvctest directory
 891 
 892 Version 1.0.7rc1 [June 9, 2000]
 893   Corrected the definition of PNG_TRANSFORM_INVERT_ALPHA  (0x0400 not 0x0200)
 894   Added contrib/visupng directory (Willem van Schaik)
 895 
 896 Version 1.0.7beta18 [June 23, 2000]
 897   Revised PNGAPI definition, and pngvcrd.c to work with __GCC__
 898     and do not redefine PNGAPI if it is passed in via a compiler directive.
 899   Revised visupng/PngFile.c to remove returns from within the Try block.
 900   Removed leading underscores from "_PNG_H" and "_PNG_SAVE_BSD_SOURCE" macros.
 901   Updated contrib/visupng/cexcept.h to version 1.0.0.
 902   Fixed bugs in pngwrite.c and pngwutil.c that prevented writing iCCP chunks.
 903 
 904 Version 1.0.7rc2 [June 28, 2000]
 905   Updated license to include disclaimers required by UCITA.
 906   Fixed "DJBPP" typo in pnggccrd.c introduced in beta18.
 907 
 908 Version 1.0.7 [July 1, 2000]
 909   Revised the definition of "trans_values" in libpng.3/libpng.txt
 910 
 911 Version 1.0.8beta1 [July 8, 2000]
 912   Added png_free(png_ptr, key) two places in pngpread.c to stop memory leaks.
 913   Changed PNG_NO_STDIO to PNG_NO_CONSOLE_IO, several places in pngrutil.c and
 914     pngwutil.c.
 915   Changed PNG_EXPORT_VAR to use PNG_IMPEXP, in pngconf.h.
 916   Removed unused "#include <assert.h>" from png.c
 917   Added WindowsCE support.
 918   Revised pnggccrd.c to work with gcc-2.95.2 and in the Cygwin environment.
 919 
 920 Version 1.0.8beta2 [July 10, 2000]
 921   Added project files to the wince directory and made further revisions
 922     of pngtest.c, pngrio.c, and pngwio.c in support of WindowsCE.
 923 
 924 Version 1.0.8beta3 [July 11, 2000]
 925   Only set the PNG_FLAG_FREE_TRNS or PNG_FREE_TRNS flag in png_handle_tRNS()
 926     for indexed-color input files to avoid potential double-freeing trans array
 927     under some unusual conditions; problem was introduced in version 1.0.6f.
 928   Further revisions to pngtest.c and files in the wince subdirectory.
 929 
 930 Version 1.0.8beta4 [July 14, 2000]
 931   Added the files pngbar.png and pngbar.jpg to the distribution.
 932   Added makefile.cygwin, and cygwin support in pngconf.h
 933   Added PNG_NO_ZALLOC_ZERO macro (makes png_zalloc skip zeroing memory)
 934 
 935 Version 1.0.8rc1 [July 16, 2000]
 936   Revised png_debug() macros and statements to eliminate compiler warnings.
 937 
 938 Version 1.0.8 [July 24, 2000]
 939   Added png_flush() in pngwrite.c, after png_write_IEND().
 940   Updated makefile.hpux to build a shared library.
 941 
 942 Version 1.0.9beta1 [November 10, 2000]
 943   Fixed typo in scripts/makefile.hpux
 944   Updated makevms.com in scripts and contrib/* and contrib/* (Martin Zinser)
 945   Fixed seqence-point bug in contrib/pngminus/png2pnm (Martin Zinser)
 946   Changed "cdrom.com" in documentation to "libpng.org"
 947   Revised pnggccrd.c to get it all working, and updated makefile.gcmmx (Greg).
 948   Changed type of "params" from voidp to png_voidp in png_read|write_png().
 949   Make sure PNGAPI and PNG_IMPEXP are defined in pngconf.h.
 950   Revised the 3 instances of WRITEFILE in pngtest.c.
 951   Relocated "msvc" and "wince" project subdirectories into "dll" subdirectory.
 952   Updated png.rc in dll/msvc project
 953   Revised makefile.dec to define and use LIBPATH and INCPATH
 954   Increased size of global png_libpng_ver[] array from 12 to 18 chars.
 955   Made global png_libpng_ver[], png_sig[] and png_pass_*[] arrays const.
 956   Removed duplicate png_crc_finish() from png_handle_bKGD() function.
 957   Added a warning when application calls png_read_update_info() multiple times.
 958   Revised makefile.cygwin
 959   Fixed bugs in iCCP support in pngrutil.c and pngwutil.c.
 960   Replaced png_set_empty_plte_permitted() with png_permit_mng_features().
 961 
 962 Version 1.0.9beta2 [November 19, 2000]
 963   Renamed the "dll" subdirectory "projects".
 964   Added borland project files to "projects" subdirectory.
 965   Set VS_FF_PRERELEASE and VS_FF_PATCHED flags in msvc/png.rc when appropriate.
 966   Add error message in png_set_compression_buffer_size() when malloc fails.
 967 
 968 Version 1.0.9beta3 [November 23, 2000]
 969   Revised PNG_LIBPNG_BUILD_TYPE macro in png.h, used in the msvc project.
 970   Removed the png_flush() in pngwrite.c that crashes some applications
 971     that don't set png_output_flush_fn.
 972   Added makefile.macosx and makefile.aix to scripts directory.
 973 
 974 Version 1.0.9beta4 [December 1, 2000]
 975   Change png_chunk_warning to png_warning in png_check_keyword().
 976   Increased the first part of msg buffer from 16 to 18 in png_chunk_error().
 977 
 978 Version 1.0.9beta5 [December 15, 2000]
 979   Added support for filter method 64 (for PNG datastreams embedded in MNG).
 980 
 981 Version 1.0.9beta6 [December 18, 2000]
 982   Revised png_set_filter() to accept filter method 64 when appropriate.
 983   Added new PNG_HAVE_PNG_SIGNATURE bit to png_ptr->mode and use it to
 984     help prevent applications from using MNG features in PNG datastreams.
 985   Added png_permit_mng_features() function.
 986   Revised libpng.3/libpng.txt.  Changed "filter type" to "filter method".
 987 
 988 Version 1.0.9rc1 [December 23, 2000]
 989   Revised test for PNG_HAVE_PNG_SIGNATURE in pngrutil.c
 990   Fixed error handling of unknown compression type in png_decompress_chunk().
 991   In pngconf.h, define __cdecl when _MSC_VER is defined.
 992 
 993 Version 1.0.9beta7 [December 28, 2000]
 994   Changed PNG_TEXT_COMPRESSION_zTXt to PNG_COMPRESSION_TYPE_BASE several places.
 995   Revised memory management in png_set_hIST and png_handle_hIST in a backward
 996     compatible manner.  PLTE and tRNS were revised similarly.
 997   Revised the iCCP chunk reader to ignore trailing garbage.
 998 
 999 Version 1.0.9beta8 [January 12, 2001]
1000   Moved pngasmrd.h into pngconf.h.
1001   Improved handling of out-of-spec garbage iCCP chunks generated by PhotoShop.
1002 
1003 Version 1.0.9beta9 [January 15, 2001]
1004   Added png_set_invalid, png_permit_mng_features, and png_mmx_supported to
1005     wince and msvc project module definition files.
1006   Minor revision of makefile.cygwin.
1007   Fixed bug with progressive reading of narrow interlaced images in pngpread.c
1008 
1009 Version 1.0.9beta10 [January 16, 2001]
1010   Do not typedef png_FILE_p in pngconf.h when PNG_NO_STDIO is defined.
1011   Fixed "png_mmx_supported" typo in project definition files.
1012 
1013 Version 1.0.9beta11 [January 19, 2001]
1014   Updated makefile.sgi to make shared library.
1015   Removed png_mmx_support() function and disabled PNG_MNG_FEATURES_SUPPORTED
1016     by default, for the benefit of DLL forward compatibility.  These will
1017     be re-enabled in version 1.2.0.
1018 
1019 Version 1.0.9rc2 [January 22, 2001]
1020   Revised cygwin support.
1021 
1022 Version 1.0.9 [January 31, 2001]
1023   Added check of cygwin's ALL_STATIC in pngconf.h
1024   Added "-nommx" parameter to contrib/gregbook/rpng2-win and rpng2-x demos.
1025 
1026 Version 1.0.10beta1 [March 14, 2001]
1027   Revised makefile.dec, makefile.sgi, and makefile.sggcc; added makefile.hpgcc.
1028   Reformatted libpng.3 to eliminate bad line breaks.
1029   Added checks for _mmx_supported in the read_filter_row function of pnggccrd.c
1030   Added prototype for png_mmx_support() near the top of pnggccrd.c
1031   Moved some error checking from png_handle_IHDR to png_set_IHDR.
1032   Added PNG_NO_READ_SUPPORTED and PNG_NO_WRITE_SUPPORTED macros.
1033   Revised png_mmx_support() function in pnggccrd.c
1034   Restored version 1.0.8 PNG_WRITE_EMPTY_PLTE_SUPPORTED behavior in pngwutil.c
1035   Fixed memory leak in contrib/visupng/PngFile.c
1036   Fixed bugs in png_combine_row() in pnggccrd.c and pngvcrd.c (C version)
1037   Added warnings when retrieving or setting gamma=0.
1038   Increased the first part of msg buffer from 16 to 18 in png_chunk_warning().
1039 
1040 Version 1.0.10rc1 [March 23, 2001]
1041   Changed all instances of memcpy, strcpy, and strlen to png_memcpy, png_strcpy,
1042     and png_strlen.
1043   Revised png_mmx_supported() function in pnggccrd.c to return proper value.
1044   Fixed bug in progressive reading (pngpread.c) with small images (height < 8).
1045 
1046 Version 1.0.10 [March 30, 2001]
1047   Deleted extraneous space (introduced in 1.0.9) from line 42 of makefile.cygwin
1048   Added beos project files (Chris Herborth)
1049 
1050 Version 1.0.11beta1 [April 3, 2001]
1051   Added type casts on several png_malloc() calls (Dimitri Papadapoulos).
1052   Removed a no-longer needed AIX work-around from pngconf.h
1053   Changed several "//" single-line comments to C-style in pnggccrd.c
1054 
1055 Version 1.0.11beta2 [April 11, 2001]
1056   Removed PNGAPI from several functions whose prototypes did not have PNGAPI.
1057   Updated scripts/pngos2.def
1058 
1059 Version 1.0.11beta3 [April 14, 2001]
1060   Added checking the results of many instances of png_malloc() for NULL
1061 
1062 Version 1.0.11beta4 [April 20, 2001]
1063   Undid the changes from version 1.0.11beta3.  Added a check for NULL return
1064     from user's malloc_fn().
1065   Removed some useless type casts of the NULL pointer.
1066   Added makefile.netbsd
1067 
1068 Version 1.0.11 [April 27, 2001]
1069   Revised makefile.netbsd
1070 
1071 Version 1.0.12beta1 [May 14, 2001]
1072   Test for Windows platform in pngconf.h when including malloc.h (Emmanuel Blot)
1073   Updated makefile.cygwin and handling of Cygwin's ALL_STATIC in pngconf.h
1074   Added some never-to-be-executed code in pnggccrd.c to quiet compiler warnings.
1075   Eliminated the png_error about apps using png_read|write_init().  Instead,
1076     libpng will reallocate the png_struct and info_struct if they are too small.
1077     This retains future binary compatibility for old applications written for
1078     libpng-0.88 and earlier.
1079 
1080 Version 1.2.0beta1 [May 6, 2001]
1081   Bumped DLLNUM to 2.
1082   Re-enabled PNG_MNG_FEATURES_SUPPORTED and enabled PNG_ASSEMBLER_CODE_SUPPORTED
1083     by default.
1084   Added runtime selection of MMX features.
1085   Added png_set_strip_error_numbers function and related macros.
1086 
1087 Version 1.2.0beta2 [May 7, 2001]
1088   Finished merging 1.2.0beta1 with version 1.0.11
1089   Added a check for attempts to read or write PLTE in grayscale PNG datastreams.
1090 
1091 Version 1.2.0beta3 [May 17, 2001]
1092   Enabled user memory function by default.
1093   Modified png_create_struct so it passes user mem_ptr to user memory allocator.
1094   Increased png_mng_features flag from png_byte to png_uint_32.
1095   Bumped shared-library (so-number) and dll-number to 3.
1096 
1097 Version 1.2.0beta4 [June 23, 2001]
1098   Check for missing profile length field in iCCP chunk and free chunk_data
1099     in case of truncated iCCP chunk.
1100   Bumped shared-library number to 3 in makefile.sgi and makefile.sggcc
1101   Bumped dll-number from 2 to 3 in makefile.cygwin
1102   Revised contrib/gregbook/rpng*-x.c to avoid a memory leak and to exit cleanly
1103     if user attempts to run it on an 8-bit display.
1104   Updated contrib/gregbook
1105   Use png_malloc instead of png_zalloc to allocate palette in pngset.c
1106   Updated makefile.ibmc
1107   Added some typecasts to eliminate gcc 3.0 warnings.  Changed prototypes
1108     of png_write_oFFS width and height from png_uint_32 to png_int_32.
1109   Updated example.c
1110   Revised prototypes for png_debug_malloc and png_debug_free in pngtest.c
1111 
1112 Version 1.2.0beta5 [August 8, 2001]
1113   Revised contrib/gregbook
1114   Revised makefile.gcmmx
1115   Revised pnggccrd.c to conditionally compile some thread-unsafe code only
1116     when PNG_THREAD_UNSAFE_OK is defined.
1117   Added tests to prevent pngwutil.c from writing a bKGD or tRNS chunk with
1118     value exceeding 2^bit_depth-1
1119   Revised makefile.sgi and makefile.sggcc
1120   Replaced calls to fprintf(stderr,...) with png_warning() in pnggccrd.c
1121   Removed restriction that do_invert_mono only operate on 1-bit opaque files
1122 
1123 Version 1.2.0 [September 1, 2001]
1124   Changed a png_warning() to png_debug() in pnggccrd.c
1125   Fixed contrib/gregbook/rpng-x.c, rpng2-x.c to avoid crash with XFreeGC().
1126 
1127 Version 1.2.1beta1 [October 19, 2001]
1128   Revised makefile.std in contrib/pngminus
1129   Include background_1 in png_struct regardless of gamma support.
1130   Revised makefile.netbsd and makefile.macosx, added makefile.darwin.
1131   Revised example.c to provide more details about using row_callback().
1132 
1133 Version 1.2.1beta2 [October 25, 2001]
1134   Added type cast to each NULL appearing in a function call, except for
1135     WINCE functions.
1136   Added makefile.so9.
1137 
1138 Version 1.2.1beta3 [October 27, 2001]
1139   Removed type casts from all NULLs.
1140   Simplified png_create_struct_2().
1141 
1142 Version 1.2.1beta4 [November 7, 2001]
1143   Revised png_create_info_struct() and png_creat_struct_2().
1144   Added error message if png_write_info() was omitted.
1145   Type cast NULLs appearing in function calls when _NO_PROTO or
1146     PNG_TYPECAST_NULL is defined.
1147 
1148 Version 1.2.1rc1 [November 24, 2001]
1149   Type cast NULLs appearing in function calls except when PNG_NO_TYPECAST_NULL
1150     is defined.
1151   Changed typecast of "size" argument to png_size_t in pngmem.c calls to
1152     the user malloc_fn, to agree with the prototype in png.h
1153   Added a pop/push operation to pnggccrd.c, to preserve Eflag (Maxim Sobolev)
1154   Updated makefile.sgi to recognize LIBPATH and INCPATH.
1155   Updated various makefiles so "make clean" does not remove previous major
1156     version of the shared library.
1157 
1158 Version 1.2.1rc2 [December 4, 2001]
1159   Always allocate 256-entry internal palette, hist, and trans arrays, to
1160     avoid out-of-bounds memory reference caused by invalid PNG datastreams.
1161   Added a check for prefix_length > data_length in iCCP chunk handler.
1162 
1163 Version 1.2.1 [December 7, 2001]
1164   None.
1165 
1166 Version 1.2.2beta1 [February 22, 2002]
1167   Fixed a bug with reading the length of iCCP profiles (Larry Reeves).
1168   Revised makefile.linux, makefile.gcmmx, and makefile.sgi to generate
1169     libpng.a, libpng12.so (not libpng.so.3), and libpng12/png.h
1170   Revised makefile.darwin to remove "-undefined suppress" option.
1171   Added checks for gamma and chromaticity values over 21474.83, which exceed
1172     the limit for PNG unsigned 32-bit integers when encoded.
1173   Revised calls to png_create_read_struct() and png_create_write_struct()
1174     for simpler debugging.
1175   Revised png_zalloc() so zlib handles errors (uses PNG_FLAG_MALLOC_NULL_MEM_OK)
1176 
1177 Version 1.2.2beta2 [February 23, 2002]
1178   Check chunk_length and idat_size for invalid (over PNG_MAX_UINT) lengths.
1179   Check for invalid image dimensions in png_get_IHDR.
1180   Added missing "fi;" in the install target of the SGI makefiles.
1181   Added install-static to all makefiles that make shared libraries.
1182   Always do gamma compensation when image is partially transparent.
1183 
1184 Version 1.2.2beta3 [March 7, 2002]
1185   Compute background.gray and background_1.gray even when color_type is RGB
1186     in case image gets reduced to gray later.
1187   Modified shared-library makefiles to install pkgconfig/libpngNN.pc.
1188   Export (with PNGAPI) png_zalloc, png_zfree, and png_handle_as_unknown
1189   Removed unused png_write_destroy_info prototype from png.h
1190   Eliminated incorrect use of width_mmx from pnggccrd.c in pixel_bytes == 8 case
1191   Added install-shared target to all makefiles that make shared libraries.
1192   Stopped a double free of palette, hist, and trans when not using free_me.
1193   Added makefile.32sunu for Sun Ultra 32 and makefile.64sunu for Sun Ultra 64.
1194 
1195 Version 1.2.2beta4 [March 8, 2002]
1196   Compute background.gray and background_1.gray even when color_type is RGB
1197     in case image gets reduced to gray later (Jason Summers).
1198   Relocated a misplaced /bin/rm in the "install-shared" makefile targets
1199   Added PNG_1_0_X macro which can be used to build a 1.0.x-compatible library.
1200 
1201 Version 1.2.2beta5 [March 26, 2002]
1202   Added missing PNGAPI to several function definitions.
1203   Check for invalid bit_depth or color_type in png_get_IHDR(), and
1204     check for missing PLTE or IHDR in png_push_read_chunk() (Matthias Clasen).
1205   Revised iTXt support to accept NULL for lang and lang_key.
1206   Compute gamma for color components of background even when color_type is gray.
1207   Changed "()" to "{}" in scripts/libpng.pc.in.
1208   Revised makefiles to put png.h and pngconf.h only in $prefix/include/libpngNN
1209   Revised makefiles to make symlink to libpng.so.NN in addition to libpngNN.so
1210 
1211 Version 1.2.2beta6 [March 31, 2002]
1212 
1213 Version 1.0.13beta1 [March 31, 2002]
1214   Prevent png_zalloc() from trying to memset memory that it failed to acquire.
1215   Add typecasts of PNG_MAX_UINT in pngset_cHRM_fixed() (Matt Holgate).
1216   Ensure that the right function (user or default) is used to free the
1217     png_struct after an error in png_create_read_struct_2().
1218 
1219 Version 1.2.2rc1 [April 7, 2002]
1220 
1221 Version 1.0.13rc1 [April 7, 2002]
1222   Save the ebx register in pnggccrd.c (Sami Farin)
1223   Add "mem_ptr = png_ptr->mem_ptr" in png_destroy_write_struct() (Paul Gardner).
1224   Updated makefiles to put headers in include/libpng and remove old include/*.h.
1225 
1226 Version 1.2.2 [April 15, 2002]
1227 
1228 Version 1.0.13 [April 15, 2002]
1229   Revised description of png_set_filter() in libpng.3/libpng.txt.
1230   Revised makefile.netbsd and added makefile.neNNbsd and makefile.freebsd
1231 
1232 Version 1.0.13patch01 [April 17, 2002]
1233 
1234 Version 1.2.2patch01 [April 17, 2002]
1235   Changed ${PNGMAJ}.${PNGVER} bug to ${PNGVER} in makefile.sgi and
1236     makefile.sggcc
1237   Fixed VER -> PNGVER typo in makefile.macosx and added install-static to
1238     install
1239   Added install: target to makefile.32sunu and makefile.64sunu
1240 
1241 Version 1.0.13patch03 [April 18, 2002]
1242 
1243 Version 1.2.2patch03 [April 18, 2002]
1244   Revised 15 makefiles to link libpng.a to libpngNN.a and the include libpng
1245   subdirectory to libpngNN subdirectory without the full pathname.
1246   Moved generation of libpng.pc from "install" to "all" in 15 makefiles.
1247 
1248 Version 1.2.3rc1 [April 28, 2002]
1249   Added install-man target to 15 makefiles (Dimitri Papadopolous-Orfanos).
1250   Added $(DESTDIR) feature to 24 makefiles (Tim Mooney)
1251   Fixed bug with $prefix, should be $(prefix) in makefile.hpux.
1252   Updated cygwin-specific portion of pngconf.h and revised makefile.cygwin
1253   Added a link from libpngNN.pc to libpng.pc in 15 makefiles.
1254   Added links from include/libpngNN/*.h to include/*.h in 24 makefiles.
1255   Revised makefile.darwin to make relative links without full pathname.
1256   Added setjmp() at the end of png_create_*_struct_2() in case user forgets
1257     to put one in their application.
1258   Restored png_zalloc() and png_zfree() prototypes to version 1.2.1 and
1259     removed them from module definition files.
1260 
1261 Version 1.2.3rc2 [May 1, 2002]
1262   Fixed bug in reporting number of channels in pngget.c and pngset.c,
1263     that was introduced in version 1.2.2beta5.
1264   Exported png_zalloc(), png_zfree(), png_default_read(), png_default_write(),
1265     png_default_flush(), and png_push_fill_buffer() and included them in
1266     module definition files.
1267   Added "libpng.pc" dependency to the "install-shared" target in 15 makefiles.
1268 
1269 Version 1.2.3rc3 [May 1, 2002]
1270   Revised prototype for png_default_flush()
1271   Remove old libpng.pc and libpngNN.pc before installing new ones.
1272 
1273 Version 1.2.3rc4 [May 2, 2002]
1274   Typos in *.def files (png_default_read|write -> png_default_read|write_data)
1275   In makefiles, changed rm libpng.NN.pc to rm libpngNN.pc
1276   Added libpng-config and libpngNN-config and modified makefiles to install
1277     them.
1278   Changed $(MANPATH) to $(DESTDIR)$(MANPATH) in makefiles
1279   Added "Win32 DLL VB" configuration to projects/msvc/libpng.dsp
1280 
1281 Version 1.2.3rc5 [May 11, 2002]
1282   Changed "error" and "message" in prototypes to "error_message" and
1283     "warning_message" to avoid namespace conflict.
1284   Revised 15 makefiles to build libpng-config from libpng-config-*.in
1285   Once more restored png_zalloc and png_zfree to regular nonexported form.
1286   Restored png_default_read|write_data, png_default_flush, png_read_fill_buffer
1287     to nonexported form, but with PNGAPI, and removed them from module def
1288     files.
1289 
1290 Version 1.2.3rc6 [May 14, 2002]
1291   Removed "PNGAPI" from png_zalloc() and png_zfree() in png.c
1292   Changed "Gz" to "Gd" in projects/msvc/libpng.dsp and zlib.dsp.
1293   Removed leftover libpng-config "sed" script from four makefiles.
1294   Revised libpng-config creating script in 16 makefiles.
1295 
1296 Version 1.2.3 [May 22, 2002]
1297   Revised libpng-config target in makefile.cygwin.
1298   Removed description of png_set_mem_fn() from documentation.
1299   Revised makefile.freebsd.
1300   Minor cosmetic changes to 15 makefiles, e.g., $(DI) = $(DESTDIR)/$(INCDIR).
1301   Revised projects/msvc/README.txt
1302   Changed -lpng to -lpngNN in LDFLAGS in several makefiles.
1303 
1304 Version 1.2.4beta1 [May 24, 2002]
1305   Added libpng.pc and libpng-config to "all:" target in 16 makefiles.
1306   Fixed bug in 16 makefiles: $(DESTDIR)/$(LIBPATH) to $(DESTDIR)$(LIBPATH)
1307   Added missing "\" before closing double quote in makefile.gcmmx.
1308   Plugged various memory leaks; added png_malloc_warn() and png_set_text_2()
1309     functions.
1310 
1311 Version 1.2.4beta2 [June 25, 2002]
1312   Plugged memory leak of png_ptr->current_text (Matt Holgate).
1313   Check for buffer overflow before reading CRC in pngpread.c (Warwick Allison)
1314   Added -soname to the loader flags in makefile.dec, makefile.sgi, and
1315     makefile.sggcc.
1316   Added "test-installed" target to makefile.linux, makefile.gcmmx,
1317     makefile.sgi, and makefile.sggcc.
1318 
1319 Version 1.2.4beta3 [June 28, 2002]
1320   Plugged memory leak of row_buf in pngtest.c when there is a png_error().
1321   Detect buffer overflow in pngpread.c when IDAT is corrupted with extra data.
1322   Added "test-installed" target to makefile.32sunu, makefile.64sunu,
1323     makefile.beos, makefile.darwin, makefile.dec, makefile.macosx,
1324     makefile.solaris, makefile.hpux, makefile.hpgcc, and makefile.so9.
1325 
1326 Version 1.2.4rc1 and 1.0.14rc1 [July 2, 2002]
1327   Added "test-installed" target to makefile.cygwin and makefile.sco.
1328   Revised pnggccrd.c to be able to back out version 1.0.x via PNG_1_0_X macro.
1329 
1330 Version 1.2.4 and 1.0.14 [July 8, 2002]
1331   Changed png_warning() to png_error() when width is too large to process.
1332 
1333 Version 1.2.4patch01 [July 20, 2002]
1334   Revised makefile.cygwin to use DLL number 12 instead of 13.
1335 
1336 Version 1.2.5beta1 [August 6, 2002]
1337   Added code to contrib/gregbook/readpng2.c to ignore unused chunks.
1338   Replaced toucan.png in contrib/gregbook (it has been corrupt since 1.0.11)
1339   Removed some stray *.o files from contrib/gregbook.
1340   Changed png_error() to png_warning() about "Too much data" in pngpread.c
1341     and about "Extra compressed data" in pngrutil.c.
1342   Prevent png_ptr->pass from exceeding 7 in png_push_finish_row().
1343   Updated makefile.hpgcc
1344   Updated png.c and pnggccrd.c handling of return from png_mmx_support()
1345 
1346 Version 1.2.5beta2 [August 15, 2002]
1347   Only issue png_warning() about "Too much data" in pngpread.c when avail_in
1348     is nonzero.
1349   Updated makefiles to install a separate libpng.so.3 with its own rpath.
1350 
1351 Version 1.2.5rc1 and 1.0.15rc1 [August 24, 2002]
1352   Revised makefiles to not remove previous minor versions of shared libraries.
1353 
1354 Version 1.2.5rc2 and 1.0.15rc2 [September 16, 2002]
1355   Revised 13 makefiles to remove "-lz" and "-L$(ZLIBLIB)", etc., from shared
1356     library loader directive.
1357   Added missing "$OBJSDLL" line to makefile.gcmmx.
1358   Added missing "; fi" to makefile.32sunu.
1359 
1360 Version 1.2.5rc3 and 1.0.15rc3 [September 18, 2002]
1361   Revised libpng-config script.
1362 
1363 Version 1.2.5 and 1.0.15 [October 3, 2002]
1364   Revised makefile.macosx, makefile.darwin, makefile.hpgcc, and makefile.hpux,
1365     and makefile.aix.
1366   Relocated two misplaced PNGAPI lines in pngtest.c
1367 
1368 Version 1.2.6beta1 [October 22, 2002]
1369   Commented out warning about uninitialized mmx_support in pnggccrd.c.
1370   Changed "IBMCPP__" flag to "__IBMCPP__" in pngconf.h.
1371   Relocated two more misplaced PNGAPI lines in pngtest.c
1372   Fixed memory overrun bug in png_do_read_filler() with 16-bit datastreams,
1373     introduced in version 1.0.2.
1374   Revised makefile.macosx, makefile.dec, makefile.aix, and makefile.32sunu.
1375 
1376 Version 1.2.6beta2 [November 1, 2002]
1377   Added libpng-config "--ldopts" output.
1378   Added "AR=ar" and "ARFLAGS=rc" and changed "ar rc" to "$(AR) $(ARFLAGS)"
1379     in makefiles.
1380 
1381 Version 1.2.6beta3 [July 18, 2004]
1382   Reverted makefile changes from version 1.2.6beta2 and some of the changes
1383     from version 1.2.6beta1; these will be postponed until version 1.2.7.
1384     Version 1.2.6 is going to be a simple bugfix release.
1385   Changed the one instance of "ln -sf" to "ln -f -s" in each Sun makefile.
1386   Fixed potential overrun in pngerror.c by using strncpy instead of memcpy.
1387   Added "#!/bin/sh" at the top of configure, for recognition of the
1388     'x' flag under Cygwin (Cosmin).
1389   Optimized vacuous tests that silence compiler warnings, in png.c (Cosmin).
1390   Added support for PNG_USER_CONFIG, in pngconf.h (Cosmin).
1391   Fixed the special memory handler for Borland C under DOS, in pngmem.c
1392     (Cosmin).
1393   Removed some spurious assignments in pngrutil.c (Cosmin).
1394   Replaced 65536 with 65536L, and 0xffff with 0xffffL, to silence warnings
1395     on 16-bit platforms (Cosmin).
1396   Enclosed shift op expressions in parentheses, to silence warnings (Cosmin).
1397   Used proper type png_fixed_point, to avoid problems on 16-bit platforms,
1398     in png_handle_sRGB() (Cosmin).
1399   Added compression_type to png_struct, and optimized the window size
1400     inside the deflate stream (Cosmin).
1401   Fixed definition of isnonalpha(), in pngerror.c and pngrutil.c (Cosmin).
1402   Fixed handling of unknown chunks that come after IDAT (Cosmin).
1403   Allowed png_error() and png_warning() to work even if png_ptr == NULL
1404     (Cosmin).
1405   Replaced row_info->rowbytes with row_bytes in png_write_find_filter()
1406     (Cosmin).
1407   Fixed definition of PNG_LIBPNG_VER_DLLNUM (Simon-Pierre).
1408   Used PNG_LIBPNG_VER and PNG_LIBPNG_VER_STRING instead of the hardcoded
1409     values in png.c (Simon-Pierre, Cosmin).
1410   Initialized png_libpng_ver[] with PNG_LIBPNG_VER_STRING (Simon-Pierre).
1411   Replaced PNG_LIBPNG_VER_MAJOR with PNG_LIBPNG_VER_DLLNUM in png.rc
1412     (Simon-Pierre).
1413   Moved the definition of PNG_HEADER_VERSION_STRING near the definitions
1414     of the other PNG_LIBPNG_VER_... symbols in png.h (Cosmin).
1415   Relocated #ifndef PNGAPI guards in pngconf.h (Simon-Pierre, Cosmin).
1416   Updated scripts/makefile.vc(a)win32 (Cosmin).
1417   Updated the MSVC project (Simon-Pierre, Cosmin).
1418   Updated the Borland C++ Builder project (Cosmin).
1419   Avoided access to asm_flags in pngvcrd.c, if PNG_1_0_X is defined (Cosmin).
1420   Commented out warning about uninitialized mmx_support in pngvcrd.c (Cosmin).
1421   Removed scripts/makefile.bd32 and scripts/pngdef.pas (Cosmin).
1422   Added extra guard around inclusion of Turbo C memory headers, in pngconf.h
1423     (Cosmin).
1424   Renamed projects/msvc/ to projects/visualc6/, and projects/borland/ to
1425     projects/cbuilder5/ (Cosmin).
1426   Moved projects/visualc6/png32ms.def to scripts/pngw32.def,
1427     and projects/visualc6/png.rc to scripts/pngw32.rc (Cosmin).
1428   Added projects/visualc6/pngtest.dsp; removed contrib/msvctest/ (Cosmin).
1429   Changed line endings to DOS style in cbuilder5 and visualc6 files, even
1430     in the tar.* distributions (Cosmin).
1431   Updated contrib/visupng/VisualPng.dsp (Cosmin).
1432   Updated contrib/visupng/cexcept.h to version 2.0.0 (Cosmin).
1433   Added a separate distribution with "configure" and supporting files (Junichi).
1434 
1435 Version 1.2.6beta4 [July 28, 2004]
1436   Added user ability to change png_size_t via a PNG_SIZE_T macro.
1437   Added png_sizeof() and png_convert_size() functions.
1438   Added PNG_SIZE_MAX (maximum value of a png_size_t variable.
1439   Added check in png_malloc_default() for (size_t)size != (png_uint_32)size
1440     which would indicate an overflow.
1441   Changed sPLT failure action from png_error to png_warning and abandon chunk.
1442   Changed sCAL and iCCP failures from png_error to png_warning and abandon.
1443   Added png_get_uint_31(png_ptr, buf) function.
1444   Added PNG_UINT_32_MAX macro.
1445   Renamed PNG_MAX_UINT to PNG_UINT_31_MAX.
1446   Made png_zalloc() issue a png_warning and return NULL on potential
1447     overflow.
1448   Turn on PNG_NO_ZALLOC_ZERO by default in version 1.2.x
1449   Revised "clobber list" in pnggccrd.c so it will compile under gcc-3.4.
1450   Revised Borland portion of png_malloc() to return NULL or issue
1451     png_error() according to setting of PNG_FLAG_MALLOC_NULL_MEM_OK.
1452   Added PNG_NO_SEQUENTIAL_READ_SUPPORTED macro to conditionally remove
1453     sequential read support.
1454   Added some "#if PNG_WRITE_SUPPORTED" blocks.
1455   Added #ifdef to remove some redundancy in png_malloc_default().
1456   Use png_malloc instead of png_zalloc to allocate the palette.
1457 
1458 Version 1.0.16rc1 and 1.2.6rc1 [August 4, 2004]
1459   Fixed buffer overflow vulnerability (CVE-2004-0597) in png_handle_tRNS().
1460   Fixed NULL dereference vulnerability (CVE-2004-0598) in png_handle_iCCP().
1461   Fixed integer overflow vulnerability (CVE-2004-0599) in png_read_png().
1462   Fixed some harmless bugs in png_handle_sBIT, etc, that would cause
1463     duplicate chunk types to go undetected.
1464   Fixed some timestamps in the -config version
1465   Rearranged order of processing of color types in png_handle_tRNS().
1466   Added ROWBYTES macro to calculate rowbytes without integer overflow.
1467   Updated makefile.darwin and removed makefile.macosx from scripts directory.
1468   Imposed default one million column, one-million row limits on the image
1469     dimensions, and added png_set_user_limits() function to override them.
1470   Revised use of PNG_SET_USER_LIMITS_SUPPORTED macro.
1471   Fixed wrong cast of returns from png_get_user_width|height_max().
1472   Changed some "keep the compiler happy" from empty statements to returns,
1473   Revised libpng.txt to remove 1.2.x stuff from the 1.0.x distribution
1474 
1475 Version 1.0.16rc2 and 1.2.6rc2 [August 7, 2004]
1476   Revised makefile.darwin and makefile.solaris.  Removed makefile.macosx.
1477   Revised pngtest's png_debug_malloc() to use png_malloc() instead of
1478     png_malloc_default() which is not supposed to be exported.
1479   Fixed off-by-one error in one of the conversions to PNG_ROWBYTES() in
1480     pngpread.c.  Bug was introduced in 1.2.6rc1.
1481   Fixed bug in RGB to RGBX transformation introduced in 1.2.6rc1.
1482   Fixed old bug in RGB to Gray transformation.
1483   Fixed problem with 64-bit compilers by casting arguments to abs()
1484     to png_int_32.
1485   Changed "ln -sf" to "ln -f -s" in three makefiles (solaris, sco, so9).
1486   Changed "HANDLE_CHUNK_*" to "PNG_HANDLE_CHUNK_*" (Cosmin)
1487   Added "-@/bin/rm -f $(DL)/$(LIBNAME).so.$(PNGMAJ)" to 15 *NIX makefiles.
1488   Added code to update the row_info->colortype in png_do_read_filler() (MSB).
1489 
1490 Version 1.0.16rc3 and 1.2.6rc3 [August 9, 2004]
1491   Eliminated use of "abs()" in testing cHRM and gAMA values, to avoid
1492     trouble with some 64-bit compilers.  Created PNG_OUT_OF_RANGE() macro.
1493   Revised documentation of png_set_keep_unknown_chunks().
1494   Check handle_as_unknown status in pngpread.c, as in pngread.c previously.
1495   Moved  "PNG_HANDLE_CHUNK_*" macros out of PNG_INTERNAL section of png.h
1496   Added "rim" definitions for CONST4 and CONST6 in pnggccrd.c
1497 
1498 Version 1.0.16rc4 and 1.2.6rc4 [August 10, 2004]
1499   Fixed mistake in pngtest.c introduced in 1.2.6rc2 (declaration of
1500     "pinfo" was out of place).
1501 
1502 Version 1.0.16rc5 and 1.2.6rc5 [August 10, 2004]
1503   Moved  "PNG_HANDLE_CHUNK_*" macros out of PNG_ASSEMBLER_CODE_SUPPORTED
1504     section of png.h where they were inadvertently placed in version rc3.
1505 
1506 Version 1.2.6 and 1.0.16 [August 15, 2004]
1507   Revised pngtest so memory allocation testing is only done when PNG_DEBUG==1.
1508 
1509 Version 1.2.7beta1 [August 26, 2004]
1510   Removed unused pngasmrd.h file.
1511   Removed references to uu.net for archived files.  Added references to
1512     PNG Spec (second edition) and the PNG ISO/IEC Standard.
1513   Added "test-dd" target in 15 makefiles, to run pngtest in DESTDIR.
1514   Fixed bug with "optimized window size" in the IDAT datastream, that
1515     causes libpng to write PNG files with incorrect zlib header bytes.
1516 
1517 Version 1.2.7beta2 [August 28, 2004]
1518   Fixed bug with sCAL chunk and big-endian machines (David Munro).
1519   Undid new code added in 1.2.6rc2 to update the color_type in
1520     png_set_filler().
1521   Added png_set_add_alpha() that updates color type.
1522 
1523 Version 1.0.17rc1 and 1.2.7rc1 [September 4, 2004]
1524   Revised png_set_strip_filler() to not remove alpha if color_type has alpha.
1525 
1526 Version 1.2.7 and 1.0.17 [September 12, 2004]
1527   Added makefile.hp64
1528   Changed projects/msvc/png32ms.def to scripts/png32ms.def in makefile.cygwin
1529 
1530 Version 1.2.8beta1 [November 1, 2004]
1531   Fixed bug in png_text_compress() that would fail to complete a large block.
1532   Fixed bug, introduced in libpng-1.2.7, that overruns a buffer during
1533     strip alpha operation in png_do_strip_filler().
1534   Added PNG_1_2_X definition in pngconf.h
1535   Use #ifdef to comment out png_info_init in png.c and png_read_init in
1536     pngread.c (as of 1.3.0)
1537 
1538 Version 1.2.8beta2 [November 2, 2004]
1539   Reduce color_type to a nonalpha type after strip alpha operation in
1540     png_do_strip_filler().
1541 
1542 Version 1.2.8beta3 [November 3, 2004]
1543   Revised definitions of PNG_MAX_UINT_32, PNG_MAX_SIZE, and PNG_MAXSUM
1544 
1545 Version 1.2.8beta4 [November 12, 2004]
1546   Fixed (again) definition of PNG_LIBPNG_VER_DLLNUM in png.h (Cosmin).
1547   Added PNG_LIBPNG_BUILD_PRIVATE in png.h (Cosmin).
1548   Set png_ptr->zstream.data_type to Z_BINARY, to avoid unnecessary detection
1549     of data type in deflate (Cosmin).
1550   Deprecated but continue to support SPECIALBUILD and PRIVATEBUILD in favor of
1551     PNG_LIBPNG_BUILD_SPECIAL_STRING and PNG_LIBPNG_BUILD_PRIVATE_STRING.
1552 
1553 Version 1.2.8beta5 [November 20, 2004]
1554   Use png_ptr->flags instead of png_ptr->transformations to pass
1555     PNG_STRIP_ALPHA info to png_do_strip_filler(), to preserve ABI
1556     compatibility.
1557   Revised handling of SPECIALBUILD, PRIVATEBUILD,
1558     PNG_LIBPNG_BUILD_SPECIAL_STRING and PNG_LIBPNG_BUILD_PRIVATE_STRING.
1559 
1560 Version 1.2.8rc1 [November 24, 2004]
1561   Moved handling of BUILD macros from pngconf.h to png.h
1562   Added definition of PNG_LIBPNG_BASE_TYPE in png.h, inadvertently
1563     omitted from beta5.
1564   Revised scripts/pngw32.rc
1565   Despammed mailing addresses by masking "@" with "at".
1566   Inadvertently installed a supposedly faster test version of pngrutil.c
1567 
1568 Version 1.2.8rc2 [November 26, 2004]
1569   Added two missing "\" in png.h
1570   Change tests in pngread.c and pngpread.c to
1571     if (png_ptr->transformations || (png_ptr->flags&PNG_FLAG_STRIP_ALPHA))
1572        png_do_read_transformations(png_ptr);
1573 
1574 Version 1.2.8rc3 [November 28, 2004]
1575   Reverted pngrutil.c to version libpng-1.2.8beta5.
1576   Added scripts/makefile.elf with supporting code in pngconf.h for symbol
1577     versioning (John Bowler).
1578 
1579 Version 1.2.8rc4 [November 29, 2004]
1580   Added projects/visualc7 (Simon-pierre).
1581 
1582 Version 1.2.8rc5 [November 29, 2004]
1583   Fixed new typo in scripts/pngw32.rc
1584 
1585 Version 1.2.8 [December 3, 2004]
1586   Removed projects/visualc7, added projects/visualc71.
1587 
1588 Version 1.2.9beta1 [February 21, 2006]
1589   Initialized some structure members in pngwutil.c to avoid gcc-4.0.0 complaints
1590   Revised man page and libpng.txt to make it clear that one should not call
1591     png_read_end or png_write_end after png_read_png or png_write_png.
1592   Updated references to png-mng-implement mailing list.
1593   Fixed an incorrect typecast in pngrutil.c
1594   Added PNG_NO_READ_SUPPORTED conditional for making a write-only library.
1595   Added PNG_NO_WRITE_INTERLACING_SUPPORTED conditional.
1596   Optimized alpha-inversion loops in pngwtran.c
1597   Moved test for nonzero gamma outside of png_build_gamma_table() in pngrtran.c
1598   Make sure num_trans is <= 256 before copying data in png_set_tRNS().
1599   Make sure num_palette is <= 256 before copying data in png_set_PLTE().
1600   Interchanged order of write_swap_alpha and write_invert_alpha transforms.
1601   Added parentheses in the definition of PNG_LIBPNG_BUILD_TYPE (Cosmin).
1602   Optimized zlib window flag (CINFO) in contrib/pngsuite/*.png (Cosmin).
1603   Updated scripts/makefile.bc32 for Borland C++ 5.6 (Cosmin).
1604   Exported png_get_uint_32, png_save_uint_32, png_get_uint_16, png_save_uint_16,
1605     png_get_int_32, png_save_int_32, png_get_uint_31 (Cosmin).
1606   Added type cast (png_byte) in png_write_sCAL() (Cosmin).
1607   Fixed scripts/makefile.cygwin (Christian Biesinger, Cosmin).
1608   Default iTXt support was inadvertently enabled.
1609 
1610 Version 1.2.9beta2 [February 21, 2006]
1611   Check for png_rgb_to_gray and png_gray_to_rgb read transformations before
1612     checking for png_read_dither in pngrtran.c
1613   Revised checking of chromaticity limits to accommodate extended RGB
1614     colorspace (John Denker).
1615   Changed line endings in some of the project files to CRLF, even in the
1616     "Unix" tar distributions (Cosmin).
1617   Made png_get_int_32 and png_save_int_32 always available (Cosmin).
1618   Updated scripts/pngos2.def, scripts/pngw32.def and projects/wince/png32ce.def
1619     with the newly exported functions.
1620   Eliminated distributions without the "configure" script.
1621   Updated INSTALL instructions.
1622 
1623 Version 1.2.9beta3 [February 24, 2006]
1624   Fixed CRCRLF line endings in contrib/visupng/VisualPng.dsp
1625   Made libpng.pc respect EXEC_PREFIX (D. P. Kreil, J. Bowler)
1626   Removed reference to pngasmrd.h from Makefile.am
1627   Renamed CHANGES to ChangeLog.
1628   Renamed LICENSE to COPYING.
1629   Renamed ANNOUNCE to NEWS.
1630   Created AUTHORS file.
1631 
1632 Version 1.2.9beta4 [March 3, 2006]
1633   Changed definition of PKGCONFIG from $prefix/lib to $libdir in configure.ac
1634   Reverted to filenames LICENSE and ANNOUNCE; removed AUTHORS and COPYING.
1635   Removed newline from the end of some error and warning messages.
1636   Removed test for sqrt() from configure.ac and configure.
1637   Made swap tables in pngtrans.c PNG_CONST (Carlo Bramix).
1638   Disabled default iTXt support that was inadvertently enabled in
1639     libpng-1.2.9beta1.
1640   Added "OS2" to list of systems that don't need underscores, in pnggccrd.c
1641   Removed libpng version and date from *.c files.
1642 
1643 Version 1.2.9beta5 [March 4, 2006]
1644   Removed trailing blanks from source files.
1645   Put version and date of latest change in each source file, and changed
1646     copyright year accordingly.
1647   More cleanup of configure.ac, Makefile.am, and associated scripts.
1648   Restored scripts/makefile.elf which was inadvertently deleted.
1649 
1650 Version 1.2.9beta6 [March 6, 2006]
1651   Fixed typo (RELEASE) in configuration files.
1652 
1653 Version 1.2.9beta7 [March 7, 2006]
1654   Removed libpng.vers and libpng.sym from libpng12_la_SOURCES in Makefile.am
1655   Fixed inconsistent #ifdef's around png_sig_bytes() and png_set_sCAL_s()
1656     in png.h.
1657   Updated makefile.elf as suggested by debian.
1658   Made cosmetic changes to some makefiles, adding LN_SF and other macros.
1659   Made some makefiles accept "exec_prefix".
1660 
1661 Version 1.2.9beta8 [March 9, 2006]
1662   Fixed some "#if defined (..." which should be "#if defined(..."
1663     Bug introduced in libpng-1.2.8.
1664   Fixed inconsistency in definition of png_default_read_data()
1665   Restored blank that was lost from makefile.sggcc "clean" target in beta7.
1666   Revised calculation of "current" and "major" for irix in ltmain.sh
1667   Changed "mkdir" to "MKDIR_P" in some makefiles.
1668   Separated PNG_EXPAND and PNG_EXPAND_tRNS.
1669   Added png_set_expand_gray_1_2_4_to_8() and deprecated
1670     png_set_gray_1_2_4_to_8() which also expands tRNS to alpha.
1671 
1672 Version 1.2.9beta9 [March 10, 2006]
1673   Include "config.h" in pngconf.h when available.
1674   Added some checks for NULL png_ptr or NULL info_ptr (timeless)
1675 
1676 Version 1.2.9beta10 [March 20, 2006]
1677   Removed extra CR from contrib/visualpng/VisualPng.dsw (Cosmin)
1678   Made pnggccrd.c PIC-compliant (Christian Aichinger).
1679   Added makefile.mingw (Wolfgang Glas).
1680   Revised pngconf.h MMX checking.
1681 
1682 Version 1.2.9beta11 [March 22, 2006]
1683   Fixed out-of-order declaration in pngwrite.c that was introduced in beta9
1684   Simplified some makefiles by using LIBSO, LIBSOMAJ, and LIBSOVER macros.
1685 
1686 Version 1.2.9rc1 [March 31, 2006]
1687   Defined PNG_USER_PRIVATEBUILD when including "pngusr.h" (Cosmin).
1688   Removed nonsensical assertion check from pngtest.c (Cosmin).
1689 
1690 Version 1.2.9 [April 14, 2006]
1691   Revised makefile.beos and added "none" selector in ltmain.sh
1692 
1693 Version 1.2.10beta1 [April 15, 2006]
1694   Renamed "config.h" to "png_conf.h" and revised Makefile.am to add
1695     -DPNG_BUILDING_LIBPNG to compile directive, and modified pngconf.h
1696     to include png_conf.h only when PNG_BUILDING_LIBPNG is defined.
1697 
1698 Version 1.2.10beta2 [April 15, 2006]
1699   Manually updated Makefile.in and configure.  Changed png_conf.h.in
1700     back to config.h.
1701 
1702 Version 1.2.10beta3 [April 15, 2006]
1703   Change png_conf.h back to config.h in pngconf.h.
1704 
1705 Version 1.2.10beta4 [April 16, 2006]
1706   Change PNG_BUILDING_LIBPNG to PNG_CONFIGURE_LIBPNG in config/Makefile*.
1707 
1708 Version 1.2.10beta5 [April 16, 2006]
1709   Added a configure check for compiling assembler code in pnggccrd.c
1710 
1711 Version 1.2.10beta6 [April 17, 2006]
1712   Revised the configure check for pnggccrd.c
1713   Moved -DPNG_CONFIGURE_LIBPNG into @LIBPNG_DEFINES@
1714   Added @LIBPNG_DEFINES@ to arguments when building libpng.sym
1715 
1716 Version 1.2.10beta7 [April 18, 2006]
1717   Change "exec_prefix=$prefix" to "exec_prefix=$(prefix)" in makefiles.
1718 
1719 Version 1.2.10rc1 [April 19, 2006]
1720   Ensure pngconf.h doesn't define both PNG_USE_PNGGCCRD and PNG_USE_PNGVCRD
1721   Fixed "LN_FS" typo in makefile.sco and makefile.solaris.
1722 
1723 Version 1.2.10rc2 [April 20, 2006]
1724   Added a backslash between -DPNG_CONFIGURE_LIBPNG and -DPNG_NO_ASSEMBLER_CODE
1725    in configure.ac and configure
1726   Made the configure warning about versioned symbols less arrogant.
1727 
1728 Version 1.2.10rc3 [April 21, 2006]
1729   Added a note in libpng.txt that png_set_sig_bytes(8) can be used when
1730     writing an embedded PNG without the 8-byte signature.
1731   Revised makefiles and configure to avoid making links to libpng.so.*
1732 
1733 Version 1.2.10 [April 23, 2006]
1734   Reverted configure to "rc2" state.
1735 
1736 Version 1.2.11beta1 [May 31, 2006]
1737   scripts/libpng.pc.in contained "configure" style version info and would
1738     not work with makefiles.
1739   The shared-library makefiles were linking to libpng.so.0 instead of
1740     libpng.so.3 compatibility as the library.
1741 
1742 Version 1.2.11beta2 [June 2, 2006]
1743   Increased sprintf buffer from 50 to 52 chars in pngrutil.c to avoid
1744     buffer overflow.
1745   Fixed bug in example.c (png_set_palette_rgb -> png_set_palette_to_rgb)
1746 
1747 Version 1.2.11beta3 [June 5, 2006]
1748   Prepended "#! /bin/sh" to ltmail.sh and contrib/pngminus/*.sh (Cosmin).
1749   Removed the accidental leftover Makefile.in~ (Cosmin).
1750   Avoided potential buffer overflow and optimized buffer in
1751     png_write_sCAL(), png_write_sCAL_s() (Cosmin).
1752   Removed the include directories and libraries from CFLAGS and LDFLAGS
1753     in scripts/makefile.gcc (Nelson A. de Oliveira, Cosmin).
1754 
1755 Version 1.2.11beta4 [June 6, 2006]
1756   Allow zero-length IDAT chunks after the entire zlib datastream, but not
1757     after another intervening chunk type.
1758 
1759 Version 1.0.19rc1, 1.2.11rc1 [June 13, 2006]
1760   Deleted extraneous square brackets from [config.h] in configure.ac
1761 
1762 Version 1.0.19rc2, 1.2.11rc2 [June 14, 2006]
1763   Added prototypes for PNG_INCH_CONVERSIONS functions to png.h
1764   Revised INSTALL and autogen.sh
1765   Fixed typo in several makefiles (-W1 should be -Wl)
1766   Added typedef for png_int_32 and png_uint_32 on 64-bit systems.
1767 
1768 Version 1.0.19rc3, 1.2.11rc3 [June 15, 2006]
1769   Removed the new typedefs for 64-bit systems (delay until version 1.4.0)
1770   Added one zero element to png_gamma_shift[] array in pngrtran.c to avoid
1771     reading out of bounds.
1772 
1773 Version 1.0.19rc4, 1.2.11rc4 [June 15, 2006]
1774   Really removed the new typedefs for 64-bit systems.
1775 
1776 Version 1.0.19rc5, 1.2.11rc5 [June 22, 2006]
1777   Removed png_sig_bytes entry from scripts/pngw32.def
1778 
1779 Version 1.0.19, 1.2.11 [June 26, 2006]
1780   None.
1781 
1782 Version 1.0.20, 1.2.12 [June 27, 2006]
1783   Really increased sprintf buffer from 50 to 52 chars in pngrutil.c to avoid
1784     buffer overflow.
1785 
1786 Version 1.2.13beta1 [October 2, 2006]
1787   Removed AC_FUNC_MALLOC from configure.ac
1788   Work around Intel-Mac compiler bug by setting PNG_NO_MMX_CODE in pngconf.h
1789   Change "logical" to "bitwise" throughout documentation.
1790   Detect and fix attempt to write wrong iCCP profile length (CVE-2006-7244)
1791 
1792 Version 1.0.21, 1.2.13 [November 14, 2006]
1793   Fix potential buffer overflow in sPLT chunk handler.
1794   Fix Makefile.am to not try to link to noexistent files.
1795   Check all exported functions for NULL png_ptr.
1796 
1797 Version 1.2.14beta1 [November 17, 2006]
1798   Relocated three misplaced tests for NULL png_ptr.
1799   Built Makefile.in with automake-1.9.6 instead of 1.9.2.
1800   Build configure with autoconf-2.60 instead of 2.59
1801 
1802 Version 1.2.14beta2 [November 17, 2006]
1803   Added some typecasts in png_zalloc().
1804 
1805 Version 1.2.14rc1 [November 20, 2006]
1806   Changed "strtod" to "png_strtod" in pngrutil.c
1807 
1808 Version 1.0.22, 1.2.14    [November 27, 2006]
1809   Added missing "$(srcdir)" in Makefile.am and Makefile.in
1810 
1811 Version 1.2.15beta1 [December 3, 2006]
1812   Generated configure with autoconf-2.61 instead of 2.60
1813   Revised configure.ac to update libpng.pc and libpng-config.
1814 
1815 Version 1.2.15beta2 [December 3, 2006]
1816   Always export MMX asm functions, just stubs if not building pnggccrd.c
1817 
1818 Version 1.2.15beta3 [December 4, 2006]
1819   Add "png_bytep" typecast to profile while calculating length in pngwutil.c
1820 
1821 Version 1.2.15beta4 [December 7, 2006]
1822   Added scripts/CMakeLists.txt
1823   Changed PNG_NO_ASSEMBLER_CODE to PNG_NO_MMX_CODE in scripts, like 1.4.0beta
1824 
1825 Version 1.2.15beta5 [December 7, 2006]
1826   Changed some instances of PNG_ASSEMBLER_* to PNG_MMX_* in pnggccrd.c
1827   Revised scripts/CMakeLists.txt
1828 
1829 Version 1.2.15beta6 [December 13, 2006]
1830   Revised scripts/CMakeLists.txt and configure.ac
1831 
1832 Version 1.2.15rc1 [December 18, 2006]
1833   Revised scripts/CMakeLists.txt
1834 
1835 Version 1.2.15rc2 [December 21, 2006]
1836   Added conditional #undef jmpbuf in pngtest.c to undo #define in AIX headers.
1837   Added scripts/makefile.nommx
1838 
1839 Version 1.2.15rc3 [December 25, 2006]
1840   Fixed shared library numbering error that was introduced in 1.2.15beta6.
1841 
1842 Version 1.2.15rc4 [December 27, 2006]
1843   Fixed handling of rgb_to_gray when png_ptr->color.gray isn't set.
1844 
1845 Version 1.2.15rc5 [December 31, 2006]
1846   Revised handling of rgb_to_gray.
1847 
1848 Version 1.2.15 [January 5, 2007]
1849   Added some (unsigned long) typecasts in pngtest.c to avoid printing errors.
1850 
1851 Version 1.2.16beta1 [January 6, 2007]
1852   Fix bugs in makefile.nommx
1853 
1854 Version 1.2.16beta2 [January 16, 2007]
1855   Revised scripts/CMakeLists.txt
1856 
1857 Version 1.2.16 [January 31, 2007]
1858   No changes.
1859 
1860 Version 1.2.17beta1 [March 6, 2007]
1861   Revised scripts/CMakeLists.txt to install both shared and static libraries.
1862   Deleted a redundant line from pngset.c.
1863 
1864 Version 1.2.17beta2 [April 26, 2007]
1865   Relocated misplaced test for png_ptr == NULL in pngpread.c
1866   Change "==" to "&" for testing PNG_RGB_TO_GRAY_ERR & PNG_RGB_TO_GRAY_WARN
1867     flags.
1868   Changed remaining instances of PNG_ASSEMBLER_* to PNG_MMX_*
1869   Added pngerror() when write_IHDR fails in deflateInit2().
1870   Added "const" to some array declarations.
1871   Mention examples of libpng usage in the libpng*.txt and libpng.3 documents.
1872 
1873 Version 1.2.17rc1 [May 4, 2007]
1874   No changes.
1875 
1876 Version 1.2.17rc2 [May 8, 2007]
1877   Moved several PNG_HAVE_* macros out of PNG_INTERNAL because applications
1878     calling set_unknown_chunk_location() need them.
1879   Changed transformation flag from PNG_EXPAND_tRNS to PNG_EXPAND in
1880     png_set_expand_gray_1_2_4_to_8().
1881   Added png_ptr->unknown_chunk to hold working unknown chunk data, so it
1882     can be free'ed in case of error.  Revised unknown chunk handling in
1883     pngrutil.c and pngpread.c to use this structure.
1884 
1885 Version 1.2.17rc3 [May 8, 2007]
1886   Revised symbol-handling in configure script.
1887 
1888 Version 1.2.17rc4 [May 10, 2007]
1889   Revised unknown chunk handling to avoid storing unknown critical chunks.
1890 
1891 Version 1.0.25 [May 15, 2007]
1892 Version 1.2.17 [May 15, 2007]
1893   Added "png_ptr->num_trans=0" before error return in png_handle_tRNS,
1894     to eliminate a vulnerability (CVE-2007-2445, CERT VU#684664)
1895 
1896 Version 1.0.26 [May 15, 2007]
1897 Version 1.2.18 [May 15, 2007]
1898   Reverted the libpng-1.2.17rc3 change to symbol-handling in configure script
1899 
1900 Version 1.2.19beta1 [May 18, 2007]
1901   Changed "const static" to "static PNG_CONST" everywhere, mostly undoing
1902     change of libpng-1.2.17beta2.  Changed other "const" to "PNG_CONST"
1903   Changed some handling of unused parameters, to avoid compiler warnings.
1904     "if (unused == NULL) return;" becomes "unused = unused".
1905 
1906 Version 1.2.19beta2 [May 18, 2007]
1907   Only use the valid bits of tRNS value in png_do_expand() (Brian Cartier)
1908 
1909 Version 1.2.19beta3 [May 19, 2007]
1910   Add some "png_byte" typecasts in png_check_keyword() and write new_key
1911   instead of key in zTXt chunk (Kevin Ryde).
1912 
1913 Version 1.2.19beta4 [May 21, 2007]
1914   Add png_snprintf() function and use it in place of sprint() for improved
1915     defense against buffer overflows.
1916 
1917 Version 1.2.19beta5 [May 21, 2007]
1918   Fixed png_handle_tRNS() to only use the valid bits of tRNS value.
1919   Changed handling of more unused parameters, to avoid compiler warnings.
1920   Removed some PNG_CONST in pngwutil.c to avoid compiler warnings.
1921 
1922 Version 1.2.19beta6 [May 22, 2007]
1923   Added some #ifdef PNG_MMX_CODE_SUPPORTED where needed in pngvcrd.c
1924   Added a special "_MSC_VER" case that defines png_snprintf to _snprintf
1925 
1926 Version 1.2.19beta7 [May 22, 2007]
1927   Squelched png_squelch_warnings() in pnggccrd.c and added
1928     an #ifdef PNG_MMX_CODE_SUPPORTED block around the declarations that caused
1929     the warnings that png_squelch_warnings was squelching.
1930 
1931 Version 1.2.19beta8 [May 22, 2007]
1932   Removed __MMX__ from test in pngconf.h.
1933 
1934 Version 1.2.19beta9 [May 23, 2007]
1935   Made png_squelch_warnings() available via PNG_SQUELCH_WARNINGS macro.
1936   Revised png_squelch_warnings() so it might work.
1937   Updated makefile.sgcc and makefile.solaris; added makefile.solaris-x86.
1938 
1939 Version 1.2.19beta10 [May 24, 2007]
1940   Resquelched png_squelch_warnings(), use "__attribute__((used))" instead.
1941 
1942 Version 1.4.0beta1 [April 20, 2006]
1943   Enabled iTXt support (changes png_struct, thus requires so-number change).
1944   Cleaned up PNG_ASSEMBLER_CODE_SUPPORTED vs PNG_MMX_CODE_SUPPORTED
1945   Eliminated PNG_1_0_X and PNG_1_2_X macros.
1946   Removed deprecated functions png_read_init, png_write_init, png_info_init,
1947     png_permit_empty_plte, png_set_gray_1_2_4_to_8, png_check_sig, and
1948     removed the deprecated macro PNG_MAX_UINT.
1949   Moved "PNG_INTERNAL" parts of png.h and pngconf.h into pngintrn.h
1950   Removed many WIN32_WCE #ifdefs (Cosmin).
1951   Reduced dependency on C-runtime library when on Windows (Simon-Pierre)
1952   Replaced sprintf() with png_sprintf() (Simon-Pierre)
1953 
1954 Version 1.4.0beta2 [April 20, 2006]
1955   Revised makefiles and configure to avoid making links to libpng.so.*
1956   Moved some leftover MMX-related defines from pngconf.h to pngintrn.h
1957   Updated scripts/pngos2.def, pngw32.def, and projects/wince/png32ce.def
1958 
1959 Version 1.4.0beta3 [May 10, 2006]
1960   Updated scripts/pngw32.def to comment out MMX functions.
1961   Added PNG_NO_GET_INT_32 and PNG_NO_SAVE_INT_32 macros.
1962   Scripts/libpng.pc.in contained "configure" style version info and would
1963     not work with makefiles.
1964   Revised pngconf.h and added pngconf.h.in, so makefiles and configure can
1965     pass defines to libpng and applications.
1966 
1967 Version 1.4.0beta4 [May 11, 2006]
1968   Revised configure.ac, Makefile.am, and many of the makefiles to write
1969     their defines in pngconf.h.
1970 
1971 Version 1.4.0beta5 [May 15, 2006]
1972   Added a missing semicolon in Makefile.am and Makefile.in
1973   Deleted extraneous square brackets from configure.ac
1974 
1975 Version 1.4.0beta6 [June 2, 2006]
1976   Increased sprintf buffer from 50 to 52 chars in pngrutil.c to avoid
1977     buffer overflow.
1978   Changed sonum from 0 to 1.
1979   Removed unused prototype for png_check_sig() from png.h
1980 
1981 Version 1.4.0beta7 [June 16, 2006]
1982   Exported png_write_sig (Cosmin).
1983   Optimized buffer in png_handle_cHRM() (Cosmin).
1984   Set pHYs = 2835 x 2835 pixels per meter, and added
1985     sCAL = 0.352778e-3 x 0.352778e-3 meters, in pngtest.png (Cosmin).
1986   Added png_set_benign_errors(), png_benign_error(), png_chunk_benign_error().
1987   Added typedef for png_int_32 and png_uint_32 on 64-bit systems.
1988   Added "(unsigned long)" typecast on png_uint_32 variables in printf lists.
1989 
1990 Version 1.4.0beta8 [June 22, 2006]
1991   Added demonstration of user chunk support in pngtest.c, to support the
1992     public sTER chunk and a private vpAg chunk.
1993 
1994 Version 1.4.0beta9 [July 3, 2006]
1995   Removed ordinals from scripts/pngw32.def and removed png_info_int and
1996     png_set_gray_1_2_4_to_8 entries.
1997   Inline call of png_get_uint_32() in png_get_uint_31().
1998   Use png_get_uint_31() to get vpAg width and height in pngtest.c
1999   Removed WINCE and Netware projects.
2000   Removed standalone Y2KINFO file.
2001 
2002 Version 1.4.0beta10 [July 12, 2006]
2003   Eliminated automatic copy of pngconf.h to pngconf.h.in from configure and
2004     some makefiles, because it was not working reliably.  Instead, distribute
2005     pngconf.h.in along with pngconf.h and cause configure and some of the
2006     makefiles to update pngconf.h from pngconf.h.in.
2007   Added pngconf.h to DEPENDENCIES in Makefile.am
2008 
2009 Version 1.4.0beta11 [August 19, 2006]
2010   Removed AC_FUNC_MALLOC from configure.ac.
2011   Added a warning when writing iCCP profile with mismatched profile length.
2012   Patched pnggccrd.c to assemble on x86_64 platforms.
2013   Moved chunk header reading into a separate function png_read_chunk_header()
2014     in pngrutil.c.  The chunk header (len+sig) is now serialized in a single
2015     operation (Cosmin).
2016   Implemented support for I/O states. Added png_ptr member io_state, and
2017     functions png_get_io_chunk_name() and png_get_io_state() in pngget.c
2018     (Cosmin).
2019   Added png_get_io_chunk_name and png_get_io_state to scripts/*.def (Cosmin).
2020   Renamed scripts/pngw32.* to scripts/pngwin.* (Cosmin).
2021   Removed the include directories and libraries from CFLAGS and LDFLAGS
2022     in scripts/makefile.gcc (Cosmin).
2023   Used png_save_uint_32() to set vpAg width and height in pngtest.c (Cosmin).
2024   Cast to proper type when getting/setting vpAg units in pngtest.c (Cosmin).
2025   Added pngintrn.h to the Visual C++ projects (Cosmin).
2026   Removed scripts/list (Cosmin).
2027   Updated copyright year in scripts/pngwin.def (Cosmin).
2028   Removed PNG_TYPECAST_NULL and used standard NULL consistently (Cosmin).
2029   Disallowed the user to redefine png_size_t, and enforced a consistent use
2030     of png_size_t across libpng (Cosmin).
2031   Changed the type of png_ptr->rowbytes, PNG_ROWBYTES() and friends
2032     to png_size_t (Cosmin).
2033   Removed png_convert_size() and replaced png_sizeof with sizeof (Cosmin).
2034   Removed some unnecessary type casts (Cosmin).
2035   Changed prototype of png_get_compression_buffer_size() and
2036     png_set_compression_buffer_size() to work with png_size_t instead of
2037     png_uint_32 (Cosmin).
2038   Removed png_memcpy_check() and png_memset_check() (Cosmin).
2039   Fixed a typo (png_byte --> png_bytep) in libpng.3 and libpng.txt (Cosmin).
2040   Clarified that png_zalloc() does not clear the allocated memory,
2041     and png_zalloc() and png_zfree() cannot be PNGAPI (Cosmin).
2042   Renamed png_mem_size_t to png_alloc_size_t, fixed its definition in
2043     pngconf.h, and used it in all memory allocation functions (Cosmin).
2044   Renamed pngintrn.h to pngpriv.h, added a comment at the top of the file
2045     mentioning that the symbols declared in that file are private, and
2046     updated the scripts and the Visual C++ projects accordingly (Cosmin).
2047   Removed circular references between pngconf.h and pngconf.h.in in
2048     scripts/makefile.vc*win32 (Cosmin).
2049   Removing trailing '.' from the warning and error messages (Cosmin).
2050   Added pngdefs.h that is built by makefile or configure, instead of
2051     pngconf.h.in (Glenn).
2052   Detect and fix attempt to write wrong iCCP profile length.
2053 
2054 Version 1.4.0beta12 [October 19, 2006]
2055   Changed "logical" to "bitwise" in the documentation.
2056   Work around Intel-Mac compiler bug by setting PNG_NO_MMX_CODE in pngconf.h
2057   Add a typecast to stifle compiler warning in pngrutil.c
2058 
2059 Version 1.4.0beta13 [November 10, 2006]
2060   Fix potential buffer overflow in sPLT chunk handler.
2061   Fix Makefile.am to not try to link to noexistent files.
2062 
2063 Version 1.4.0beta14 [November 15, 2006]
2064   Check all exported functions for NULL png_ptr.
2065 
2066 Version 1.4.0beta15 [November 17, 2006]
2067   Relocated two misplaced tests for NULL png_ptr.
2068   Built Makefile.in with automake-1.9.6 instead of 1.9.2.
2069   Build configure with autoconf-2.60 instead of 2.59
2070   Add "install: all" in Makefile.am so "configure; make install" will work.
2071 
2072 Version 1.4.0beta16 [November 17, 2006]
2073   Added a typecast in png_zalloc().
2074 
2075 Version 1.4.0beta17 [December 4, 2006]
2076   Changed "new_key[79] = '\0';" to "(*new_key)[79] = '\0';" in pngwutil.c
2077   Add "png_bytep" typecast to profile while calculating length in pngwutil.c
2078 
2079 Version 1.4.0beta18 [December 7, 2006]
2080   Added scripts/CMakeLists.txt
2081 
2082 Version 1.4.0beta19 [May 16, 2007]
2083   Revised scripts/CMakeLists.txt
2084   Rebuilt configure and Makefile.in with newer tools.
2085   Added conditional #undef jmpbuf in pngtest.c to undo #define in AIX headers.
2086   Added scripts/makefile.nommx
2087 
2088 Version 1.4.0beta20 [July 9, 2008]
2089   Moved several PNG_HAVE_* macros from pngpriv.h to png.h because applications
2090     calling set_unknown_chunk_location() need them.
2091   Moved several macro definitions from pngpriv.h to pngconf.h
2092   Merge with changes to the 1.2.X branch, as of 1.2.30beta04.
2093   Deleted all use of the MMX assembler code and Intel-licensed optimizations.
2094   Revised makefile.mingw
2095 
2096 Version 1.4.0beta21 [July 21, 2008]
2097   Moved local array "chunkdata" from pngrutil.c to the png_struct, so
2098     it will be freed by png_read_destroy() in case of a read error (Kurt
2099     Christensen).
2100 
2101 Version 1.4.0beta22 [July 21, 2008]
2102   Change "purpose" and "buffer" to png_ptr->chunkdata to avoid memory leaking.
2103 
2104 Version 1.4.0beta23 [July 22, 2008]
2105   Change "chunkdata = NULL" to "png_ptr->chunkdata = NULL" several places in
2106     png_decompress_chunk().
2107 
2108 Version 1.4.0beta24 [July 25, 2008]
2109   Change all remaining "chunkdata" to "png_ptr->chunkdata" in
2110     png_decompress_chunk(), and remove "chunkdata" from parameter list.
2111   Put a call to png_check_chunk_name() in png_read_chunk_header().
2112   Revised png_check_chunk_name() to reject a name with a lowercase 3rd byte.
2113   Removed two calls to png_check_chunk_name() occurring later in the process.
2114   Define PNG_NO_ERROR_NUMBERS by default in pngconf.h
2115 
2116 Version 1.4.0beta25 [July 30, 2008]
2117   Added a call to png_check_chunk_name() in pngpread.c
2118   Reverted png_check_chunk_name() to accept a name with a lowercase 3rd byte.
2119   Added png_push_have_buffer() function to pngpread.c
2120   Eliminated PNG_BIG_ENDIAN_SUPPORTED and associated png_get_* macros.
2121   Made inline expansion of png_get_*() optional with PNG_USE_READ_MACROS.
2122   Eliminated all PNG_USELESS_TESTS and PNG_CORRECT_PALETTE_SUPPORTED code.
2123   Synced contrib directory and configure files with libpng-1.2.30beta06.
2124   Eliminated no-longer-used pngdefs.h (but it's still built in the makefiles)
2125   Relocated a misplaced "#endif /* PNG_NO_WRITE_FILTER */" in pngwutil.c
2126 
2127 Version 1.4.0beta26 [August 4, 2008]
2128   Removed png_push_have_buffer() function in pngpread.c.  It increased the
2129     compiled library size slightly.
2130   Changed "-Wall" to "-W -Wall" in the CFLAGS in all makefiles (Cosmin Truta)
2131   Declared png_ptr "volatile" in pngread.c and pngwrite.c to avoid warnings.
2132   Updated contrib/visupng/cexcept.h to version 2.0.1
2133   Added PNG_LITERAL_CHARACTER macros for #, [, and ].
2134 
2135 Version 1.4.0beta27 [August 5, 2008]
2136   Revised usage of PNG_LITERAL_SHARP in pngerror.c.
2137   Moved newline character from individual png_debug messages into the
2138     png_debug macros.
2139   Allow user to #define their own png_debug, png_debug1, and png_debug2.
2140 
2141 Version 1.4.0beta28 [August 5, 2008]
2142   Revised usage of PNG_LITERAL_SHARP in pngerror.c.
2143   Added PNG_STRING_NEWLINE macro
2144 
2145 Version 1.4.0beta29 [August 9, 2008]
2146   Revised usage of PNG_STRING_NEWLINE to work on non-ISO compilers.
2147   Added PNG_STRING_COPYRIGHT macro.
2148   Added non-ISO versions of png_debug macros.
2149 
2150 Version 1.4.0beta30 [August 14, 2008]
2151   Added premultiplied alpha feature (Volker Wiendl).
2152 
2153 Version 1.4.0beta31 [August 18, 2008]
2154   Moved png_set_premultiply_alpha from pngtrans.c to pngrtran.c
2155   Removed extra crc check at the end of png_handle_cHRM().  Bug introduced
2156     in libpng-1.4.0beta20.
2157 
2158 Version 1.4.0beta32 [August 19, 2008]
2159   Added PNG_WRITE_FLUSH_SUPPORTED block around new png_flush() call.
2160   Revised PNG_NO_STDIO version of png_write_flush()
2161 
2162 Version 1.4.0beta33 [August 20, 2008]
2163   Added png_get|set_chunk_cache_max() to limit the total number of sPLT,
2164     text, and unknown chunks that can be stored.
2165 
2166 Version 1.4.0beta34 [September 6, 2008]
2167   Shortened tIME_string to 29 bytes in pngtest.c
2168   Fixed off-by-one error introduced in png_push_read_zTXt() function in
2169     libpng-1.2.30beta04/pngpread.c (Harald van Dijk)
2170 
2171 Version 1.4.0beta35 [October 6, 2008]
2172   Changed "trans_values" to "trans_color".
2173   Changed so-number from 0 to 14.  Some OS do not like 0.
2174   Revised makefile.darwin to fix shared library numbering.
2175   Change png_set_gray_1_2_4_to_8() to png_set_expand_gray_1_2_4_to_8()
2176     in example.c (debian bug report)
2177 
2178 Version 1.4.0beta36 [October 25, 2008]
2179   Sync with tEXt vulnerability fix in libpng-1.2.33rc02.
2180 
2181 Version 1.4.0beta37 [November 13, 2008]
2182   Added png_check_cHRM in png.c and moved checking from pngget.c, pngrutil.c,
2183     and pngwrite.c
2184 
2185 Version 1.4.0beta38 [November 22, 2008]
2186   Added check for zero-area RGB cHRM triangle in png_check_cHRM() and
2187     png_check_cHRM_fixed().
2188 
2189 Version 1.4.0beta39 [November 23, 2008]
2190   Revised png_warning() to write its message on standard output by default
2191     when warning_fn is NULL.
2192 
2193 Version 1.4.0beta40 [November 24, 2008]
2194   Eliminated png_check_cHRM().  Instead, always use png_check_cHRM_fixed().
2195   In png_check_cHRM_fixed(), ensure white_y is > 0, and removed redundant
2196     check for all-zero coordinates that is detected by the triangle check.
2197 
2198 Version 1.4.0beta41 [November 26, 2008]
2199   Fixed string vs pointer-to-string error in png_check_keyword().
2200   Rearranged test expressions in png_check_cHRM_fixed() to avoid internal
2201     overflows.
2202   Added PNG_NO_CHECK_cHRM conditional.
2203 
2204 Version 1.4.0beta42, 43 [December 1, 2008]
2205   Merge png_debug with version 1.2.34beta04.
2206 
2207 Version 1.4.0beta44 [December 6, 2008]
2208   Removed redundant check for key==NULL before calling png_check_keyword()
2209     to ensure that new_key gets initialized and removed extra warning
2210     (Merge with version 1.2.34beta05 -- Arvan Pritchard).
2211 
2212 Version 1.4.0beta45 [December 9, 2008]
2213   In png_write_png(), respect the placement of the filler bytes in an earlier
2214     call to png_set_filler() (Jim Barry).
2215 
2216 Version 1.4.0beta46 [December 10, 2008]
2217   Undid previous change and added PNG_TRANSFORM_STRIP_FILLER_BEFORE and
2218     PNG_TRANSFORM_STRIP_FILLER_AFTER conditionals and deprecated
2219     PNG_TRANSFORM_STRIP_FILLER (Jim Barry).
2220 
2221 Version 1.4.0beta47 [December 15, 2008]
2222   Support for dithering was disabled by default, because it has never
2223     been well tested and doesn't work very well.  The code has not
2224     been removed, however, and can be enabled by building libpng with
2225     PNG_READ_DITHER_SUPPORTED defined.
2226 
2227 Version 1.4.0beta48 [February 14, 2009]
2228   Added new exported function png_calloc().
2229   Combined several instances of png_malloc(); png_memset() into png_calloc().
2230   Removed prototype for png_freeptr() that was added in libpng-1.4.0beta24
2231     but was never defined.
2232 
2233 Version 1.4.0beta49 [February 28, 2009]
2234   Added png_fileno() macro to pngconf.h, used in pngwio.c
2235   Corrected order of #ifdef's in png_debug definition in png.h
2236   Fixed bug introduced in libpng-1.4.0beta48 with the memset arguments
2237     for pcal_params.
2238   Fixed order of #ifdef directives in the png_debug defines in png.h
2239     (bug introduced in libpng-1.2.34/1.4.0beta29).
2240   Revised comments in png_set_read_fn() and png_set_write_fn().
2241 
2242 Version 1.4.0beta50 [March 18, 2009]
2243   Use png_calloc() instead of png_malloc() to allocate big_row_buf when
2244     reading an interlaced file, to avoid a possible UMR.
2245   Undid revision of PNG_NO_STDIO version of png_write_flush().  Users
2246     having trouble with fflush() can build with PNG_NO_WRITE_FLUSH defined
2247     or supply their own flush_fn() replacement.
2248   Revised libpng*.txt and png.h documentation about use of png_write_flush()
2249     and png_set_write_fn().
2250   Removed fflush() from pngtest.c.
2251   Added "#define PNG_NO_WRITE_FLUSH" to contrib/pngminim/encoder/pngusr.h
2252 
2253 Version 1.4.0beta51 [March 21, 2009]
2254   Removed new png_fileno() macro from pngconf.h .
2255 
2256 Version 1.4.0beta52 [March 27, 2009]
2257   Relocated png_do_chop() ahead of building gamma tables in pngrtran.c
2258     This avoids building 16-bit gamma tables unnecessarily.
2259   Removed fflush() from pngtest.c.
2260   Added "#define PNG_NO_WRITE_FLUSH" to contrib/pngminim/encoder/pngusr.h
2261   Added a section on differences between 1.0.x and 1.2.x to libpng.3/libpng.txt
2262 
2263 Version 1.4.0beta53 [April 1, 2009]
2264   Removed some remaining MMX macros from pngpriv.h
2265   Fixed potential memory leak of "new_name" in png_write_iCCP() (Ralph Giles)
2266 
2267 Version 1.4.0beta54 [April 13, 2009]
2268   Added "ifndef PNG_SKIP_SETJMP_CHECK" block in pngconf.h to allow
2269     application code writers to bypass the check for multiple inclusion
2270     of setjmp.h when they know that it is safe to ignore the situation.
2271   Eliminated internal use of setjmp() in pngread.c and pngwrite.c
2272   Reordered ancillary chunks in pngtest.png to be the same as what
2273     pngtest now produces, and made some cosmetic changes to pngtest output.
2274   Eliminated deprecated png_read_init_3() and png_write_init_3() functions.
2275 
2276 Version 1.4.0beta55 [April 15, 2009]
2277   Simplified error handling in pngread.c and pngwrite.c by putting
2278     the new png_read_cleanup() and png_write_cleanup() functions inline.
2279 
2280 Version 1.4.0beta56 [April 25, 2009]
2281   Renamed "user_chunk_data" to "my_user_chunk_data" in pngtest.c to suppress
2282     "shadowed declaration" warning from gcc-4.3.3.
2283   Renamed "gamma" to "png_gamma" in pngset.c to avoid "shadowed declaration"
2284     warning about a global "gamma" variable in math.h on some platforms.
2285 
2286 Version 1.4.0beta57 [May 2, 2009]
2287   Removed prototype for png_freeptr() that was added in libpng-1.4.0beta24
2288     but was never defined (again).
2289   Rebuilt configure scripts with autoconf-2.63 instead of 2.62
2290   Removed pngprefs.h and MMX from makefiles
2291 
2292 Version 1.4.0beta58 [May 14, 2009]
2293   Changed pngw32.def to pngwin.def in makefile.mingw (typo was introduced
2294     in beta57).
2295   Clarified usage of sig_bit versus sig_bit_p in example.c (Vincent Torri)
2296 
2297 Version 1.4.0beta59 [May 15, 2009]
2298   Reformated sources in libpng style (3-space intentation, comment format)
2299   Fixed typo in libpng docs (PNG_FILTER_AVE should be PNG_FILTER_AVG)
2300   Added sections about the git repository and our coding style to the
2301     documentation
2302   Relocated misplaced #endif in pngwrite.c, sCAL chunk handler.
2303 
2304 Version 1.4.0beta60 [May 19, 2009]
2305   Conditionally compile png_read_finish_row() which is not used by
2306     progressive readers.
2307   Added contrib/pngminim/preader to demonstrate building minimal progressive
2308     decoder, based on contrib/gregbook with embedded libpng and zlib.
2309 
2310 Version 1.4.0beta61 [May 20, 2009]
2311   In contrib/pngminim/*, renamed "makefile.std" to "makefile", since there
2312     is only one makefile in those directories, and revised the README files
2313     accordingly.
2314   More reformatting of comments, mostly to capitalize sentences.
2315 
2316 Version 1.4.0beta62 [June 2, 2009]
2317   Added "#define PNG_NO_WRITE_SWAP" to contrib/pngminim/encoder/pngusr.h
2318     and "define PNG_NO_READ_SWAP" to decoder/pngusr.h and preader/pngusr.h
2319   Reformatted several remaining "else statement" into two lines.
2320   Added a section to the libpng documentation about using png_get_io_ptr()
2321     in configure scripts to detect the presence of libpng.
2322 
2323 Version 1.4.0beta63 [June 15, 2009]
2324   Revised libpng*.txt and libpng.3 to mention calling png_set_IHDR()
2325     multiple times and to specify the sample order in the tRNS chunk,
2326     because the ISO PNG specification has a typo in the tRNS table.
2327   Changed several PNG_UNKNOWN_CHUNK_SUPPORTED to
2328     PNG_HANDLE_AS_UNKNOWN_SUPPORTED, to make the png_set_keep mechanism
2329     available for ignoring known chunks even when not saving unknown chunks.
2330   Adopted preference for consistent use of "#ifdef" and "#ifndef" versus
2331     "#if defined()" and "if !defined()" where possible.
2332 
2333 Version 1.4.0beta64 [June 24, 2009]
2334   Eliminated PNG_LEGACY_SUPPORTED code.
2335   Moved the various unknown chunk macro definitions outside of the
2336     PNG_READ|WRITE_ANCILLARY_CHUNK_SUPPORTED blocks.
2337 
2338 Version 1.4.0beta65 [June 26, 2009]
2339   Added a reference to the libpng license in each file.
2340 
2341 Version 1.4.0beta66 [June 27, 2009]
2342   Refer to the libpng license instead of the libpng license in each file.
2343 
2344 Version 1.4.0beta67 [July 6, 2009]
2345   Relocated INVERT_ALPHA within png_read_png() and png_write_png().
2346   Added high-level API transform PNG_TRANSFORM_GRAY_TO_RGB.
2347   Added an "xcode" project to the projects directory (Alam Arias).
2348 
2349 Version 1.4.0beta68 [July 19, 2009]
2350   Avoid some tests in filter selection in pngwutil.c
2351 
2352 Version 1.4.0beta69 [July 25, 2009]
2353   Simplified the new filter-selection test.  This runs faster in the
2354     common "PNG_ALL_FILTERS" and PNG_FILTER_NONE cases.
2355   Removed extraneous declaration from the new call to png_read_gray_to_rgb()
2356     (bug introduced in libpng-1.4.0beta67).
2357   Fixed up xcode project (Alam Arias)
2358   Added a prototype for png_64bit_product() in png.c
2359 
2360 Version 1.4.0beta70 [July 27, 2009]
2361   Avoid a possible NULL dereference in debug build, in png_set_text_2().
2362     (bug introduced in libpng-0.95, discovered by Evan Rouault)
2363 
2364 Version 1.4.0beta71 [July 29, 2009]
2365   Rebuilt configure scripts with autoconf-2.64.
2366 
2367 Version 1.4.0beta72 [August 1, 2009]
2368   Replaced *.tar.lzma with *.tar.xz in distribution.  Get the xz codec
2369     from <http://tukaani.org/xz>.
2370 
2371 Version 1.4.0beta73 [August 1, 2009]
2372   Reject attempt to write iCCP chunk with negative embedded profile length
2373     (JD Chen) (CVE-2009-5063).
2374 
2375 Version 1.4.0beta74 [August 8, 2009]
2376   Changed png_ptr and info_ptr member "trans" to "trans_alpha".
2377 
2378 Version 1.4.0beta75 [August 21, 2009]
2379   Removed an extra png_debug() recently added to png_write_find_filter().
2380   Fixed incorrect #ifdef in pngset.c regarding unknown chunk support.
2381 
2382 Version 1.4.0beta76 [August 22, 2009]
2383   Moved an incorrectly located test in png_read_row() in pngread.c
2384 
2385 Version 1.4.0beta77 [August 27, 2009]
2386   Removed lpXYZ.tar.bz2 (with CRLF), KNOWNBUG, libpng-x.y.z-KNOWNBUG.txt,
2387     and the "noconfig" files from the distribution.
2388   Moved CMakeLists.txt from scripts into the main libpng directory.
2389   Various bugfixes and improvements to CMakeLists.txt (Philip Lowman)
2390 
2391 Version 1.4.0beta78 [August 31, 2009]
2392   Converted all PNG_NO_* tests to PNG_*_SUPPORTED everywhere except pngconf.h
2393   Eliminated PNG_NO_FREE_ME and PNG_FREE_ME_SUPPORTED macros.
2394   Use png_malloc plus a loop instead of png_calloc() to initialize
2395     row_pointers in png_read_png().
2396 
2397 Version 1.4.0beta79 [September 1, 2009]
2398   Eliminated PNG_GLOBAL_ARRAYS and PNG_LOCAL_ARRAYS; always use local arrays.
2399   Eliminated PNG_CALLOC_SUPPORTED macro and always provide png_calloc().
2400 
2401 Version 1.4.0beta80 [September 17, 2009]
2402   Removed scripts/libpng.icc
2403   Changed typecast of filler from png_byte to png_uint_16 in png_set_filler().
2404     (Dennis Gustafsson)
2405   Fixed typo introduced in beta78 in pngtest.c ("#if def " should be "#ifdef ")
2406 
2407 Version 1.4.0beta81 [September 23, 2009]
2408   Eliminated unused PNG_FLAG_FREE_* defines from pngpriv.h
2409   Expanded TAB characters in pngrtran.c
2410   Removed PNG_CONST from all "PNG_CONST PNG_CHNK" declarations to avoid
2411     compiler complaints about doubly declaring things "const".
2412   Changed all "#if [!]defined(X)" to "if[n]def X" where possible.
2413   Eliminated unused png_ptr->row_buf_size
2414 
2415 Version 1.4.0beta82 [September 25, 2009]
2416   Moved redundant IHDR checking into new png_check_IHDR() in png.c
2417     and report all errors found in the IHDR data.
2418   Eliminated useless call to png_check_cHRM() from pngset.c
2419 
2420 Version 1.4.0beta83 [September 25, 2009]
2421   Revised png_check_IHDR() to eliminate bogus complaint about filter_type.
2422 
2423 Version 1.4.0beta84 [September 30, 2009]
2424   Fixed some inconsistent indentation in pngconf.h
2425   Revised png_check_IHDR() to add a test for width variable less than 32-bit.
2426 
2427 Version 1.4.0beta85 [October 1, 2009]
2428   Revised png_check_IHDR() again, to check info_ptr members instead of
2429     the contents of the returned parameters.
2430 
2431 Version 1.4.0beta86 [October 9, 2009]
2432   Updated the "xcode" project (Alam Arias).
2433   Eliminated a shadowed declaration of "pp" in png_handle_sPLT().
2434 
2435 Version 1.4.0rc01 [October 19, 2009]
2436   Trivial cosmetic changes.
2437 
2438 Version 1.4.0beta87 [October 30, 2009]
2439   Moved version 1.4.0 back into beta.
2440 
2441 Version 1.4.0beta88 [October 30, 2009]
2442   Revised libpng*.txt section about differences between 1.2.x and 1.4.0
2443     because most of the new features have now been ported back to 1.2.41
2444 
2445 Version 1.4.0beta89 [November 1, 2009]
2446   More bugfixes and improvements to CMakeLists.txt (Philip Lowman)
2447   Removed a harmless extra png_set_invert_alpha() from pngwrite.c
2448   Apply png_user_chunk_cache_max within png_decompress_chunk().
2449   Merged libpng-1.2.41.txt with libpng-1.4.0.txt where appropriate.
2450 
2451 Version 1.4.0beta90 [November 2, 2009]
2452   Removed all remaining WIN32_WCE #ifdefs except those involving the
2453     time.h "tm" structure
2454 
2455 Version 1.4.0beta91 [November 3, 2009]
2456   Updated scripts/pngw32.def and projects/wince/png32ce.def
2457   Copied projects/wince/png32ce.def to the scripts directory.
2458   Added scripts/makefile.wce
2459   Patched ltmain.sh for wince support.
2460   Added PNG_CONVERT_tIME_SUPPORTED macro.
2461 
2462 Version 1.4.0beta92 [November 4, 2009]
2463   Make inclusion of time.h in pngconf.h depend on PNG_CONVERT_tIME_SUPPORTED
2464   Make #define PNG_CONVERT_tIME_SUPPORTED depend on PNG_WRITE_tIME_SUPPORTED
2465   Revised libpng*.txt to describe differences from 1.2.40 to 1.4.0 (instead
2466     of differences from 1.2.41 to 1.4.0)
2467 
2468 Version 1.4.0beta93 [November 7, 2009]
2469   Added PNG_DEPSTRUCT, PNG_DEPRECATED, PNG_USE_RESULT, PNG_NORETURN, and
2470     PNG_ALLOCATED macros to detect deprecated direct access to the
2471     png_struct or info_struct members and other deprecated usage in
2472     applications (John Bowler).
2473   Updated scripts/makefile* to add "-DPNG_CONFIGURE_LIBPNG" to CFLAGS,
2474     to prevent warnings about direct access to png structs by libpng
2475     functions while building libpng.  They need to be tested, especially
2476     those using compilers other than gcc.
2477   Updated projects/visualc6 and visualc71 with "/d PNG_CONFIGURE_LIBPNG".
2478     They should work but still need to be updated to remove
2479     references to pnggccrd.c or pngvcrd.c and ASM building.
2480   Added README.txt to the beos, cbuilder5, netware, and xcode projects warning
2481     that they need to be updated, to remove references to pnggccrd.c and
2482     pngvcrd.c and to depend on pngpriv.h
2483   Removed three direct references to read_info_ptr members in pngtest.c
2484     that were detected by the new PNG_DEPSTRUCT macro.
2485   Moved the png_debug macro definitions and the png_read_destroy(),
2486     png_write_destroy() and png_far_to_near() prototypes from png.h
2487     to pngpriv.h (John Bowler)
2488   Moved the synopsis lines for png_read_destroy(), png_write_destroy()
2489     png_debug(), png_debug1(), and png_debug2() from libpng.3 to libpngpf.3.
2490 
2491 Version 1.4.0beta94 [November 9, 2009]
2492   Removed the obsolete, unused pnggccrd.c and pngvcrd.c files.
2493   Updated CMakeLists.txt to add "-DPNG_CONFIGURE_LIBPNG" to the definitions.
2494   Removed dependency of pngtest.o on pngpriv.h in the makefiles.
2495   Only #define PNG_DEPSTRUCT, etc. in pngconf.h if not already defined.
2496 
2497 Version 1.4.0beta95 [November 10, 2009]
2498   Changed png_check_sig() to !png_sig_cmp() in contrib programs.
2499   Added -DPNG_CONFIGURE_LIBPNG to contrib/pngminm/*/makefile
2500   Changed png_check_sig() to !png_sig_cmp() in contrib programs.
2501   Corrected the png_get_IHDR() call in contrib/gregbook/readpng2.c
2502   Changed pngminim/*/gather.sh to stop trying to remove pnggccrd.c and pngvcrd.c
2503   Added dependency on pngpriv.h in contrib/pngminim/*/makefile
2504 
2505 Version 1.4.0beta96 [November 12, 2009]
2506   Renamed scripts/makefile.wce to scripts/makefile.cegcc
2507   Revised Makefile.am to use libpng.sys while building libpng.so
2508     so that only PNG_EXPORT functions are exported.
2509   Removed the deprecated png_check_sig() function/macro.
2510   Removed recently removed function names from scripts/*.def
2511   Revised pngtest.png to put chunks in the same order written by pngtest
2512     (evidently the same change made in libpng-1.0beta54 was lost).
2513   Added PNG_PRIVATE macro definition in pngconf.h for possible future use.
2514 
2515 Version 1.4.0beta97 [November 13, 2009]
2516   Restored pngtest.png to the libpng-1.4.0beta7 version.
2517   Removed projects/beos and netware.txt; no one seems to be supporting them.
2518   Revised Makefile.in
2519 
2520 Version 1.4.0beta98 [November 13, 2009]
2521   Added the "xcode" project to zip distributions,
2522   Fixed a typo in scripts/pngwin.def introduced in beta97.
2523 
2524 Version 1.4.0beta99 [November 14, 2009]
2525   Moved libpng-config.in and libpng.pc-configure.in out of the scripts
2526     directory, to libpng-config.in and libpng-pc.in, respectively, and
2527     modified Makefile.am and configure.ac accordingly.  Now "configure"
2528     needs nothing from the "scripts" directory.
2529   Avoid redefining PNG_CONST in pngconf.h
2530 
2531 Version 1.4.0beta100 [November 14, 2009]
2532   Removed ASM builds from projects/visualc6 and projects/visualc71
2533   Removed scripts/makefile.nommx and makefile.vcawin32
2534   Revised CMakeLists.txt to account for new location of libpng-config.in
2535     and libpng-pc.in
2536   Updated INSTALL to reflect removal and relocation of files.
2537 
2538 Version 1.4.0beta101 [November 14, 2009]
2539   Restored the binary files (*.jpg, *.png, some project files) that were
2540     accidentally deleted from the zip and 7z distributions when the xcode
2541     project was added.
2542 
2543 Version 1.4.0beta102 [November 18, 2009]
2544   Added libpng-config.in and libpng-pc.in to the zip and 7z distributions.
2545   Fixed a typo in projects/visualc6/pngtest.dsp, introduced in beta100.
2546   Moved descriptions of makefiles and other scripts out of INSTALL into
2547     scripts/README.txt
2548   Updated the copyright year in scripts/pngwin.rc from 2006 to 2009.
2549 
2550 Version 1.4.0beta103 [November 21, 2009]
2551   Removed obsolete comments about ASM from projects/visualc71/README_zlib.txt
2552   Align row_buf on 16-byte boundary in memory.
2553   Restored the PNG_WRITE_FLUSH_AFTER_IEND_SUPPORTED guard around the call
2554     to png_flush() after png_write_IEND().  See 1.4.0beta32, 1.4.0beta50
2555     changes above and 1.2.30, 1.2.30rc01 and rc03 in 1.2.41 CHANGES.  Someone
2556     needs this feature.
2557   Make the 'png_jmpbuf' macro expand to a call that records the correct
2558     longjmp function as well as returning a pointer to the setjmp
2559     jmp_buf buffer, and marked direct access to jmpbuf 'deprecated'.
2560     (John Bowler)
2561 
2562 Version 1.4.0beta104 [November 22, 2009]
2563   Removed png_longjmp_ptr from scripts/*.def and libpng.3
2564   Rebuilt configure scripts with autoconf-2.65
2565 
2566 Version 1.4.0beta105 [November 25, 2009]
2567   Use fast integer PNG_DIVIDE_BY_255() or PNG_DIVIDE_BY_65535()
2568     to accomplish alpha premultiplication when
2569     PNG_READ_COMPOSITE_NODIV_SUPPORTED is defined.
2570   Changed "/255" to "/255.0" in background calculations to make it clear
2571     that the 255 is used as a double.
2572 
2573 Version 1.4.0beta106 [November 27, 2009]
2574   Removed premultiplied alpha feature.
2575 
2576 Version 1.4.0beta107 [December 4, 2009]
2577   Updated README
2578   Added "#define PNG_NO_PEDANTIC_WARNINGS" in the libpng source files.
2579   Removed "-DPNG_CONFIGURE_LIBPNG" from the makefiles and projects.
2580   Revised scripts/makefile.netbsd, makefile.openbsd, and makefile.sco
2581     to put png.h and pngconf.h in $prefix/include, like the other scripts,
2582     instead of in $prefix/include/libpng.  Also revised makefile.sco
2583     to put them in $prefix/include/libpng15 instead of in
2584     $prefix/include/libpng/libpng15.
2585 
2586 Version 1.4.0beta108 [December 11, 2009]
2587   Removed leftover "-DPNG_CONFIGURE_LIBPNG" from contrib/pngminim/*/makefile
2588   Relocated png_do_chop() to its original position in pngrtran.c; the
2589     change in version 1.2.41beta08 caused transparency to be handled wrong
2590     in some 16-bit datastreams (Yusaku Sugai).
2591 
2592 Version 1.4.0beta109 [December 13, 2009]
2593   Added "bit_depth" parameter to the private png_build_gamma_table() function.
2594   Pass bit_depth=8 to png_build_gamma_table() when bit_depth is 16 but the
2595     PNG_16_TO_8 transform has been set, to avoid unnecessary build of 16-bit
2596     tables.
2597 
2598 Version 1.4.0rc02 [December 20, 2009]
2599   Declared png_cleanup_needed "volatile" in pngread.c and pngwrite.c
2600 
2601 Version 1.4.0rc03 [December 22, 2009]
2602   Renamed libpng-pc.in back to libpng.pc.in and revised CMakeLists.txt
2603     (revising the change in 1.4.0beta99)
2604 
2605 Version 1.4.0rc04 [December 25, 2009]
2606   Swapped PNG_UNKNOWN_CHUNKS_SUPPORTED and PNG_HANDLE_AS_UNKNOWN_SUPPORTED
2607     in pngset.c to be consistent with other changes in version 1.2.38.
2608 
2609 Version 1.4.0rc05 [December 25, 2009]
2610   Changed "libpng-pc.in" to "libpng.pc.in" in configure.ac, configure, and
2611     Makefile.in to be consistent with changes in libpng-1.4.0rc03
2612 
2613 Version 1.4.0rc06 [December 29, 2009]
2614   Reverted the gamma_table changes from libpng-1.4.0beta109.
2615   Fixed some indentation errors.
2616 
2617 Version 1.4.0rc07 [January 1, 2010]
2618   Revised libpng*.txt and libpng.3 about 1.2.x->1.4.x differences.
2619   Use png_calloc() instead of png_malloc(); png_memset() in pngrutil.c
2620   Update copyright year to 2010.
2621 
2622 Version 1.4.0rc08 [January 2, 2010]
2623   Avoid deprecated references to png_ptr-io_ptr and png_ptr->error_ptr
2624     in pngtest.c
2625 
2626 Version 1.4.0 [January 3, 2010]
2627   No changes.
2628 
2629 Version 1.4.1beta01 [January 8, 2010]
2630   Updated CMakeLists.txt for consistent indentation and to avoid an
2631     unclosed if-statement warning (Philip Lowman).
2632   Revised Makefile.am and Makefile.in to remove references to Y2KINFO,
2633     KNOWNBUG, and libpng.la (Robert Schwebel).
2634   Revised the makefiles to install the same files and symbolic
2635     links as configure, except for libpng.la and libpng14.la.
2636   Make png_set|get_compression_buffer_size() available even when
2637     PNG_WRITE_SUPPORTED is not enabled.
2638   Revised Makefile.am and Makefile.in to simplify their maintenance.
2639   Revised scripts/makefile.linux to install a link to libpng14.so.14.1
2640 
2641 Version 1.4.1beta02 [January 9, 2010]
2642   Revised the rest of the makefiles to install a link to libpng14.so.14.1
2643 
2644 Version 1.4.1beta03 [January 10, 2010]
2645   Removed png_set_premultiply_alpha() from scripts/*.def
2646 
2647 Version 1.4.1rc01 [January 16, 2010]
2648   No changes.
2649 
2650 Version 1.4.1beta04 [January 23, 2010]
2651   Revised png_decompress_chunk() to improve speed and memory usage when
2652     decoding large chunks.
2653   Added png_set|get_chunk_malloc_max() functions.
2654 
2655 Version 1.4.1beta05 [January 26, 2010]
2656   Relocated "int k" declaration in pngtest.c to minimize its scope.
2657 
2658 Version 1.4.1beta06 [January 28, 2010]
2659   Revised png_decompress_chunk() to use a two-pass method suggested by
2660     John Bowler.
2661 
2662 Version 1.4.1beta07 [February 6, 2010]
2663   Folded some long lines in the source files.
2664   Added defineable PNG_USER_CHUNK_CACHE_MAX, PNG_USER_CHUNK_MALLOC_MAX,
2665     and a PNG_USER_LIMITS_SUPPORTED flag.
2666   Eliminated use of png_ptr->irowbytes and reused the slot in png_ptr as
2667     png_ptr->png_user_chunk_malloc_max.
2668   Revised png_push_save_buffer() to do fewer but larger png_malloc() calls.
2669 
2670 Version 1.4.1beta08 [February 6, 2010]
2671   Minor cleanup and updating of dates and copyright year.
2672 
2673 Version 1.5.0beta01 [February 7, 2010]
2674   Moved declaration of png_struct into private pngstruct.h and png_info
2675     into pnginfo.h
2676 
2677 Version 1.4.1beta09 and 1.5.0beta02 [February 7, 2010]
2678   Reverted to original png_push_save_buffer() code.
2679 
2680 Version 1.4.1beta10 and 1.5.0beta03 [February 8, 2010]
2681   Return allocated "old_buffer" in png_push_save_buffer() before
2682     calling png_error(), to avoid a potential memory leak.
2683   Updated configure script to use SO number 15.
2684 
2685 Version 1.5.0beta04 [February 9, 2010]
2686   Removed malformed "incomplete struct declaration" of png_info from png.h
2687 
2688 Version 1.5.0beta05 [February 12, 2010]
2689   Removed PNG_DEPSTRUCT markup in pngstruct.h and pnginfo.h, and undid the
2690     linewrapping that it entailed.
2691   Revised comments in pngstruct.h and pnginfo.h and added pointers to
2692     the libpng license.
2693   Changed PNG_INTERNAL to PNG_EXPOSE_INTERNAL_STRUCTURES
2694   Removed the cbuilder5 project, which has not been updated to 1.4.0.
2695 
2696 Version 1.4.1beta12 and 1.5.0beta06 [February 14, 2010]
2697   Fixed type declaration of png_get_chunk_malloc_max() in pngget.c (Daisuke
2698     Nishikawa)
2699 
2700 Version 1.5.0beta07 [omitted]
2701 
2702 Version 1.5.0beta08 [February 19, 2010]
2703   Changed #ifdef PNG_NO_STDIO_SUPPORTED to #ifdef PNG_NO_CONSOLE_IO_SUPPORTED
2704     wherever png_snprintf() is used to construct error and warning messages.
2705   Noted in scripts/makefile.mingw that it expects to be run under MSYS.
2706   Removed obsolete unused MMX-querying support from contrib/gregbook
2707   Added exported png_longjmp() function.
2708   Removed the AIX redefinition of jmpbuf in png.h
2709   Added -D_ALLSOURCE in configure.ac, makefile.aix, and CMakeLists.txt
2710     when building on AIX.
2711 
2712 Version 1.5.0beta09 [February 19, 2010]
2713   Removed -D_ALLSOURCE from configure.ac, makefile.aix, and CMakeLists.txt.
2714   Changed the name of png_ptr->jmpbuf to png_ptr->png_jmpbuf in pngstruct.h
2715 
2716 Version 1.5.0beta10 [February 25, 2010]
2717   Removed unused gzio.c from contrib/pngminim gather and makefile scripts
2718   Removed replacement error handlers from contrib/gregbook.  Because of
2719     the new png_longjmp() function they are no longer needed.
2720 
2721 Version 1.5.0beta11 [March 6, 2010]
2722   Removed checking for already-included setjmp.h from pngconf.h
2723   Fixed inconsistent indentations and made numerous cosmetic changes.
2724   Revised the "SEE ALSO" style of libpng.3, libpngpf.3, and png.5
2725 
2726 Version 1.5.0beta12 [March 9, 2010]
2727   Moved "#include png.h" inside pngpriv.h and removed "#include png.h" from
2728     the source files, along with "#define PNG_EXPOSE_INTERNAL_STRUCTURES"
2729     and "#define PNG_NO_PEDANTIC_WARNINGS" (John Bowler).
2730   Created new pngdebug.h and moved debug definitions there.
2731 
2732 Version 1.5.0beta13 [March 10, 2010]
2733   Protect pngstruct.h, pnginfo.h, and pngdebug.h from being included twice.
2734   Revise the "#ifdef" blocks in png_inflate() so it will compile when neither
2735     PNG_USER_CHUNK_MALLOC_MAX nor PNG_SET_CHUNK_MALLOC_LIMIT_SUPPORTED
2736     is defined.
2737   Removed unused png_measure_compressed_chunk() from pngpriv.h and libpngpf.3
2738   Moved the 'config.h' support from pngconf.h to pngpriv.h
2739   Removed PNGAPI from the png_longjmp_ptr typedef.
2740   Eliminated dependence of pngtest.c on the private pngdebug.h file.
2741   Make all png_debug macros into *unterminated* statements or
2742     expressions (i.e. a trailing ';' must always be added) and correct
2743     the format statements in various png_debug messages.
2744 
2745 Version 1.5.0beta14 [March 14, 2010]
2746   Removed direct access to png_ptr->io_ptr from the Windows code in pngtest.c
2747   Revised Makefile.am to account for recent additions and replacements.
2748   Corrected CE and OS/2 DEF files (scripts/png*def) for symbols removed and
2749     added ordinal numbers to the Windows DEF file and corrected the duplicated
2750     ordinal numbers on CE symbols that are commented out.
2751   Added back in export symbols that can be present in the Windows build but
2752     are disabled by default.
2753   PNG_EXPORT changed to include an 'ordinal' field for DEF file generation.
2754     PNG_CALLBACK added to make callback definitions uniform.  PNGAPI split
2755     into PNGCAPI (base C form), PNGAPI (exports) and PNGCBAPI (callbacks),
2756     and appropriate changes made to all files.  Cygwin builds re-hinged to
2757     allow procedure call standard changes and to remove the need for the DEF
2758     file (fixes build on Cygwin).
2759   Enabled 'attribute' warnings that are relevant to library APIs and callbacks.
2760   Changed rules for generation of the various symbol files and added a new
2761     rule for a DEF file (which is also added to the distribution).
2762   Updated the symbol file generation to stop it adding spurious spaces
2763     to EOL (coming from preprocessor macro expansion).  Added a facility
2764     to join tokens in the output and rewrite *.dfn to use this.
2765   Eliminated scripts/*.def in favor of libpng.def; updated projects/visualc71
2766     and removed scripts/makefile.cygwin.
2767   Made PNG_BUILD_DLL safe: it can be set whenever a DLL is being built.
2768   Removed the include of sys/types.h - apparently unnecessary now on the
2769     platforms on which it happened (all but Mac OS and RISC OS).
2770   Moved the Mac OS test into pngpriv.h (the only place it is used.)
2771 
2772 Version 1.5.0beta15 [March 17, 2010]
2773   Added symbols.chk target to Makefile.am to validate the symbols in png.h
2774     against the new DEF file scripts/symbols.def.
2775   Changed the default DEF file back to pngwin.def.
2776   Removed makefile.mingw.
2777   Eliminated PNG_NO_EXTERN and PNG_ALL_EXTERN
2778 
2779 Version 1.5.0beta16 [April 1, 2010]
2780   Make png_text_struct independent of PNG_iTXt_SUPPORTED, so that
2781     fields are initialized in all configurations.  The READ/WRITE
2782     macros (PNG_(READ|WRITE)_iTXt_SUPPORTED) still function as
2783     before to disable code to actually read or write iTXt chunks
2784     and iTXt_SUPPORTED can be used to detect presence of either
2785     read or write support (but it is probably better to check for
2786     the one actually required - read or write.)
2787   Combined multiple png_warning() calls for a single error.
2788   Restored the macro definition of png_check_sig().
2789 
2790 Version 1.5.0beta17 [April 17, 2010]
2791   Added some "(long)" typecasts to printf calls in png_handle_cHRM().
2792   Documented the fact that png_set_dither() was disabled since libpng-1.4.0.
2793   Reenabled png_set_dither() but renamed it to png_set_quantize() to reflect
2794     more accurately what it actually does.  At the same time, renamed
2795     the PNG_DITHER_[RED,GREEN_BLUE]_BITS macros to
2796     PNG_QUANTIZE_[RED,GREEN,BLUE]_BITS.
2797   Added some "(long)" typecasts to printf calls in png_handle_cHRM().
2798   Freeze build-time only configuration in the build.
2799     In all prior versions of libpng most configuration options
2800     controlled by compiler #defines had to be repeated by the
2801     application code that used libpng.  This patch changes this
2802     so that compilation options that can only be changed at build
2803     time are frozen in the build.  Options that are compiler
2804     dependent (and those that are system dependent) are evaluated
2805     each time - pngconf.h holds these.  Options that can be changed
2806     per-file in the application are in png.h.  Frozen options are
2807     in the new installed header file pnglibconf.h (John Bowler)
2808   Removed the xcode project because it has not been updated to work
2809     with libpng-1.5.0.
2810   Removed the ability to include optional pngusr.h
2811 
2812 Version 1.5.0beta18 [April 17, 2010]
2813   Restored the ability to include optional pngusr.h
2814   Moved replacements for png_error() and png_warning() from the
2815     contrib/pngminim project to pngerror.c, for use when warnings or
2816     errors are disabled via PNG_NO_WARN or PNG_NO_ERROR_TEXT, to avoid
2817     storing unneeded error/warning text.
2818   Updated contrib/pngminim project to work with the new pnglibconf.h
2819   Added some PNG_NO_* defines to contrib/pngminim/*/pngusr.h to save space.
2820 
2821 Version 1.5.0beta19 [April 24, 2010]
2822   Added PNG_{READ,WRITE}_INT_FUNCTIONS_SUPPORTED.  This allows the functions
2823     to read and write ints to be disabled independently of PNG_USE_READ_MACROS,
2824     which allows libpng to be built with the functions even though the default
2825     is to use the macros - this allows applications to choose at app build
2826     time whether or not to use macros (previously impossible because the
2827     functions weren't in the default build.)
2828   Changed Windows calling convention back to __cdecl for API functions.
2829     For Windows/x86 platforms only:
2830       __stdcall is no longer needed for Visual Basic, so libpng-1.5.0 uses
2831       __cdecl throughout (both API functions and callbacks) on Windows/x86
2832       platforms.
2833   Replaced visualc6 and visualc71 projects with new vstudio project
2834   Relaxed the overly-restrictive permissions of some files.
2835 
2836 Version 1.5.0beta20 [April 24, 2010]
2837   Relaxed more overly-restrictive permissions of some files.
2838 
2839 Version 1.5.0beta21 [April 27, 2010]
2840   Removed some unwanted binary bytes and changed CRLF to NEWLINE in the new
2841     vstudio project files, and some trivial editing of some files in the
2842     scripts directory.
2843   Set PNG_NO_READ_BGR, PNG_NO_IO_STATE, and PNG_NO_TIME_RFC1123 in
2844     contrib/pngminim/decoder/pngusr.h to make a smaller decoder application.
2845 
2846 Version 1.5.0beta22 [April 28, 2010]
2847   Fixed dependencies of GET_INT_32 - it does not require READ_INT_FUNCTIONS
2848     because it has a macro equivalent.
2849   Improved the options.awk script; added an "everything off" option.
2850   Revised contrib/pngminim to use the "everything off" option in pngusr.dfa.
2851 
2852 Version 1.5.0beta23 [April 29, 2010]
2853   Corrected PNG_REMOVED macro to take five arguments.
2854     The macro was documented with two arguments (name,ordinal), however
2855     the symbol checking .dfn files assumed five arguments.  The five
2856     argument form seems more useful so it is changed to that.
2857   Corrected PNG_UNKNOWN_CHUNKS_SUPPORTED to PNG_HANDLE_AS_UNKNOWN_SUPPORTED
2858     in gregbook/readpng2.c
2859   Corrected protection of png_get_user_transform_ptr. The API declaration in
2860     png.h is removed if both READ and WRITE USER_TRANSFORM are turned off
2861     but was left defined in pngtrans.c
2862   Added logunsupported=1 to cause pnglibconf.h to document disabled options.
2863     This makes the installed pnglibconf.h more readable but causes no
2864     other change.  The intention is that users of libpng will find it
2865     easier to understand if an API they need is missing.
2866   Include png_reset_zstream() in png.c only when PNG_READ_SUPPORTED is defined.
2867   Removed dummy_inflate.c from contrib/pngminim/encoder
2868   Removed contrib/pngminim/*/gather.sh; gathering is now done in the makefile.
2869 
2870 Version 1.5.0beta24 [May 7, 2010]
2871   Use bitwise "&" instead of arithmetic mod in pngrutil.c calculation of the
2872     offset of the png_ptr->rowbuf pointer into png_ptr->big_row_buf.
2873   Added more blank lines for readability.
2874 
2875 Version 1.5.0beta25 [June 18, 2010]
2876   In pngpread.c: png_push_have_row() add check for new_row > height
2877   Removed the now-redundant check for out-of-bounds new_row from example.c
2878 
2879 Version 1.5.0beta26 [June 18, 2010]
2880   In pngpread.c: png_push_process_row() add check for too many rows.
2881 
2882 Version 1.5.0beta27 [June 18, 2010]
2883   Removed the check added in beta25 as it is now redundant.
2884 
2885 Version 1.5.0beta28 [June 20, 2010]
2886   Rewrote png_process_IDAT_data to consistently treat extra data as warnings
2887     and handle end conditions more cleanly.
2888   Removed the new (beta26) check in png_push_process_row().
2889 
2890 Version 1.5.0beta29 [June 21, 2010]
2891   Revised scripts/options.awk to work on Sunos (but still doesn't work)
2892   Added comment to options.awk and contrib/pngminim/*/makefile to try nawk.
2893 
2894 Version 1.5.0beta30 [June 22, 2010]
2895   Stop memory leak when reading a malformed sCAL chunk.
2896 
2897 Version 1.5.0beta31 [June 26, 2010]
2898   Revised pngpread.c patch of beta28 to avoid an endless loop.
2899   Removed some trailing blanks.
2900 
2901 Version 1.5.0beta32 [June 26, 2010]
2902   Removed leftover scripts/options.patch and scripts/options.rej
2903 
2904 Version 1.5.0beta33 [July 6, 3010]
2905   Made FIXED and FLOATING options consistent in the APIs they enable and
2906     disable.  Corrected scripts/options.awk to handle both command line
2907     options and options specified in the .dfa files.
2908   Changed char *msg to PNG_CONST char *msg in pngrutil.c
2909   Make png_set_sRGB_gAMA_and_cHRM set values using either the fixed or
2910     floating point APIs, but not both.
2911   Reversed patch to remove error handler when the jmp_buf is stored in the
2912     main program structure, not the png_struct.
2913     The error handler is needed because the default handler in libpng will
2914     always use the jmp_buf in the library control structure; this is never
2915     set.  The gregbook code is a useful example because, even though it
2916     uses setjmp/longjmp, it shows how error handling can be implemented
2917     using control mechanisms not directly supported by libpng.  The
2918     technique will work correctly with mechanisms such as Microsoft
2919     Structure Exceptions or C++ exceptions (compiler willing - note that gcc
2920     does not by default support interworking of C and C++ error handling.)
2921   Reverted changes to call png_longjmp in contrib/gregbook where it is not
2922     appropriate.  If mainprog->jmpbuf is used by setjmp, then png_longjmp
2923     cannot be used.
2924   Changed "extern PNG_EXPORT" to "PNG_EXPORT" in png.h (Jan Nijtmans)
2925   Changed "extern" to "PNG_EXTERN" in pngpriv.h (except for the 'extern "C" {')
2926 
2927 Version 1.5.0beta34 [July 12, 2010]
2928   Put #ifndef PNG_EXTERN, #endif around the define PNG_EXTERN in pngpriv.h
2929 
2930 Version 1.5.0beta35 [July 24, 2010]
2931   Removed some newly-added TAB characters.
2932   Added -DNO_PNG_SNPRINTF to CFLAGS in scripts/makefile.dj2
2933   Moved the definition of png_snprintf() outside of the enclosing
2934     #ifdef blocks in pngconf.h
2935 
2936 Version 1.5.0beta36 [July 29, 2010]
2937   Patches by John Bowler:
2938   Fixed point APIs are now supported throughout (no missing APIs).
2939   Internal fixed point arithmetic support exists for all internal floating
2940     point operations.
2941   sCAL validates the floating point strings it is passed.
2942   Safe, albeit rudimentary, Watcom support is provided by PNG_API_RULE==2
2943   Two new APIs exist to get the number of passes without turning on the
2944     PNG_INTERLACE transform and to get the number of rows in the current
2945     pass.
2946   A new test program, pngvalid.c, validates the gamma code.
2947   Errors in the 16-bit gamma correction (overflows) have been corrected.
2948   cHRM chunk testing is done consistently (previously the floating point
2949     API bypassed it, because the test really didn't work on FP, now the test
2950     is performed on the actual values to be stored in the PNG file so it
2951     works in the FP case too.)
2952   Most floating point APIs now simply call the fixed point APIs after
2953     converting the values to the fixed point form used in the PNG file.
2954   The standard headers no longer include zlib.h, which is currently only
2955     required for pngstruct.h and can therefore be internal.
2956   Revised png_get_int_32 to undo the PNG two's complement representation of
2957     negative numbers.
2958 
2959 Version 1.5.0beta37 [July 30, 2010]
2960   Added a typecast in png_get_int_32() in png.h and pngrutil.h to avoid
2961     a compiler warning.
2962   Replaced oFFs 0,0 with oFFs -10,20 in pngtest.png
2963 
2964 Version 1.5.0beta38 [July 31, 2010]
2965   Implemented remaining "_fixed" functions.
2966   Corrected a number of recently introduced warnings mostly resulting from
2967     safe but uncast assignments to shorter integers.  Also added a zlib
2968     VStudio release library project because the latest zlib Official Windows
2969     build does not include such a thing.
2970   Revised png_get_int_16() to be similar to png_get_int_32().
2971   Restored projects/visualc71.
2972 
2973 Version 1.5.0beta39 [August 2, 2010]
2974   VisualC/GCC warning fixes, VisualC build fixes
2975   The changes include support for function attributes in VC in addition to
2976     those already present in GCC - necessary because without these some
2977     warnings are unavoidable.  Fixes include signed/unsigned fixes in
2978     pngvalid and checks with gcc -Wall -Wextra -Wunused.
2979   VC requires function attributes on function definitions as well as
2980     declarations, PNG_FUNCTION has been added to enable this and the
2981     relevant function definitions changed.
2982 
2983 Version 1.5.0beta40 [August 6, 2010]
2984   Correct use of _WINDOWS_ in pngconf.h
2985   Removed png_mem_ #defines; they are no longer used.
2986   Added the sRGB chunk to pngtest.png
2987 
2988 Version 1.5.0beta41 [August 11, 2010]
2989   Added the cHRM chunk to pngtest.png
2990   Don't try to use version-script with cygwin/mingw.
2991   Revised contrib/gregbook to work under cygwin/mingw.
2992 
2993 Version 1.5.0beta42 [August 18, 2010]
2994   Add .dll.a to the list of extensions to be symlinked by Makefile.am (Yaakov)
2995   Made all API functions that have const arguments and constant string
2996     literal pointers declare them (John Bowler).
2997 
2998 Version 1.5.0beta43 [August 20, 2010]
2999   Removed spurious tabs, shorten long lines (no source change)
3000     Also added scripts/chkfmt to validate the format of all the files that can
3001     reasonably be validated (it is suggested to run "make distclean" before
3002     checking, because some machine generated files have long lines.)
3003   Reformatted the CHANGES file to be more consistent throughout.
3004   Made changes to address various issues identified by GCC, mostly
3005     signed/unsigned and shortening problems on assignment but also a few
3006     difficult to optimize (for GCC) loops.
3007   Fixed non-GCC fixed point builds.  In png.c a declaration was misplaced
3008     in an earlier update.  Fixed to declare the auto variables at the head.
3009   Use cexcept.h in pngvalid.c.
3010 
3011 Version 1.5.0beta44 [August 24, 2010]
3012   Updated CMakeLists.txt to use CMAKE_INSTALL_LIBDIR variable; useful for
3013     installing libpng in /usr/lib64 (Funda Wang).
3014   Revised CMakeLists.txt to put the man pages in share/man/man* not man/man*
3015   Revised CMakeLists.txt to make symlinks instead of copies when installing.
3016   Changed PNG_LIB_NAME from pngNN to libpngNN in CMakeLists.txt (Philip Lowman)
3017   Implemented memory checks within pngvalid
3018   Reformatted/rearranged pngvalid.c to assist use of progressive reader.
3019   Check interlaced images in pngvalid
3020   Clarified pngusr.h comments in pnglibconf.dfa
3021   Simplified the pngvalid error-handling code now that cexcept.h is in place.
3022   Implemented progressive reader in pngvalid.c for standard tests
3023   Implemented progressive read in pngvalid.c gamma tests
3024   Turn on progressive reader in pngvalid.c by default and tidy code.
3025 
3026 Version 1.5.0beta45 [August 26, 2010]
3027   Added an explicit make step to projects/vstudio for pnglibconf.h
3028     Also corrected zlib.vcxproj into which Visual Studio had introduced
3029     what it calls an "authoring error".  The change to make pnglibconf.h
3030     simply copies the file; in the future it may actually generate the
3031     file from scripts/pnglibconf.dfa as the other build systems do.
3032   Changed pngvalid to work when floating point APIs are disabled
3033   Renamed the prebuilt scripts/pnglibconf.h to scripts/pnglibconf.h.prebuilt
3034   Supply default values for PNG_USER_PRIVATEBUILD and PNG_USER_DLLFNAME_POSTFIX
3035     in pngpriv.h in case the user neglected to define them in their pngusr.h
3036 
3037 Version 1.5.0beta46 [August 28, 2010]
3038   Added new private header files to libpng_sources in CMakeLists.txt
3039   Added PNG_READ_16BIT, PNG_WRITE_16BIT, and PNG_16BIT options.
3040   Added reference to scripts/pnglibconf.h.prebuilt in the visualc71 project.
3041 
3042 Version 1.5.0beta47 [September 11, 2010]
3043   Fixed a number of problems with 64-bit compilation reported by Visual
3044     Studio 2010 (John Bowler).
3045 
3046 Version 1.5.0beta48 [October 4, 2010]
3047   Updated CMakeLists.txt (Philip Lowman).
3048   Revised autogen.sh to recognize and use $AUTOCONF, $AUTOMAKE, $AUTOHEADER,
3049     $AUTOPOINT, $ACLOCAL and $LIBTOOLIZE
3050   Fixed problem with symbols creation in Makefile.am which was assuming that
3051     all versions of ccp write to standard output by default (Martin Banky). The
3052     bug was introduced in libpng-1.2.9beta5.
3053   Removed unused mkinstalldirs.
3054 
3055 Version 1.5.0beta49 [October 8, 2010]
3056   Undid Makefile.am revision of 1.5.0beta48.
3057 
3058 Version 1.5.0beta50 [October 14, 2010]
3059   Revised Makefile.in to account for mkinstalldirs being removed.
3060   Added some "(unsigned long)" typecasts in printf statements in pngvalid.c.
3061   Suppressed a compiler warning in png_handle_sPLT().
3062   Check for out-of-range text compression mode in png_set_text().
3063 
3064 Version 1.5.0beta51 [October 15, 2010]
3065   Changed embedded dates to "(PENDING RELEASE) in beta releases (and future
3066     rc releases) to minimize the difference between releases.
3067 
3068 Version 1.5.0beta52 [October 16, 2010]
3069   Restored some of the embedded dates (in png.h, png.c, documentation, etc.)
3070 
3071 Version 1.5.0beta53 [October 18, 2010]
3072   Updated INSTALL to mention using "make maintainer-clean" and to remove
3073     obsolete statement about a custom ltmain.sh
3074   Disabled "color-tests" by default in Makefile.am so it will work with
3075     automake versions earlier than 1.11.1
3076   Use document name "libpng-manual.txt" instead of "libpng-<version>.txt"
3077     to simplify version differences.
3078   Removed obsolete remarks about setjmp handling from INSTALL.
3079   Revised and renamed the typedef in png.h and png.c that was designed
3080     to catch library and header mismatch.
3081 
3082 Version 1.5.0beta54 [November 10, 2010]
3083   Require 48 bytes, not 64 bytes, for big_row_buf in overflow checks.
3084   Used a consistent structure for the pngget.c functions.
3085 
3086 Version 1.5.0beta55 [November 21, 2010]
3087   Revised png_get_uint_32, png_get_int_32, png_get_uint_16 (Cosmin)
3088   Moved reading of file signature into png_read_sig (Cosmin)
3089   Fixed atomicity of chunk header serialization (Cosmin)
3090   Added test for io_state in pngtest.c (Cosmin)
3091   Added "#!/bin/sh" at the top of contrib/pngminim/*/gather.sh scripts.
3092   Changes to remove gcc warnings (John Bowler)
3093     Certain optional gcc warning flags resulted in warnings in libpng code.
3094     With these changes only -Wconversion and -Wcast-qual cannot be turned on.
3095     Changes are trivial rearrangements of code.  -Wconversion is not possible
3096     for pngrutil.c (because of the widespread use of += et al on variables
3097     smaller than (int) or (unsigned int)) and -Wcast-qual is not possible
3098     with pngwio.c and pngwutil.c because the 'write' callback and zlib
3099     compression both fail to declare their input buffers with 'const'.
3100 
3101 Version 1.5.0beta56 [December 7, 2010]
3102   Added the private PNG_UNUSED() macro definition in pngpriv.h.
3103   Added some commentary about PNG_EXPORT in png.h and pngconf.h
3104   Revised PNG_EXPORT() macro and added PNG_EXPORTA() macro, with the
3105     objective of simplifying and improving the cosmetic appearance of png.h.
3106   Fixed some incorrect "=" macro names in pnglibconf.dfa
3107   Included documentation of changes in 1.5.0 from 1.4.x in libpng-manual.txt
3108 
3109 Version 1.5.0beta57 [December 9, 2010]
3110   Documented the pngvalid gamma error summary with additional comments and
3111     print statements.
3112   Improved missing symbol handling in checksym.awk; symbols missing in both
3113     the old and new files can now be optionally ignored, treated as errors
3114     or warnings.
3115   Removed references to pngvcrd.c and pnggccrd.c from the vstudio project.
3116   Updated "libpng14" to "libpng15" in the visualc71 project.
3117   Enabled the strip16 tests in pngvalid.`
3118   Don't display test results (except PASS/FAIL) when running "make test".
3119     Instead put them in pngtest-log.txt
3120   Added "--with-zprefix=<string>" to configure.ac
3121   Updated the prebuilt configuration files to autoconf version 2.68
3122 
3123 Version 1.5.0beta58 [December 19, 2010]
3124   Fixed interlace image handling and add test cases (John Bowler)
3125   Fixed the clean rule in Makefile.am to remove pngtest-log.txt
3126   Made minor changes to work around warnings in gcc 3.4
3127 
3128 Version 1.5.0rc01 [December 27, 2010]
3129   No changes.
3130 
3131 Version 1.5.0rc02 [December 27, 2010]
3132   Eliminated references to the scripts/*.def files in project/visualc71.
3133 
3134 Version 1.5.0rc03 [December 28, 2010]
3135   Eliminated scripts/*.def and revised Makefile.am accordingly
3136 
3137 Version 1.5.0rc04 [December 29, 2010]
3138   Fixed bug in background transformation handling in pngrtran.c (it was
3139     looking for the flag in png_ptr->transformations instead of in
3140     png_ptr->flags) (David Raymond).
3141 
3142 Version 1.5.0rc05 [December 31, 2010]
3143   Fixed typo in a comment in CMakeLists.txt (libpng14 => libpng15) (Cosmin)
3144 
3145 Version 1.5.0rc06 [January 4, 2011]
3146   Changed the new configure option "zprefix=string" to "zlib-prefix=string"
3147 
3148 Version 1.5.0rc07 [January 4, 2011]
3149   Updated copyright year.
3150 
3151 Version 1.5.0 [January 6, 2011]
3152   No changes.
3153 
3154 version 1.5.1beta01 [January 8, 2011]
3155   Added description of png_set_crc_action() to the manual.
3156   Added a note in the manual that the type of the iCCP profile was changed
3157     from png_charpp to png_bytepp in png_get_iCCP().  This change happened
3158     in version 1.5.0beta36 but is not noted in the CHANGES.  Similarly,
3159     it was changed from png_charpp to png_const_bytepp in png_set_iCCP().
3160   Ensure that png_rgb_to_gray ignores palette mapped images, if libpng
3161     internally happens to call it with one, and fixed a failure to handle
3162     palette mapped images correctly.  This fixes CVE-2690.
3163 
3164 Version 1.5.1beta02 [January 14, 2011]
3165   Fixed a bug in handling of interlaced images (bero at arklinux.org).
3166   Updated CMakeLists.txt (Clifford Yapp)
3167 
3168 Version 1.5.1beta03 [January 14, 2011]
3169   Fixed typecasting of some png_debug() statements (Cosmin)
3170 
3171 Version 1.5.1beta04 [January 16, 2011]
3172   Updated documentation of png_set|get_tRNS() (Thomas Klausner).
3173   Mentioned in the documentation that applications must #include "zlib.h"
3174     if they need access to anything in zlib.h, and that a number of
3175     macros such as png_memset() are no longer accessible by applications.
3176   Corrected pngvalid gamma test "sample" function to access all of the color
3177     samples of each pixel, instead of sampling the red channel three times.
3178   Prefixed variable names index, div, exp, gamma with "png_" to avoid "shadow"
3179     warnings, and (mistakenly) changed png_exp() to exp().
3180 
3181 Version 1.5.1beta05 [January 16, 2011]
3182   Changed variable names png_index, png_div, png_exp, and png_gamma to
3183     char_index, divisor, exp_b10, and gamma_val, respectively, and
3184     changed exp() back to png_exp().
3185 
3186 Version 1.5.1beta06 [January 20, 2011]
3187   Prevent png_push_crc_skip() from hanging while reading an unknown chunk
3188     or an over-large compressed zTXt chunk with the progressive reader.
3189   Eliminated more GCC "shadow" warnings.
3190   Revised png_fixed() in png.c to avoid compiler warning about reaching the
3191     end without returning anything.
3192 
3193 Version 1.5.1beta07 [January 22, 2011]
3194   In the manual, describe the png_get_IHDR() arguments in the correct order.
3195   Added const_png_structp and const_png_infop types, and used them in
3196     prototypes for most png_get_*() functions.
3197 
3198 Version 1.5.1beta08 [January 23, 2011]
3199   Added png_get_io_chunk_type() and deprecated png_get_io_chunk_name()
3200   Added synopses for the IO_STATE functions and other missing synopses
3201     to the manual. Removed the synopses from libpngpf.3 because they
3202     were out of date and no longer useful.  Better information can be
3203     obtained by reading the prototypes and comments in pngpriv.h
3204   Attempted to fix cpp on Solaris with S. Studio 12 cc, fix build
3205     Added a make macro DFNCPP that is a CPP that will accept the tokens in
3206     a .dfn file and adds configure stuff to test for such a CPP.  ./configure
3207     should fail if one is not available.
3208   Corrected const_png_ in png.h to png_const_ to avoid polluting the namespace.
3209   Added png_get_current_row_number and png_get_current_pass_number for the
3210     benefit of the user transform callback.
3211   Added png_process_data_pause and png_process_data_skip for the benefit of
3212     progressive readers that need to stop data processing or want to optimize
3213     skipping of unread data (e.g., if the reader marks a chunk to be skipped.)
3214 
3215 Version 1.5.1beta09 [January 24, 2011]
3216   Enhanced pngvalid, corrected an error in gray_to_rgb, corrected doc error.
3217     pngvalid contains tests of transforms, which tests are currently disabled
3218     because they are incompletely tested.  gray_to_rgb was failing to expand
3219     the bit depth for smaller bit depth images; this seems to be a long
3220     standing error and resulted, apparently, in invalid output
3221     (CVE-2011-0408, CERT VU#643140).  The documentation did not accurately
3222     describe what libpng really does when converting RGB to gray.
3223 
3224 Version 1.5.1beta10 [January 27, 2010]
3225   Fixed incorrect examples of callback prototypes in the manual, that were
3226     introduced in libpng-1.0.0.
3227   In addition the order of the png_get_uint macros with respect to the
3228     relevant function definitions has been reversed.  This helps the
3229     preprocessing of the symbol files be more robust.  Furthermore, the
3230     symbol file preprocessing now uses -DPNG_NO_USE_READ_MACROS even when
3231     the library may actually be built with PNG_USE_READ_MACROS; this stops
3232     the read macros interfering with the symbol file format.
3233   Made the manual, synopses, and function prototypes use the function
3234     argument names file_gamma, int_file_gamma, and srgb_intent consistently.
3235 
3236 Version 1.5.1beta11 [January 28, 2011]
3237   Changed PNG_UNUSED from "param=param;" to "{if(param){}}".
3238   Corrected local variable type in new API png_process_data_skip()
3239     The type was self-evidently incorrect but only causes problems on 64-bit
3240     architectures.
3241   Added transform tests to pngvalid and simplified the arguments.
3242 
3243 Version 1.5.1rc01 [January 29, 2011]
3244   No changes.
3245 
3246 Version 1.5.1rc02 [January 31, 2011]
3247   Added a request in the manual that applications do not use "png_" or
3248     "PNG_" to begin any of their own symbols.
3249   Changed PNG_UNUSED to "(void)param;" and updated the commentary in pngpriv.h
3250 
3251 Version 1.5.1 [February 3, 2011]
3252   No changes.
3253 
3254 Version 1.5.2beta01 [February 13, 2011]
3255   More -Wshadow fixes for older gcc compilers.  Older gcc versions apparently
3256     check formal parameters names in function declarations (as well as
3257     definitions) to see if they match a name in the global namespace.
3258   Revised PNG_EXPORTA macro to not use an empty parameter, to accommodate the
3259     old VisualC++ preprocessor.
3260   Turned on interlace handling in png_read_png().
3261   Fixed gcc pedantic warnings.
3262   Handle longjmp in Cygwin.
3263   Fixed png_get_current_row_number() in the interlaced case.
3264   Cleaned up ALPHA flags and transformations.
3265   Implemented expansion to 16 bits.
3266 
3267 Version 1.5.2beta02 [February 19, 2011]
3268   Fixed mistake in the descriptions of user read_transform and write_transform
3269     function prototypes in the manual.  The row_info struct is png_row_infop.
3270   Reverted png_get_current_row_number() to previous (1.5.2beta01) behavior.
3271   Corrected png_get_current_row_number documentation
3272   Fixed the read/write row callback documentation.
3273     This documents the current behavior, where the callback is called after
3274     every row with information pertaining to the next row.
3275 
3276 Version 1.5.2beta03 [March 3, 2011]
3277   Fixed scripts/makefile.vcwin32
3278   Updated contrib/pngsuite/README to add the word "modify".
3279   Define PNG_ALLOCATED to blank when _MSC_VER<1300.
3280 
3281 Version 1.5.2rc01 [March 19, 2011]
3282   Define remaining attributes to blank when MSC_VER<1300.
3283   ifdef out mask arrays in pngread.c when interlacing is not supported.
3284 
3285 Version 1.5.2rc02 [March 22, 2011]
3286   Added a hint to try CPP=/bin/cpp if "cpp -E" fails in scripts/pnglibconf.mak
3287     and in contrib/pngminim/*/makefile, eg., on SunOS 5.10, and removed "strip"
3288     from the makefiles.
3289   Fixed a bug (present since libpng-1.0.7) that makes png_handle_sPLT() fail
3290     to compile when PNG_NO_POINTER_INDEXING is defined (Chubanov Kirill)
3291 
3292 Version 1.5.2rc03 [March 24, 2011]
3293   Don't include standard header files in png.h while building the symbol table,
3294     to avoid cpp failure on SunOS (introduced PNG_BUILDING_SYMBOL_TABLE macro).
3295 
3296 Version 1.5.2 [March 31, 2011]
3297   No changes.
3298 
3299 Version 1.5.3beta01 [April 1, 2011]
3300   Re-initialize the zlib compressor before compressing non-IDAT chunks.
3301   Added API functions (png_set_text_compression_level() and four others) to
3302     set parameters for zlib compression of non-IDAT chunks.
3303 
3304 Version 1.5.3beta02 [April 3, 2011]
3305   Updated scripts/symbols.def with new API functions.
3306   Only compile the new zlib re-initializing code when text or iCCP is
3307     supported, using PNG_WRITE_COMPRESSED_TEXT_SUPPORTED macro.
3308   Improved the optimization of the zlib CMF byte (see libpng-1.2.6beta03).
3309   Optimize the zlib CMF byte in non-IDAT compressed chunks
3310 
3311 Version 1.5.3beta03 [April 16, 2011]
3312   Fixed gcc -ansi -pedantic compile. A strict ANSI system does not have
3313     snprintf, and the "__STRICT_ANSI__" detects that condition more reliably
3314     than __STDC__ (John Bowler).
3315   Removed the PNG_PTR_NORETURN attribute because it too dangerous. It tells
3316     the compiler that a user supplied callback (the error handler) does not
3317     return, yet there is no guarantee in practice that the application code
3318     will correctly implement the error handler because the compiler only
3319     issues a warning if there is a mistake (John Bowler).
3320   Removed the no-longer-used PNG_DEPSTRUCT macro.
3321   Updated the zlib version to 1.2.5 in the VStudio project.
3322   Fixed 64-bit builds where png_uint_32 is smaller than png_size_t in
3323     pngwutil.c (John Bowler).
3324   Fixed bug with stripping the filler or alpha channel when writing, that
3325     was introduced in libpng-1.5.2beta01 (bug report by Andrew Church).
3326 
3327 Version 1.5.3beta04 [April 27, 2011]
3328   Updated pngtest.png with the new zlib CMF optimization.
3329   Cleaned up conditional compilation code and of background/gamma handling
3330     Internal changes only except a new option to avoid compiling the
3331     png_build_grayscale_palette API (which is not used at all internally.)
3332     The main change is to move the transform tests (READ_TRANSFORMS,
3333     WRITE_TRANSFORMS) up one level to the caller of the APIs.  This avoids
3334     calls to spurious functions if all transforms are disabled and slightly
3335     simplifies those functions.  Pngvalid modified to handle this.
3336     A minor change is to stop the strip_16 and expand_16 interfaces from
3337     disabling each other; this allows the future alpha premultiplication
3338     code to use 16-bit intermediate values while still producing 8-bit output.
3339     png_do_background and png_do_gamma have been simplified to take a single
3340     pointer to the png_struct rather than pointers to every item required
3341     from the png_struct. This makes no practical difference to the internal
3342     code.
3343   A serious bug in the pngvalid internal routine 'standard_display_init' has
3344     been fixed - this failed to initialize the red channel and accidentally
3345     initialized the alpha channel twice.
3346   Changed png_struct jmp_buf member name from png_jmpbuf to tmp_jmpbuf to
3347     avoid a possible clash with the png_jmpbuf macro on some platforms.
3348 
3349 Version 1.5.3beta05 [May 6, 2011]
3350   Added the "_POSIX_SOURCE" feature test macro to ensure libpng sees the
3351     correct API. _POSIX_SOURCE is defined in pngpriv.h, pngtest.c and
3352     pngvalid.c to ensure that POSIX conformant systems disable non-POSIX APIs.
3353   Removed png_snprintf and added formatted warning messages.  This change adds
3354     internal APIs to allow png_warning messages to have parameters without
3355     requiring the host OS to implement snprintf.  As a side effect the
3356     dependency of the tIME-supporting RFC1132 code on stdio is removed and
3357     PNG_NO_WARNINGS does actually work now.
3358   Pass "" instead of '\0' to png_default_error() in png_err().  This mistake
3359     was introduced in libpng-1.2.20beta01.  This fixes CVE-2011-2691.
3360   Added PNG_WRITE_OPTIMIZE_CMF_SUPPORTED macro to make the zlib "CMF" byte
3361     optimization configurable.
3362   IDAT compression failed if preceded by a compressed text chunk (bug
3363     introduced in libpng-1.5.3beta01-02).  This was because the attempt to
3364     reset the zlib stream in png_write_IDAT happened after the first IDAT
3365     chunk had been deflated - much too late.  In this change internal
3366     functions were added to claim/release the z_stream and, hopefully, make
3367     the code more robust.  Also deflateEnd checking is added - previously
3368     libpng would ignore an error at the end of the stream.
3369 
3370 Version 1.5.3beta06 [May 8, 2011]
3371   Removed the -D_ALL_SOURCE from definitions for AIX in CMakeLists.txt
3372   Implemented premultiplied alpha support: png_set_alpha_mode API
3373 
3374 Version 1.5.3beta07 [May 11, 2011]
3375   Added expand_16 support to the high level interface.
3376   Added named value and 'flag' gamma support to png_set_gamma.  Made a minor
3377     change from the previous (unreleased) ABI/API to hide the exact value used
3378     for Macs - it's not a good idea to embed this in the ABI!
3379   Moved macro definitions for PNG_HAVE_IHDR, PNG_HAVE_PLTE, and PNG_AFTER_IDAT
3380     from pngpriv.h to png.h because they must be visible to applications
3381     that call png_set_unknown_chunks().
3382   Check for up->location !PNG_AFTER_IDAT when writing unknown chunks
3383     before IDAT.
3384 
3385 Version 1.5.3beta08 [May 16, 2011]
3386   Improved "pngvalid --speed" to exclude more of pngvalid from the time.
3387   Documented png_set_alpha_mode(), other changes in libpng.3/libpng-manual.txt
3388   The cHRM chunk now sets the defaults for png_set_rgb_to_gray() (when negative
3389     parameters are supplied by the caller), while in the absence of cHRM
3390     sRGB/Rec 709 values are still used.  This introduced a divide-by-zero
3391     bug in png_handle_cHRM().
3392   The bKGD chunk no longer overwrites the background value set by
3393     png_set_background(), allowing the latter to be used before the file
3394     header is read. It never performed any useful function to override
3395     the default anyway.
3396   Added memory overwrite and palette image checks to pngvalid.c
3397     Previously palette image code was poorly checked. Since the transformation
3398     code has a special palette path in most cases this was a severe weakness.
3399   Minor cleanup and some extra checking in pngrutil.c and pngrtran.c. When
3400     expanding an indexed image, always expand to RGBA if transparency is
3401     present.
3402 
3403 Version 1.5.3beta09 [May 17, 2011]
3404   Reversed earlier 1.5.3 change of transformation order; move png_expand_16
3405     back where it was.  The change doesn't work because it requires 16-bit
3406     gamma tables when the code only generates 8-bit ones.  This fails
3407     silently; the libpng code just doesn't do any gamma correction.  Moving
3408     the tests back leaves the old, inaccurate, 8-bit gamma calculations, but
3409     these are clearly better than none!
3410 
3411 Version 1.5.3beta10 [May 20, 2011]
3412 
3413   png_set_background() and png_expand_16() did not work together correctly.
3414     This problem is present in 1.5.2; if png_set_background is called with
3415     need_expand false and the matching 16 bit color libpng erroneously just
3416     treats it as an 8-bit color because of where png_do_expand_16 is in the
3417     transform list.  This simple fix reduces the supplied colour to 8-bits,
3418     so it gets smashed, but this is better than the current behavior.
3419   Added tests for expand16, more fixes for palette image tests to pngvalid.
3420     Corrects the code for palette image tests and disables attempts to
3421     validate palette colors.
3422 
3423 Version 1.5.3rc01 [June 3, 2011]
3424   No changes.
3425 
3426 Version 1.5.3rc02 [June 8, 2011]
3427   Fixed uninitialized memory read in png_format_buffer() (Bug report by
3428     Frank Busse, CVE-2011-2501, related to CVE-2004-0421).
3429 
3430 Version 1.5.3beta11 [June 11, 2011]
3431   Fixed png_handle_sCAL which is broken in 1.5. This fixes CVE 2011-2692.
3432   Added sCAL to pngtest.png
3433   Revised documentation about png_set_user_limits() to say that it also affects
3434     png writing.
3435   Revised handling of png_set_user_limits() so that it can increase the
3436     limit beyond the PNG_USER_WIDTH|HEIGHT_MAX; previously it could only
3437     reduce it.
3438   Make the 16-to-8 scaling accurate. Dividing by 256 with no rounding is
3439     wrong (high by one) 25% of the time. Dividing by 257 with rounding is
3440     wrong in 128 out of 65536 cases. Getting the right answer all the time
3441     without division is easy.
3442   Added "_SUPPORTED" to the PNG_WRITE_CUSTOMIZE_ZTXT_COMPRESSION macro.
3443   Added projects/owatcom, an IDE project for OpenWatcom to replace
3444     scripts/makefile.watcom.  This project works with OpenWatcom 1.9. The
3445     IDE autogenerates appropriate makefiles (libpng.mk) for batch processing.
3446     The project is configurable, unlike the Visual Studio project, so long
3447     as the developer has an awk.
3448   Changed png_set_gAMA to limit the gamma value range so that the inverse
3449     of the stored value cannot overflow the fixed point representation,
3450     and changed other things OpenWatcom warns about.
3451   Revised pngvalid.c to test PNG_ALPHA_MODE_SUPPORTED correctly. This allows
3452     pngvalid to build when ALPHA_MODE is not supported, which is required if
3453     it is to build on libpng 1.4.
3454   Removed string/memory macros that are no longer used and are not
3455     necessarily fully supportable, particularly png_strncpy and png_snprintf.
3456   Added log option to pngvalid.c and attempted to improve gamma messages.
3457 
3458 Version 1.5.3 [omitted]
3459   People found the presence of a beta release following an rc release
3460     to be confusing; therefore we bump the version to libpng-1.5.4beta01
3461     and there will be no libpng-1.5.3 release.
3462 
3463 Version 1.5.4beta01 [June 14, 2011]
3464   Made it possible to undefine PNG_READ_16_TO_8_ACCURATE_SCALE_SUPPORTED
3465     to get the same (inaccurate) output as libpng-1.5.2 and earlier.
3466   Moved definitions of PNG_HAVE_IHDR, PNG_AFTER_IDAT, and PNG_HAVE_PLTE
3467     outside of an unknown-chunk block in png.h because they are also
3468     needed for other uses.
3469 
3470 Version 1.5.4beta02 [June 14, 2011]
3471   Fixed and clarified LEGACY 16-to-8 scaling code.
3472   Added png_set_chop_16() API, to match inaccurate results from previous
3473     libpng versions.
3474   Removed the ACCURATE and LEGACY options (they are no longer useable)
3475   Use the old scaling method for background if png_set_chop_16() was
3476     called.
3477   Made png_set_chop_16() API removeable by disabling PNG_CHOP_16_TO_8_SUPPORTED
3478 
3479 Version 1.5.4beta03 [June 15, 2011]
3480   Fixed a problem in png_do_expand_palette() exposed by optimization in
3481     1.5.3beta06
3482   Also removed a spurious and confusing "trans" member ("trans") from png_info.
3483   The palette expand optimization prevented expansion to an intermediate RGBA
3484     form if tRNS was present but alpha was marked to be stripped; this exposed
3485     a check for tRNS in png_do_expand_palette() which is inconsistent with the
3486     code elsewhere in libpng.
3487   Correction to the expand_16 code; removed extra instance of
3488     png_set_scale_16_to_8 from pngpriv.h
3489 
3490 Version 1.5.4beta04 [June 16, 2011]
3491   Added a missing "#ifdef PNG_READ_BACKGROUND_SUPPORTED/#endif" in pngrtran.c
3492   Added PNG_TRANSFORM_CHOP_16 to the high-level read transforms.
3493   Made PNG_READ_16_TO_8_ACCURATE_SCALE configurable again.  If this is
3494     not enabled, png_set_strip_16() and png_do_scale_16_to_8() aren't built.
3495   Revised contrib/visupng, gregbook, and pngminim to demonstrate chop_16_to_8
3496 
3497 Version 1.5.4beta05 [June 16, 2011]
3498   Renamed png_set_strip_16() to png_set_scale_16() and renamed
3499     png_set_chop_16() to png_set_strip(16) in an attempt to minimize the
3500     behavior changes between libpng14 and libpng15.
3501 
3502 Version 1.5.4beta06 [June 18, 2011]
3503   Fixed new bug that was causing both strip_16 and scale_16 to be applied.
3504 
3505 Version 1.5.4beta07 [June 19, 2011]
3506   Fixed pngvalid, simplified macros, added checking for 0 in sCAL.
3507     The ACCURATE scale macro is no longer defined in 1.5 - call the
3508     png_scale_16_to_8 API.  Made sure that PNG_READ_16_TO_8 is still defined
3509     if the png_strip_16_to_8 API is present.  png_check_fp_number now
3510     maintains some state so that positive, negative and zero values are
3511     identified.  sCAL uses these to be strictly spec conformant.
3512 
3513 Version 1.5.4beta08 [June 23, 2011]
3514   Fixed pngvalid if ACCURATE_SCALE is defined.
3515   Updated scripts/pnglibconf.h.prebuilt.
3516 
3517 Version 1.5.4rc01 [June 30, 2011]
3518   Define PNG_ALLOCATED to "restrict" only if MSC_VER >= 1400.
3519 
3520 Version 1.5.4 [July 7, 2011]
3521   No changes.
3522 
3523 Version 1.5.5beta01 [July 13, 2011]
3524   Fixed some typos and made other minor changes in the manual.
3525   Updated contrib/pngminus/makefile.std (Samuli Souminen)
3526 
3527 Version 1.5.5beta02 [July 14, 2011]
3528   Revised Makefile.am and Makefile.in to look in the right directory for
3529     pnglibconf.h.prebuilt
3530 
3531 Version 1.5.5beta03 [July 27, 2011]
3532   Enabled compilation with g++ compiler.  This compiler does not recognize
3533     the file extension, so it always compiles with C++ rules.  Made minor
3534     changes to pngrutil.c to cast results where C++ expects it but C does not.
3535   Minor editing of libpng.3 and libpng-manual.txt.
3536 
3537 Version 1.5.5beta04 [July 29, 2011]
3538   Revised CMakeLists.txt (Clifford Yapp)
3539   Updated commentary about the png_rgb_to_gray() default coefficients
3540     in the manual and in pngrtran.c
3541 
3542 Version 1.5.5beta05 [August 17, 2011]
3543   Prevent unexpected API exports from non-libpng DLLs on Windows.  The "_DLL"
3544     is removed from the test of whether a DLL is being built (this erroneously
3545     caused the libpng APIs to be marked as DLL exports in static builds under
3546     Microsoft Visual Studio).  Almost all of the libpng building configuration
3547     is moved from pngconf.h to pngpriv.h, but PNG_DLL_EXPORT remains in
3548     pngconf.h, though, so that it is colocated with the import definition (it
3549     is no longer used anywhere in the installed headers).  The VStudio project
3550     definitions have been cleaned up: "_USRDLL" has been removed from the
3551     static library builds (this was incorrect), and PNG_USE_DLL has been added
3552     to pngvalid to test the functionality (pngtest does not supply it,
3553     deliberately).  The spurious "_EXPORTS" has been removed from the
3554     libpng build (all these errors were a result of copy/paste between project
3555     configurations.)
3556   Added new types and internal functions for CIE RGB end point handling to
3557     pngpriv.h (functions yet to be implemented).
3558 
3559 Version 1.5.5beta06 [August 26, 2011]
3560   Ensure the CMAKE_LIBRARY_OUTPUT_DIRECTORY is set in CMakeLists.txt
3561     (Clifford Yap)
3562   Fixes to rgb_to_gray and cHRM XYZ APIs (John Bowler):
3563     The rgb_to_gray code had errors when combined with gamma correction.
3564     Some pixels were treated as true grey when they weren't and such pixels
3565     and true grey ones were not gamma corrected (the original value of the
3566     red component was used instead).  APIs to get and set cHRM using color
3567     space end points have been added and the rgb_to_gray code that defaults
3568     based on cHRM, and the divide-by-zero bug in png_handle_cHRM (CERT
3569     VU#477046, CVE-2011-3328, introduced in 1.5.4) have been corrected.
3570   A considerable number of tests has been added to pngvalid for the
3571     rgb_to_gray transform.
3572   Arithmetic errors in rgb_to_gray whereby the calculated gray value was
3573     truncated to the bit depth rather than rounded have been fixed except in
3574     the 8-bit non-gamma-corrected case (where consistency seems more important
3575     than correctness.)  The code still has considerable inaccuracies in the
3576     8-bit case because 8-bit linear arithmetic is used.
3577 
3578 Version 1.5.5beta07 [September 7, 2011]
3579   Added "$(ARCH)" option to makefile.darwin
3580   Added SunOS support to configure.ac and Makefile.am
3581   Changed png_chunk_benign_error() to png_warning() in png.c, in
3582     png_XYZ_from_xy_checked().
3583 
3584 Version 1.5.5beta08 [September 10, 2011]
3585   Fixed 64-bit compilation errors (gcc). The errors fixed relate
3586     to conditions where types that are 32 bits in the GCC 32-bit
3587     world (uLong and png_size_t) become 64 bits in the 64-bit
3588     world.  This produces potential truncation errors which the
3589     compiler correctly flags.
3590   Relocated new HAVE_SOLARIS_LD definition in configure.ac
3591   Constant changes for 64-bit compatibility (removal of L suffixes). The
3592     16-bit cases still use "L" as we don't have a 16-bit test system.
3593 
3594 Version 1.5.5rc01 [September 15, 2011]
3595   Removed "L" suffixes in pngpriv.h
3596 
3597 Version 1.5.5 [September 22, 2011]
3598   No changes.
3599 
3600 Version 1.5.6beta01 [September 22, 2011]
3601   Fixed some 64-bit type conversion warnings in pngrtran.c
3602   Moved row_info from png_struct to a local variable.
3603   The various interlace mask arrays have been made into arrays of
3604     bytes and made PNG_CONST and static (previously some arrays were
3605     marked PNG_CONST and some weren't).
3606   Additional checks have been added to the transform code to validate the
3607     pixel depths after the transforms on both read and write.
3608   Removed some redundant code from pngwrite.c, in png_destroy_write_struct().
3609   Changed chunk reading/writing code to use png_uint_32 instead of png_byte[4].
3610     This removes the need to allocate temporary strings for chunk names on
3611     the stack in the read/write code.  Unknown chunk handling still uses the
3612     string form because this is exposed in the API.
3613 
3614 Version 1.5.6beta02 [September 26, 2011]
3615   Added a note in the manual the png_read_update_info() must be called only
3616     once with a particular info_ptr.
3617   Fixed a typo in the definition of the new PNG_STRING_FROM_CHUNK(s,c) macro.
3618 
3619 Version 1.5.6beta03 [September 28, 2011]
3620   Revised test-pngtest.sh to report FAIL when pngtest fails.
3621   Added "--strict" option to pngtest, to report FAIL when the failure is
3622     only because the resulting valid files are different.
3623   Revised CMakeLists.txt to work with mingw and removed some material from
3624     CMakeLists.txt that is no longer useful in libpng-1.5.
3625 
3626 Version 1.5.6beta04 [October 5, 2011]
3627   Fixed typo in Makefile.in and Makefile.am ("-M Wl" should be "-M -Wl")."
3628 
3629 Version 1.5.6beta05 [October 12, 2011]
3630   Speed up png_combine_row() for interlaced images. This reduces the generality
3631     of the code, allowing it to be optimized for Adam7 interlace.  The masks
3632     passed to png_combine_row() are now generated internally, avoiding
3633     some code duplication and localizing the interlace handling somewhat.
3634   Align png_struct::row_buf - previously it was always unaligned, caused by
3635     a bug in the code that attempted to align it; the code needs to subtract
3636     one from the pointer to take account of the filter byte prepended to
3637     each row.
3638   Optimized png_combine_row() when rows are aligned. This gains a small
3639     percentage for 16-bit and 32-bit pixels in the typical case where the
3640     output row buffers are appropriately aligned. The optimization was not
3641     previously possible because the png_struct buffer was always misaligned.
3642   Fixed bug in png_write_chunk_header() debug print, introduced in 1.5.6beta01.
3643 
3644 Version 1.5.6beta06 [October 17, 2011]
3645   Removed two redundant tests for uninitialized row.
3646   Fixed a relatively harmless memory overwrite in compressed text writing
3647     with a 1 byte zlib buffer.
3648   Add ability to call png_read_update_info multiple times to pngvalid.c.
3649   Fixes for multiple calls to png_read_update_info. These fixes attend to
3650     most of the errors revealed in pngvalid, however doing the gamma work
3651     twice results in inaccuracies that can't be easily fixed.  There is now
3652     a warning in the code if this is going to happen.
3653   Turned on multiple png_read_update_info in pngvalid transform tests.
3654   Prevent libpng from overwriting unused bits at the end of the image when
3655     it is not byte aligned, while reading. Prior to libpng-1.5.6 libpng would
3656     overwrite the partial byte at the end of each row if the row width was not
3657     an exact multiple of 8 bits and the image is not interlaced.
3658 
3659 Version 1.5.6beta07 [October 21, 2011]
3660   Made png_ptr->prev_row an aligned pointer into png_ptr->big_prev_row
3661     (Mans Rullgard).
3662 
3663 Version 1.5.6rc01 [October 26, 2011]
3664   Changed misleading "Missing PLTE before cHRM" warning to "Out of place cHRM"
3665 
3666 Version 1.5.6rc02 [October 27, 2011]
3667   Added LSR() macro to defend against buggy compilers that evaluate non-taken
3668     code branches and complain about out-of-range shifts.
3669 
3670 Version 1.5.6rc03 [October 28, 2011]
3671   Renamed the LSR() macro to PNG_LSR() and added PNG_LSL() macro.
3672   Fixed compiler warnings with Intel and MSYS compilers. The logical shift
3673     fix for Microsoft Visual C is required by other compilers, so this
3674     enables that fix for all compilers when using compile-time constants.
3675     Under MSYS 'byte' is a name declared in a system header file, so we
3676     changed the name of a local variable to avoid the warnings that result.
3677   Added #define PNG_ALIGN_TYPE PNG_ALIGN_NONE to contrib/pngminim/*/pngusr.h
3678 
3679 Version 1.5.6 [November 3, 2011]
3680   No changes.
3681 
3682 Version 1.5.7beta01 [November 4, 2011]
3683   Added support for ARM processor, when decoding all PNG up-filtered rows
3684     and any other-filtered rows with 3 or 4 bytes per pixel (Mans Rullgard).
3685   Fixed bug in pngvalid on early allocation failure; fixed type cast in
3686     pngmem.c; pngvalid would attempt to call png_error() if the allocation
3687     of a png_struct or png_info failed. This would probably have led to a
3688     crash.  The pngmem.c implementation of png_malloc() included a cast
3689     to png_size_t which would fail on large allocations on 16-bit systems.
3690   Fix for the preprocessor of the Intel C compiler. The preprocessor
3691     splits adjacent @ signs with a space; this changes the concatenation
3692     token from @-@-@ to PNG_JOIN; that should work with all compiler
3693     preprocessors.
3694   Paeth filter speed improvements from work by Siarhei Siamashka. This
3695     changes the 'Paeth' reconstruction function to improve the GCC code
3696     generation on x86. The changes are only part of the suggested ones;
3697     just the changes that definitely improve speed and remain simple.
3698     The changes also slightly increase the clarity of the code.
3699 
3700 Version 1.5.7beta02 [November 11, 2011]
3701   Check compression_type parameter in png_get_iCCP and remove spurious
3702     casts. The compression_type parameter is always assigned to, so must
3703     be non-NULL. The cast of the profile length potentially truncated the
3704     value unnecessarily on a 16-bit int system, so the cast of the (byte)
3705     compression type to (int) is specified by ANSI-C anyway.
3706   Fixed FP division by zero in pngvalid.c; the 'test_pixel' code left
3707     the sBIT fields in the test pixel as 0, which resulted in a floating
3708     point division by zero which was irrelevant but causes systems where
3709     FP exceptions cause a crash. Added code to pngvalid to turn on FP
3710     exceptions if the appropriate glibc support is there to ensure this is
3711     tested in the future.
3712   Updated scripts/pnglibconf.mak and scripts/makefile.std to handle the
3713     new PNG_JOIN macro.
3714   Added versioning to pnglibconf.h comments.
3715   Simplified read/write API initial version; basic read/write tested on
3716     a variety of images, limited documentation (in the header file.)
3717   Installed more accurate linear to sRGB conversion tables. The slightly
3718     modified tables reduce the number of 16-bit values that
3719     convert to an off-by-one 8-bit value.  The "makesRGB.c" code that was used
3720     to generate the tables is now in a contrib/sRGBtables sub-directory.
3721 
3722 Version 1.5.7beta03 [November 17, 2011]
3723   Removed PNG_CONST from the sRGB table declarations in pngpriv.h and png.c
3724   Added run-time detection of NEON support.
3725   Added contrib/libtests; includes simplified API test and timing test and
3726     a color conversion utility for rapid checking of failed 'pngstest' results.
3727   Multiple transform bug fixes plus a work-round for double gamma correction.
3728     libpng does not support more than one transform that requires linear data
3729     at once - if this is tried typically the results is double gamma
3730     correction. Since the simplified APIs can need rgb to gray combined with
3731     a compose operation it is necessary to do one of these outside the main
3732     libpng transform code. This check-in also contains fixes to various bugs
3733     in the simplified APIs themselves and to some bugs in compose and rgb to
3734     gray (on palette) itself.
3735   Fixes for C++ compilation using g++ When libpng source is compiled
3736     using g++. The compiler imposes C++ rules on the C source; thus it
3737     is desirable to make the source work with either C or C++ rules
3738     without throwing away useful error information.  This change adds
3739     png_voidcast to allow C semantic (void*) cases or the corresponding
3740     C++ static_cast operation, as appropriate.
3741   Added --noexecstack to assembler file compilation. GCC does not set
3742     this on assembler compilation, even though it does on C compilation.
3743     This creates security issues if assembler code is enabled; the
3744     work-around is to set it by default in the flags for $(CCAS)
3745   Work around compilers that don't support declaration of const data. Some
3746     compilers fault 'extern const' data declarations (because the data is
3747     not initialized); this turns on const-ness only for compilers where
3748     this is known to work.
3749 
3750 Version 1.5.7beta04 [November 17, 2011]
3751   Since the gcc driver does not recognize the --noexecstack flag, we must
3752     use the -Wa prefix to have it passed through to the assembler.
3753     Also removed a duplicate setting of this flag.
3754   Added files that were omitted from the libpng-1.5.7beta03 zip distribution.
3755 
3756 Version 1.5.7beta05 [November 25, 2011]
3757   Removed "zTXt" from warning in generic chunk decompression function.
3758   Validate time settings passed to png_set_tIME() and png_convert_to_rfc1123()
3759     (Frank Busse). Note: This prevented CVE-2015-7981 from affecting
3760     libpng-1.5.7 and later.
3761   Added MINGW support to CMakeLists.txt
3762   Reject invalid compression flag or method when reading the iTXt chunk.
3763   Backed out 'simplified' API changes. The API seems too complex and there
3764     is a lack of consensus or enthusiasm for the proposals.  The API also
3765     reveals significant bugs inside libpng (double gamma correction and the
3766     known bug of being unable to retrieve a corrected palette). It seems
3767     better to wait until the bugs, at least, are corrected.
3768   Moved pngvalid.c into contrib/libtests
3769   Rebuilt Makefile.in, configure, etc., with autoconf-2.68
3770 
3771 Version 1.5.7rc01 [December 1, 2011]
3772   Replaced an "#if" with "#ifdef" in pngrtran.c
3773   Revised #if PNG_DO_BC block in png.c (use #ifdef and add #else)
3774 
3775 Version 1.5.7rc02 [December 5, 2011]
3776   Revised project files and contrib/pngvalid/pngvalid.c to account for
3777     the relocation of pngvalid into contrib/libtests.
3778   Revised pngconf.h to use " __declspec(restrict)" only when MSC_VER >= 1400,
3779     as in libpng-1.5.4.
3780   Put CRLF line endings in the owatcom project files.
3781 
3782 Version 1.5.7rc03 [December 7, 2011]
3783   Updated CMakeLists.txt to account for the relocation of pngvalid.c
3784 
3785 Version 1.5.7 [December 15, 2011]
3786   Minor fixes to pngvalid.c for gcc 4.6.2 compatibility to remove warnings
3787     reported by earlier versions.
3788   Fixed minor memset/sizeof errors in pngvalid.c.
3789 
3790 Version 1.6.0beta01 [December 15, 2011]
3791   Removed machine-generated configure files from the GIT repository (they will
3792     continue to appear in the tarball distributions and in the libpng15 and
3793     earlier GIT branches).
3794   Restored the new 'simplified' API, which was started in libpng-1.5.7beta02
3795     but later deleted from libpng-1.5.7beta05.
3796   Added example programs for the new 'simplified' API.
3797   Added ANSI-C (C90) headers and require them, and take advantage of the
3798     change. Also fixed some of the projects/* and contrib/* files that needed
3799     updates for libpng16 and the move of pngvalid.c.
3800     With this change the required ANSI-C header files are assumed to exist: the
3801     implementation must provide float.h, limits.h, stdarg.h and stddef.h and
3802     libpng relies on limits.h and stddef.h existing and behaving as defined
3803     (the other two required headers aren't used).  Non-ANSI systems that don't
3804     have stddef.h or limits.h will have to provide an appropriate fake
3805     containing the relevant types and #defines.
3806   Dropped support for 16-bit platforms. The use of FAR/far has been eliminated
3807     and the definition of png_alloc_size_t is now controlled by a flag so
3808     that 'small size_t' systems can select it if necessary.  Libpng 1.6 may
3809     not currently work on such systems -- it seems likely that it will
3810     ask 'malloc' for more than 65535 bytes with any image that has a
3811     sufficiently large row size (rather than simply failing to read such
3812     images).
3813   New tools directory containing tools used to generate libpng code.
3814   Fixed race conditions in parallel make builds. With higher degrees of
3815     parallelism during 'make' the use of the same temporary file names such
3816     as 'dfn*' can result in a race where a temporary file from one arm of the
3817     build is deleted or overwritten in another arm.  This changes the
3818     temporary files for suffix rules to always use $* and ensures that the
3819     non-suffix rules use unique file names.
3820 
3821 Version 1.6.0beta02 [December 21, 2011]
3822   Correct configure builds where build and source directories are separate.
3823     The include path of 'config.h' was erroneously made relative in pngvalid.c
3824     in libpng 1.5.7.
3825 
3826 Version 1.6.0beta03 [December 22, 2011]
3827   Start-up code size improvements, error handler flexibility. These changes
3828     alter how the tricky allocation of the initial png_struct and png_info
3829     structures are handled. png_info is now handled in pretty much the same
3830     way as everything else, except that the allocations handle NULL return
3831     silently.  png_struct is changed in a similar way on allocation and on
3832     deallocation a 'safety' error handler is put in place (which should never
3833     be required).  The error handler itself is changed to permit mismatches
3834     in the application and libpng error buffer size; however, this means a
3835     silent change to the API to return the jmp_buf if the size doesn't match
3836     the size from the libpng compilation; libpng now allocates the memory and
3837     this may fail.  Overall these changes result in slight code size
3838     reductions; however, this is a reduction in code that is always executed
3839     so is particularly valuable.  Overall on a 64-bit system the libpng DLL
3840     decreases in code size by 1733 bytes.  pngerror.o increases in size by
3841     about 465 bytes because of the new functionality.
3842   Added png_convert_to_rfc1123_buffer() and deprecated png_convert_to_rfc1123()
3843     to avoid including a spurious buffer in the png_struct.
3844 
3845 Version 1.6.0beta04 [December 30, 2011]
3846   Regenerated configure scripts with automake-1.11.2
3847   Eliminated png_info_destroy(). It is now used only in png.c and only calls
3848     one other internal function and memset().
3849   Enabled png_get_sCAL_fixed() if floating point APIs are enabled. Previously
3850     it was disabled whenever internal fixed point arithmetic was selected,
3851     which meant it didn't exist even on systems where FP was available but not
3852     preferred.
3853   Added pngvalid.c compile time checks for const APIs.
3854   Implemented 'restrict' for png_info and png_struct. Because of the way
3855     libpng works both png_info and png_struct are always accessed via a
3856     single pointer.  This means adding C99 'restrict' to the pointer gives
3857     the compiler some opportunity to optimize the code.  This change allows
3858     that.
3859   Moved AC_MSG_CHECKING([if libraries can be versioned]) later to the proper
3860     location in configure.ac (Gilles Espinasse).
3861   Changed png_memcpy to C assignment where appropriate. Changed all those
3862     uses of png_memcpy that were doing a simple assignment to assignments
3863     (all those cases where the thing being copied is a non-array C L-value).
3864   Added some error checking to png_set_*() routines.
3865   Removed the reference to the non-exported function png_memcpy() from
3866     example.c.
3867   Fixed the Visual C 64-bit build - it requires jmp_buf to be aligned, but
3868     it had become misaligned.
3869   Revised contrib/pngminus/pnm2png.c to avoid warnings when png_uint_32
3870     and unsigned long are of different sizes.
3871 
3872 Version 1.6.0beta05 [January 15, 2012]
3873   Updated manual with description of the simplified API (copied from png.h)
3874   Fix bug in pngerror.c: some long warnings were being improperly truncated
3875     (CVE-2011-3464, bug introduced in libpng-1.5.3beta05).
3876 
3877 Version 1.6.0beta06 [January 24, 2012]
3878   Added palette support to the simplified APIs. This commit
3879     changes some of the macro definitions in png.h, app code
3880     may need corresponding changes.
3881   Increased the formatted warning buffer to 192 bytes.
3882   Added color-map support to simplified API. This is an initial version for
3883     review; the documentation has not yet been updated.
3884   Fixed Min/GW uninstall to remove libpng.dll.a
3885 
3886 Version 1.6.0beta07 [January 28, 2012]
3887   Eliminated Intel icc/icl compiler warnings. The Intel (GCC derived)
3888     compiler issues slightly different warnings from those issued by the
3889     current vesions of GCC. This eliminates those warnings by
3890     adding/removing casts and small code rewrites.
3891   Updated configure.ac from autoupdate: added --enable-werror option.
3892     Also some layout regularization and removal of introduced tab characters
3893     (replaced with 3-character indentation).  Obsolete macros identified by
3894     autoupdate have been removed; the replacements are all in 2.59 so
3895     the pre-req hasn't been changed.  --enable-werror checks for support
3896     for -Werror (or the given argument) in the compiler.  This mimics the
3897     gcc configure option by allowing -Werror to be turned on safely; without
3898     the option the tests written in configure itself fail compilation because
3899     they cause compiler warnings.
3900   Rewrote autogen.sh to run autoreconf instead of running tools one-by-one.
3901   Conditionalize the install rules for MINGW and CYGWIN in CMakeLists.txt and
3902     set CMAKE_LIBRARY_OUTPUT_DIRECTORY to "lib" on all platforms (C. Yapp).
3903   Freeze libtool files in the 'scripts' directory. This version of autogen.sh
3904     attempts to dissuade people from running it when it is not, or should not,
3905     be necessary.  In fact, autogen.sh does not work when run in a libpng
3906     directory extracted from a tar distribution anymore. You must run it in
3907     a GIT clone instead.
3908   Added two images to contrib/pngsuite (1-bit and 2-bit transparent grayscale),
3909     and renamed three whose names were inconsistent with those in
3910     pngsuite/README.txt.
3911 
3912 Version 1.6.0beta08 [February 1, 2012]
3913   Fixed Image::colormap misalignment in pngstest.c
3914   Check libtool/libtoolize version number (2.4.2) in configure.ac
3915   Divide test-pngstest.sh into separate pngstest runs for basic and
3916     transparent images.
3917   Moved automake options to AM_INIT_AUTOMAKE in configure.ac
3918   Added color-tests, silent-rules (Not yet implemented in Makefile.am) and
3919     version checking to configure.ac
3920   Improved pngstest speed by not doing redundant tests and add const to
3921     the background parameter of png_image_finish_read. The --background
3922     option is now done automagically only when required, so that commandline
3923     option no longer exists.
3924   Cleaned up pngpriv.h to consistently declare all functions and data.
3925     Also eliminated PNG_CONST_DATA, which is apparently not needed but we
3926     can't be sure until it is gone.
3927   Added symbol prefixing that allows all the libpng external symbols
3928     to be prefixed (suggested by Reuben Hawkins).
3929   Updated "ftbb*.png" list in the owatcom and vstudio projects.
3930   Fixed 'prefix' builds on clean systems. The generation of pngprefix.h
3931     should not require itself.
3932   Updated INSTALL to explain that autogen.sh must be run in a GIT clone,
3933     not in a libpng directory extracted from a tar distribution.
3934 
3935 Version 1.6.0beta09 [February 1, 2012]
3936   Reverted the prebuilt configure files to libpng-1.6.0beta05 condition.
3937 
3938 Version 1.6.0beta10 [February 3, 2012]
3939   Added Z_SOLO for zlib-1.2.6+ and correct pngstest tests
3940   Updated list of test images in CMakeLists.txt
3941   Updated the prebuilt configure files to current condition.
3942   Revised INSTALL information about autogen.sh; it works in tar distributions.
3943 
3944 Version 1.6.0beta11 [February 16, 2012]
3945   Fix character count in pngstest command in projects/owatcom/pngstest.tgt
3946   Revised test-pngstest.sh to report PASS/FAIL for each image.
3947   Updated documentation about the simplified API.
3948   Corrected estimate of error in libpng png_set_rgb_to_gray API.  The API is
3949     extremely inaccurate for sRGB conversions because it uses an 8-bit
3950     intermediate linear value and it does not use the sRGB transform, so it
3951     suffers from the known instability in gamma transforms for values close
3952     to 0 (see Poynton).  The net result is that the calculation has a maximum
3953     error of 14.99/255; 0.5/255^(1/2.2).  pngstest now uses 15 for the
3954     permitted 8-bit error. This may still not be enough because of arithmetic
3955     error.
3956   Removed some unused arrays (with #ifdef) from png_read_push_finish_row().
3957   Fixed a memory overwrite bug in simplified read of RGB PNG with
3958     non-linear gamma Also bugs in the error checking in pngread.c and changed
3959     quite a lot of the checks in pngstest.c to be correct; either correctly
3960     written or not over-optimistic.  The pngstest changes are insufficient to
3961     allow all possible RGB transforms to be passed; pngstest cmppixel needs
3962     to be rewritten to make it clearer which errors it allows and then changed
3963     to permit known inaccuracies.
3964   Removed tests for no-longer-used *_EMPTY_PLTE_SUPPORTED from pngstruct.h
3965   Fixed fixed/float API export conditionals. 1) If FIXED_POINT or
3966     FLOATING_POINT options were switched off, png.h ended up with lone ';'
3967     characters.  This is not valid ANSI-C outside a function.  The ';'
3968     characters have been moved inside the definition of PNG_FP_EXPORT and
3969     PNG_FIXED_EXPORT. 2) If either option was switched off, the declaration
3970     of the corresponding functions were completely omitted, even though some
3971     of them are still used internally.  The result is still valid, but
3972     produces warnings from gcc with some warning options (including -Wall). The
3973     fix is to cause png.h to declare the functions with PNG_INTERNAL_FUNCTION
3974     when png.h is included from pngpriv.h.
3975   Check for invalid palette index while reading paletted PNG.  When one is
3976     found, issue a warning and increase png_ptr->num_palette accordingly.
3977     Apps are responsible for checking to see if that happened.
3978 
3979 Version 1.6.0beta12 [February 18, 2012]
3980   Do not increase num_palette on invalid_index.
3981   Relocated check for invalid palette index to pngrtran.c, after unpacking
3982     the sub-8-bit pixels.
3983   Fixed CVE-2011-3026 buffer overrun bug.  This bug was introduced when
3984     iCCP chunk support was added at libpng-1.0.6. Deal more correctly with the
3985     test on iCCP chunk length. Also removed spurious casts that may hide
3986     problems on 16-bit systems.
3987 
3988 Version 1.6.0beta13 [February 24, 2012]
3989   Eliminated redundant png_push_read_tEXt|zTXt|iTXt|unknown code from
3990     pngpread.c and use the sequential png_handle_tEXt, etc., in pngrutil.c;
3991     now that png_ptr->buffer is inaccessible to applications, the special
3992     handling is no longer useful.
3993   Added PNG_SAFE_LIMITS feature to pnglibconf.dfa, pngpriv.h, and new
3994     pngusr.dfa to reset the user limits to safe ones if PNG_SAFE_LIMITS is
3995     defined.  To enable, use "CPPFLAGS=-DPNG_SAFE_LIMITS_SUPPORTED=1" on the
3996     configure command or put #define PNG_SAFE_LIMITS_SUPPORTED in
3997     pnglibconf.h.prebuilt and pnglibconf.h.
3998 
3999 Version 1.6.0beta14 [February 27, 2012]
4000   Added information about the new limits in the manual.
4001   Updated Makefile.in
4002 
4003 Version 1.6.0beta15 [March 2, 2012]
4004   Removed unused "current_text" members of png_struct and the png_free()
4005     of png_ptr->current_text from pngread.c
4006   Rewrote pngstest.c for substantial speed improvement.
4007   Fixed transparent pixel and 16-bit rgb tests in pngstest and removed a
4008     spurious check in pngwrite.c
4009   Added PNG_IMAGE_FLAG_FAST for the benefit of applications that store
4010     intermediate files, or intermediate in-memory data, while processing
4011     image data with the simplified API.  The option makes the files larger
4012     but faster to write and read.  pngstest now uses this by default; this
4013     can be disabled with the --slow option.
4014   Improved pngstest fine tuning of error numbers, new test file generator.
4015     The generator generates images that test the full range of sample values,
4016     allow the error numbers in pngstest to be tuned and checked.  makepng
4017     also allows generation of images with extra chunks, although this is
4018     still work-in-progress.
4019   Added check for invalid palette index while reading.
4020   Fixed some bugs in ICC profile writing. The code should now accept
4021     all potentially valid ICC profiles and reject obviously invalid ones.
4022     It now uses png_error() to do so rather than casually writing a PNG
4023     without the necessary color data.
4024   Removed whitespace from the end of lines in all source files and scripts.
4025 
4026 Version 1.6.0beta16 [March 6, 2012]
4027   Relocated palette-index checking function from pngrutil.c to pngtrans.c
4028   Added palette-index checking while writing.
4029   Changed png_inflate() and calling routines to avoid overflow problems.
4030     This is an intermediate check-in that solves the immediate problems and
4031     introduces one performance improvement (avoiding a copy via png_ptr->zbuf.)
4032     Further changes will be made to make ICC profile handling more secure.
4033   Fixed build warnings (MSVC, GCC, GCC v3). Cygwin GCC with default options
4034     declares 'index' as a global, causing a warning if it is used as a local
4035     variable.  GCC 64-bit warns about assigning a (size_t) (unsigned 64-bit)
4036     to an (int) (signed 32-bit).  MSVC, however, warns about using the
4037     unary '-' operator on an unsigned value (even though it is well defined
4038     by ANSI-C to be ~x+1).  The padding calculation was changed to use a
4039     different method.  Removed the tests on png_ptr->pass.
4040   Added contrib/libtests/tarith.c to test internal arithmetic functions from
4041     png.c. This is a libpng maintainer program used to validate changes to the
4042     internal arithmetic functions.
4043   Made read 'inflate' handling like write 'deflate' handling. The read
4044     code now claims and releases png_ptr->zstream, like the write code.
4045     The bug whereby the progressive reader failed to release the zstream
4046     is now fixed, all initialization is delayed, and the code checks for
4047     changed parameters on deflate rather than always calling
4048     deflatedEnd/deflateInit.
4049   Validate the zTXt strings in pngvalid.
4050   Added code to validate the windowBits value passed to deflateInit2().
4051     If the call to deflateInit2() is wrong a png_warning will be issued
4052     (in fact this is harmless, but the PNG data produced may be sub-optimal).
4053 
4054 Version 1.6.0beta17 [March 10, 2012]
4055   Fixed PNG_LIBPNG_BUILD_BASE_TYPE definition. 
4056   Reject all iCCP chunks after the first, even if the first one is invalid.
4057   Deflate/inflate was reworked to move common zlib calls into single
4058     functions [rw]util.c.  A new shared keyword check routine was also added
4059     and the 'zbuf' is no longer allocated on progressive read.  It is now
4060     possible to call png_inflate() incrementally.  A warning is no longer
4061     issued if the language tag or translated keyword in the iTXt chunk
4062     has zero length.
4063   If benign errors are disabled use maximum window on ancillary inflate.
4064     This works round a bug introduced in 1.5.4 where compressed ancillary
4065     chunks could end up with a too-small windowBits value in the deflate
4066     header.
4067 
4068 Version 1.6.0beta18 [March 16, 2012]
4069   Issue a png_benign_error() instead of png_warning() about bad palette index.
4070   In pngtest, treat benign errors as errors if "-strict" is present.
4071   Fixed an off-by-one error in the palette index checking function.
4072   Fixed a compiler warning under Cygwin (Windows-7, 32-bit system)
4073   Revised example.c to put text strings in a temporary character array
4074     instead of directly assigning string constants to png_textp members.
4075     This avoids compiler warnings when -Wwrite-strings is enabled.
4076   Added output flushing to aid debugging under Visual Studio. Unfortunately
4077     this is necessary because the VS2010 output window otherwise simply loses
4078     the error messages on error (they weren't flushed to the window before
4079     the process exited, apparently!)
4080   Added configuration support for benign errors and changed the read
4081     default. Also changed some warnings in the iCCP and sRGB handling
4082     from to benign errors. Configuration now makes read benign
4083     errors warnings and write benign errors to errors by default (thus
4084     changing the behavior on read).  The simplified API always forces
4085     read benign errors to warnings (regardless of the system default, unless
4086     this is disabled in which case the simplified API can't be built.)
4087 
4088 Version 1.6.0beta19 [March 18, 2012]
4089   Work around for duplicate row start calls; added warning messages.
4090     This turns on PNG_FLAG_DETECT_UNINITIALIZED to detect app code that
4091     fails to call one of the 'start' routines (not enabled in libpng-1.5
4092     because it is technically an API change, since it did normally work
4093     before.)  It also makes duplicate calls to png_read_start_row (an
4094     internal function called at the start of the image read) benign, as
4095     they were before changes to use png_inflate_claim. Somehow webkit is
4096     causing this to happen; this is probably a mis-feature in the zlib
4097     changes so this commit is only a work-round.
4098   Removed erroneous setting of DETECT_UNINITIALIZED and added more
4099     checks. The code now does a png_error if an attempt is made to do the
4100     row initialization twice; this is an application error and it has
4101     serious consequences because the transform data in png_struct is
4102     changed by each call.
4103   Added application error reporting and added chunk names to read
4104     benign errors; also added --strict to pngstest - not enabled
4105     yet because a warning is produced.
4106   Avoid the double gamma correction warning in the simplified API.
4107     This allows the --strict option to pass in the pngstest checks
4108 
4109 Version 1.6.0beta20 [March 29, 2012]
4110   Changed chunk handler warnings into benign errors, incrementally load iCCP
4111   Added checksum-icc.c to contrib/tools
4112   Prevent PNG_EXPAND+PNG_SHIFT doing the shift twice.
4113   Recognize known sRGB ICC profiles while reading; prefer writing the
4114     iCCP profile over writing the sRGB chunk, controlled by the
4115     PNG_sRGB_PROFILE_CHECKS option.
4116   Revised png_set_text_2() to avoid potential memory corruption (fixes
4117     CVE-2011-3048, also known as CVE-2012-3425).
4118 
4119 Version 1.6.0beta21 [April 27, 2012]
4120   Revised scripts/makefile.darwin: use system zlib; remove quotes around
4121     architecture list; add missing ppc architecture; add architecture options
4122     to shared library link; don't try to create a shared lib based on missing
4123     RELEASE variable.
4124   Enable png_set_check_for_invalid_index() for both read and write.
4125   Removed #ifdef PNG_HANDLE_AS_UNKNOWN_SUPPORTED in pngpriv.h around
4126     declaration of png_handle_unknown().
4127   Added -lssp_nonshared in a comment in scripts/makefile.freebsd
4128     and changed deprecated NOOBJ and NOPROFILE to NO_OBJ and NO_PROFILE.
4129 
4130 Version 1.6.0beta22 [May 23, 2012]
4131   Removed need for -Wno-cast-align with clang.  clang correctly warns on
4132     alignment increasing pointer casts when -Wcast-align is passed. This
4133     fixes the cases that clang warns about either by eliminating the
4134     casts from png_bytep to png_uint_16p (pngread.c), or, for pngrutil.c
4135     where the cast is previously verified or pngstest.c where it is OK, by
4136     introducing new png_aligncast macros to do the cast in a way that clang
4137     accepts.
4138 
4139 Version 1.6.0beta23 [June 6, 2012]
4140   Revised CMakeLists.txt to not attempt to make a symlink under mingw.
4141   Made fixes for new optimization warnings from gcc 4.7.0. The compiler
4142     performs an optimization which is safe; however it then warns about it.
4143     Changing the type of 'palette_number' in pngvalid.c removes the warning.
4144   Do not depend upon a GCC feature macro being available for use in generating
4145     the linker mapfile symbol prefix.
4146   Improved performance of new do_check_palette_indexes() function (only
4147     update the value when it actually increases, move test for whether
4148     the check is wanted out of the function.
4149 
4150 Version 1.6.0beta24 [June 7, 2012]
4151   Don't check palette indexes if num_palette is 0 (as it can be in MNG files).
4152 
4153 Version 1.6.0beta25 [June 16, 2012]
4154   Revised png_set_keep_unknown_chunks() so num_chunks < 0 means ignore all
4155     unknown chunks and all known chunks except for IHDR, PLTE, tRNS, IDAT,
4156     and IEND.  Previously it only meant ignore all unknown chunks, the
4157     same as num_chunks == 0. Revised png_image_skip_unused_chunks() to
4158     provide a list of chunks to be processed instead of a list of chunks to
4159     ignore.  Revised contrib/gregbook/readpng2.c accordingly.
4160 
4161 Version 1.6.0beta26 [July 10, 2012]
4162   Removed scripts/makefile.cegcc from the *.zip and *.7z distributions; it
4163     depends on configure, which is not included in those archives.
4164   Moved scripts/chkfmt to contrib/tools.
4165   Changed "a+w" to "u+w" in Makefile.in to fix CVE-2012-3386.
4166 
4167 Version 1.6.0beta27 [August 11, 2012]
4168   Do not compile PNG_DEPRECATED, PNG_ALLOC and PNG_PRIVATE when __GNUC__ < 3.
4169   Do not use __restrict when GNUC is <= 3.1
4170   Removed references to png_zalloc() and png_zfree() from the manual.
4171   Fixed configurations where floating point is completely disabled.  Because
4172     of the changes to support symbol prefixing PNG_INTERNAL_FUNCTION declares
4173     floating point APIs during libpng builds even if they are completely
4174     disabled. This requires the png floating point types (png_double*) to be
4175     declared even though the functions are never actually defined.  This
4176     change provides a dummy definition so that the declarations work, yet any
4177     implementation will fail to compile because of an incomplete type.
4178   Re-eliminated the use of strcpy() in pngtest.c.  An unnecessary use of
4179     strcpy() was accidentally re-introduced in libpng16; this change replaces
4180     it with strncpy().
4181   Eliminated use of png_sizeof(); use sizeof() instead.
4182   Use a consistent style for (sizeof type) and (sizeof (array))
4183   Cleanup of png_set_filler().  This function does very different things on
4184     read and write.  In libpng 1.6 the two cases can be distinguished and
4185     considerable code cleanup, and extra error checking, is possible.  This
4186     makes calls on the write side that have no effect be ignored with a
4187     png_app_error(), which can be disabled in the app using
4188     png_set_benign_errors(), and removes the spurious use of usr_channels
4189     on the read side.
4190   Insist on autotools 1.12.1 for git builds because there are security issues
4191     with 1.12 and insisting on anything less would allow 1.12 to be used.
4192   Removed info_ptr->signature[8] from WRITE-only builds.
4193   Add some conditions for compiling png_fixed().  This is a small function
4194     but it requires "-lm" on some platforms.
4195   Cause pngtest --strict to fail on any warning from libpng (not just errors)
4196     and cause it not to fail at the comparison step if libpng lacks support
4197     for writing chunks that it reads from the input (currently only implemented
4198     for compressed text chunks).
4199   Make all three "make check" test programs work without READ or WRITE support.
4200     Now "make check" will succeed even if libpng is compiled with -DPNG_NO_READ
4201     or -DPNG_NO_WRITE.  The tests performed are reduced, but the basic reading
4202     and writing of a PNG file is always tested by one or more of the tests.
4203   Consistently use strlen(), memset(), memcpy(), and memcmp() instead of the
4204     png_strlen(), png_memset(), png_memcpy(), and png_memcmp() macros.
4205   Removed the png_sizeof(), png_strlen(), png_memset(), png_memcpy(), and
4206     png_memcmp() macros.
4207   Work around gcc 3.x and Microsoft Visual Studio 2010 complaints. Both object
4208     to the split initialization of num_chunks.
4209 
4210 Version 1.6.0beta28 [August 29, 2012]
4211   Unknown handling fixes and clean up. This adds more correct option
4212     control of the unknown handling, corrects the pre-existing bug where
4213     the per-chunk 'keep' setting is ignored and makes it possible to skip
4214     IDAT chunks in the sequential reader (broken in earlier 1.6 versions).
4215     There is a new test program, test-unknown.c, which is a work in progress
4216     (not currently part of the test suite).  Comments in the header files now
4217     explain how the unknown handling works.
4218   Allow fine grain control of unknown chunk APIs. This change allows
4219     png_set_keep_unknown_chunks() to be turned off if not required and causes
4220     both read and write to behave appropriately (on read this is only possible
4221     if the user callback is used to handle unknown chunks).  The change
4222     also removes the support for storing unknown chunks in the info_struct
4223     if the only unknown handling enabled is via the callback, allowing libpng
4224     to be configured with callback reading and none of the unnecessary code.
4225   Corrected fix for unknown handling in pngtest. This reinstates the
4226     libpng handling of unknown chunks other than vpAg and sTER (including
4227     unsafe-to-copy chunks which were dropped before) and eliminates the
4228     repositioning of vpAg and sTER in pngtest.png by changing pngtest.png
4229     (so the chunks are where libpng would put them).
4230   Added "tunknown" test and corrected a logic error in png_handle_unknown()
4231     when SAVE support is absent.  Moved the shell test scripts for
4232     contrib/libtests from the libpng top directory to contrib/libtests.
4233     png_handle_unknown() must always read or skip the chunk, if
4234     SAVE_UNKNOWN_CHUNKS is turned off *and* the application does not set
4235     a user callback an unknown chunk will not be read, leading to a read
4236     error, which was revealed by the "tunknown" test.
4237   Cleaned up and corrected ICC profile handling.
4238     contrib/libtests/makepng: corrected 'rgb' and 'gray' cases.  profile_error
4239     messages could be truncated; made a correct buffer size calculation and
4240     adjusted pngerror.c appropriately. png_icc_check_* checking improved;
4241     changed the functions to receive the correct color type of the PNG on read
4242     or write and check that it matches the color space of the profile (despite
4243     what the comments said before, there is danger in assuming the app will
4244     cope correctly with an RGB profile on a grayscale image and, since it
4245     violates the PNG spec, allowing it is certain to produce inconsistent
4246     app behavior and might even cause app crashes.) Check that profiles
4247     contain the tags needed to process the PNG (tags all required by the ICC
4248     spec). Removed unused PNG_STATIC from pngpriv.h.
4249 
4250 Version 1.6.0beta29 [September 4, 2012]
4251   Fixed the simplified API example programs to add the *colormap parameter
4252     to several of he API and improved the error message if the version field
4253     is not set.
4254   Added contrib/examples/* to the *.zip and *.7z distributions.
4255   Updated simplified API synopses and description of the png_image structure
4256     in the manual.
4257   Made makepng and pngtest produce identical PNGs, add "--relaxed" option
4258     to pngtest. The "--relaxed" option turns off the benign errors that are
4259     enabled by default in pre-RC builds. makepng can now write ICC profiles
4260     where the length has not been extended to a multiple of 4, and pngtest
4261     now intercepts all libpng errors, allowing the previously-introduced
4262     "--strict test" on no warnings to actually work.
4263   Improved ICC profile handling including cHRM chunk generation and fixed
4264     Cygwin+MSVC build errors. The ICC profile handling now includes more
4265     checking.  Several errors that caused rejection of the profile are now
4266     handled with a warning in such a way that the invalid profiles will be
4267     read by default in release (but not pre-RC) builds but will not be
4268     written by default.  The easy part of handling the cHRM chunk is written,
4269     where the ICC profile contains the required data.  The more difficult
4270     part plus guessing a gAMA value requires code to pass selected RGB values
4271     through the profile.
4272 
4273 Version 1.6.0beta30 [October 24, 2012]
4274   Changed ICC profile matrix/vector types to not depend on array type rules.
4275     By the ANSI-C standard the new types should be identical to the previous
4276     versions, and all known versions of gcc tested with the previous versions
4277     except for GCC-4.2.1 work with this version.  The change makes the ANSI-C
4278     rule that const applied to an array of elements applies instead to the
4279     elements in the array moot by explicitly applying const to the base
4280     elements of the png_icc_matrix and png_icc_vector types. The accidental
4281     (harmless) 'const' previously applied to the parameters of two of the
4282     functions have also been removed.
4283   Added a work around for GCC 4.2 optimization bug.
4284   Marked the broken (bad white point) original HP sRGB profiles correctly and
4285     correct comments.
4286   Added -DZ_SOLO to contrib/pngminim/*/makefile to work with zlib-1.2.7
4287   Use /MDd for vstudio debug builds. Also added pngunkown to the vstudio
4288     builds, fixed build errors and corrected a minor exit code error in
4289     pngvalid if the 'touch' file name is invalid.
4290   Add updated WARNING file to projects/vstudio from libpng 1.5/vstudio
4291   Fixed build when using #define PNG_NO_READ_GAMMA in png_do_compose() in
4292     pngrtran.c (Domani Hannes).
4293 
4294 Version 1.6.0beta31 [November 1, 2012]
4295   Undid the erroneous change to vstudio/pngvalid build in libpng-1.6.0beta30.
4296   Made pngvalid so that it will build outside the libpng source tree.
4297   Made builds -DPNG_NO_READ_GAMMA compile (the unit tests still fail).
4298   Made PNG_NO_READ_GAMMA switch off interfaces that depend on READ_GAMMA.
4299     Prior to 1.6.0 switching off READ_GAMMA did unpredictable things to the
4300     interfaces that use it (specifically, png_do_background in 1.4 would
4301     simply display composite for grayscale images but do composition
4302     with the incorrect arithmetic for color ones). In 1.6 the semantic
4303     of -DPNG_NO_READ_GAMMA is changed to simply disable any interface that
4304     depends on it; this obliges people who set it to consider whether they
4305     really want it off if they happen to use any of the interfaces in
4306     question (typically most users who disable it won't).
4307   Fixed GUIDs in projects/vstudio. Some were duplicated or missing,
4308     resulting in VS2010 having to update the files.
4309   Removed non-working ICC profile support code that was mostly added to
4310     libpng-1.6.0beta29 and beta30. There was too much code for too little
4311     gain; implementing full ICC color correction may be desirable but is left
4312     up to applications.
4313 
4314 Version 1.6.0beta32 [November 25, 2012]
4315   Fixed an intermittent SEGV in pngstest due to an uninitialized array element.
4316   Added the ability for contrib/libtests/makepng.c to make a PNG with just one
4317     color. This is useful for debugging pngstest color inaccuracy reports.
4318   Fixed error checking in the simplified write API (Olaf van der Spek)
4319   Made png_user_version_check() ok to use with libpng version 1.10.x and later.
4320 
4321 Version 1.6.0beta33 [December 15, 2012]
4322   Fixed typo in png.c (PNG_SET_CHUNK_MALLOC_MAX should be PNG_CHUNK_MALLOC_MAX)
4323     that causes the MALLOC_MAX limit not to work (John Bowler)
4324   Change png_warning() to png_app_error() in pngwrite.c and comment the
4325     fall-through condition.
4326   Change png_warning() to png_app_warning() in png_write_tRNS().
4327   Rearranged the ARM-NEON optimizations: Isolated the machine specific code
4328     to the hardware subdirectory and added comments to pngrutil.c so that
4329     implementors of other optimizations know what to do.
4330   Fixed cases of unquoted DESTDIR in Makefile.am
4331   Rebuilt Makefile.in, etc., with autoconf-2.69 and automake-1.12.5.
4332 
4333 Version 1.6.0beta34 [December 19, 2012]
4334   Cleaned up whitespace in the synopsis portion of the manpage "libpng.3"
4335   Disassembled the version number in scripts/options.awk (necessary for
4336     building on SunOs).
4337 
4338 Version 1.6.0beta35 [December 23, 2012]
4339   Made default Zlib compression settings be configurable. This adds #defines to
4340     pnglibconf.h to control the defaults.
4341   Fixed Windows build issues, enabled ARM compilation. Various warnings issued
4342     by earlier versions of GCC fixed for Cygwin and Min/GW (which both use old
4343     GCCs.) ARM support is enabled by default in zlib.props (unsupported by
4344     Microsoft) and ARM compilation is made possible by deleting the check for
4345     x86. The test programs cannot be run because they are not signed.
4346 
4347 Version 1.6.0beta36 [January 2, 2013]
4348   Discontinued distributing libpng-1.x.x.tar.bz2.
4349   Discontinued distributing libpng-1.7.0-1.6.0-diff.txt and similar.
4350   Rebuilt configure with autoconf-2.69 (inadvertently not done in beta33)
4351   Fixed 'make distcheck' on SUN OS - libpng.so was not being removed
4352 
4353 Version 1.6.0beta37 [January 10, 2013]
4354   Fixed conceivable but difficult to repro overflow. Also added two test
4355     programs to generate and test a PNG which should have the problem.
4356 
4357 Version 1.6.0beta39 [January 19, 2013]
4358   Again corrected attempt at overflow detection in png_set_unknown_chunks()
4359   (CVE-2013-7353).  Added overflow detection in png_set_sPLT() and
4360   png_set_text_2() (CVE-2013-7354).
4361 
4362 Version 1.6.0beta40 [January 20, 2013]
4363   Use consistent handling of overflows in text, sPLT and unknown png_set_* APIs
4364 
4365 Version 1.6.0rc01 [January 26, 2013]
4366   No changes.
4367 
4368 Version 1.6.0rc02 [February 4, 2013]
4369   Added png_get_palette_max() function.
4370 
4371 Version 1.6.0rc03 [February 5, 2013]
4372   Fixed the png_get_palette_max API.
4373 
4374 Version 1.6.0rc04 [February 7, 2013]
4375   Turn serial tests back on (recently turned off by autotools upgrade).
4376 
4377 Version 1.6.0rc05 [February 8, 2013]
4378   Update manual about png_get_palette_max().
4379 
4380 Version 1.6.0rc06 [February 9, 2013]
4381   Fixed missing dependency in --prefix builds The intermediate
4382     internal 'prefix.h' file can only be generated correctly after
4383     pnglibconf.h, however the dependency was not in Makefile.am.  The
4384     symptoms are unpredictable depending on the order make chooses to
4385     build pngprefix.h and pnglibconf.h, often the error goes unnoticed
4386     because there is a system pnglibconf.h to use instead.
4387 
4388 Version 1.6.0rc07 [February 10, 2013]
4389   Enclosed the new png_get_palette_max in #ifdef PNG_GET_PALETTE_MAX_SUPPORTED
4390     block, and revised pnglibconf.h and pnglibconf.h.prebuilt accordingly.
4391 
4392 Version 1.6.0rc08 [February 10, 2013]
4393   Fix typo in png.h #ifdef
4394 
4395 Version 1.6.0 [February 14, 2013]
4396   No changes.
4397 
4398 Version 1.6.1beta01 [February 16, 2013]
4399   Made symbol prefixing work with the ARM neon optimizations. Also allow
4400     pngpriv.h to be included for preprocessor definitions only, so it can
4401     be used in non-C/C++ files. Back ported from libpng 1.7.
4402   Made sRGB check numbers consistent.
4403   Ported libpng 1.5 options.awk/dfn file handling to 1.6, fixed one bug.
4404   Removed cc -E workround, corrected png_get_palette_max API Tested on
4405     SUN OS cc 5.9, which demonstrates the tokenization problem previously
4406     avoided by using /lib/cpp.  Since all .dfn output is now protected in
4407     double quotes unless it is to be macro substituted the fix should
4408     work everywhere.
4409   Enabled parallel tests - back ported from libpng-1.7.
4410   scripts/pnglibconf.dfa formatting improvements back ported from libpng17.
4411   Fixed a race condition in the creation of the build 'scripts' directory
4412     while building with a parallel make.
4413   Use approved/supported Android method to check for NEON, use Linux/POSIX
4414     1003.1 API to check /proc/self/auxv avoiding buffer allocation and other
4415     library calls (ported from libpng15).
4416 
4417 Version 1.6.1beta02 [February 19, 2013]
4418   Use parentheses more consistently in "#if defined(MACRO)" tests.
4419   Folded long lines.
4420   Reenabled code to allow zero length PLTE chunks for MNG.
4421 
4422 Version 1.6.1beta03 [February 22, 2013]
4423   Fixed ALIGNED_MEMORY support.
4424   Added a new configure option:
4425     --enable-arm-neon=always will stop the run-time checks. New checks
4426     within arm/arm_init.c will cause the code not to be compiled unless
4427     __ARM_NEON__ is set. This should make it fail safe (if someone asks
4428     for it on then the build will fail if it can't be done.)
4429   Updated the INSTALL document.
4430 
4431 Version 1.6.1beta04 [February 27, 2013]
4432   Revised INSTALL to recommend using CPPFLAGS instead of INCLUDES.
4433   Revised scripts/makefile.freebsd to respect ZLIBLIB and ZLIBINC.
4434   Revised scripts/dfn.awk to work with the buggy MSYS awk that has trouble
4435     with CRLF line endings.
4436 
4437 Version 1.6.1beta05 [March 1, 2013]
4438   Avoid a possible memory leak in contrib/gregbook/readpng.c
4439 
4440 Version 1.6.1beta06 [March 4, 2013]
4441   Better documentation of unknown handling API interactions.
4442   Corrected Android builds and corrected libpng.vers with symbol
4443     prefixing.  It also makes those tests compile and link on Android.
4444   Added an API png_set_option() to set optimization options externally,
4445     providing an alternative and general solution for the non-portable
4446     run-time tests used by the ARM Neon code, using the PNG_ARM_NEON option.
4447   The order of settings vs options in pnglibconf.h is reversed to allow
4448     settings to depend on options and options can now set (or override) the
4449     defaults for settings.
4450 
4451 Version 1.6.1beta07 [March 7, 2013]
4452   Corrected simplified API default gamma for color-mapped output, added
4453     a flag to change default. In 1.6.0 when the simplified API was used
4454     to produce color-mapped output from an input image with no gamma
4455     information the gamma assumed for the input could be different from
4456     that assumed for non-color-mapped output.  In particular 16-bit depth
4457     input files were assumed to be sRGB encoded, whereas in the 'direct'
4458     case they were assumed to have linear data.  This was an error.  The
4459     fix makes the simplified API treat all input files the same way and
4460     adds a new flag to the png_image::flags member to allow the
4461     application/user to specify that 16-bit files contain sRGB data
4462     rather than the default linear.
4463   Fixed bugs in the pngpixel and makepng test programs.
4464 
4465 Version 1.6.1beta08 [March 7, 2013]
4466   Fixed CMakelists.txt to allow building a single variant of the library
4467     (Claudio Bley):
4468   Introduced a PNG_LIB_TARGETS variable that lists all activated library
4469     targets.  It is an error if this variable ends up empty, ie. you have
4470     to build at least one library variant.
4471   Made the *_COPY targets only depend on library targets actually being build.
4472   Use PNG_LIB_TARGETS to unify a code path.
4473   Changed the CREATE_SYMLINK macro to expect the full path to a file as the
4474     first argument. When symlinking the filename component of that path is
4475     determined and used as the link target.
4476   Use copy_if_different in the CREATE_SYMLINK macro.
4477 
4478 Version 1.6.1beta09 [March 13, 2013]
4479   Eliminated two warnings from the Intel C compiler. The warnings are
4480     technically valid, although a reasonable treatment of division would
4481     show it to be incorrect.
4482 
4483 Version 1.6.1rc01 [March 21, 2013]
4484   No changes.
4485 
4486 Version 1.6.1 [March 28, 2013]
4487   No changes.
4488 
4489 Version 1.6.2beta01 [April 14, 2013]
4490   Updated documentation of 1.5.x to 1.6.x changes in iCCP chunk handling.
4491   Fixed incorrect warning of excess deflate data. End condition - the
4492     warning would be produced if the end of the deflate stream wasn't read
4493     in the last row.  The warning is harmless.
4494   Corrected the test on user transform changes on read. It was in the
4495     png_set of the transform function, but that doesn't matter unless the
4496     transform function changes the rowbuf size, and that is only valid if
4497     transform_info is called.
4498   Corrected a misplaced closing bracket in contrib/libtests/pngvalid.c
4499     (Flavio Medeiros).
4500   Corrected length written to uncompressed iTXt chunks (Samuli Suominen).
4501     Bug was introduced in libpng-1.6.0.
4502 
4503 Version 1.6.2rc01 [April 18, 2013]
4504   Added contrib/tools/fixitxt.c, to repair the erroneous iTXt chunk length
4505     written by libpng-1.6.0 and 1.6.1.
4506   Disallow storing sRGB information when the sRGB is not supported.
4507 
4508 Version 1.6.2rc02 [April 18, 2013]
4509   Merge pngtest.c with libpng-1.7.0
4510 
4511 Version 1.6.2rc03 [April 22, 2013]
4512   Trivial spelling cleanup.
4513 
4514 Version 1.6.2rc04 and 1.6.2rc05 [omitted]
4515 
4516 Version 1.6.2rc06 [April 24, 2013]
4517   Reverted to version 1.6.2rc03.  Recent changes to arm/neon support
4518     have been ported to libpng-1.7.0beta09 and will reappear in version
4519     1.6.3beta01.
4520 
4521 Version 1.6.2 [April 25, 2013]
4522   No changes.
4523 
4524 Version 1.6.3beta01 [April 25, 2013]
4525   Revised stack marking in arm/filter_neon.S and configure.ac.
4526   Ensure that NEON filter stuff is completely disabled when switched 'off'.
4527     Previously the ARM NEON specific files were still built if the option
4528     was switched 'off' as opposed to being explicitly disabled.
4529 
4530 Version 1.6.3beta02 [April 26, 2013]
4531   Test for 'arm*' not just 'arm' in the host_cpu configure variable.
4532   Rebuilt the configure scripts.
4533 
4534 Version 1.6.3beta03 [April 30, 2013]
4535   Expanded manual paragraph about writing private chunks, particularly
4536     the need to call png_set_keep_unknown_chunks() when writing them.
4537   Avoid dereferencing NULL pointer possibly returned from
4538     png_create_write_struct() (Andrew Church).
4539 
4540 Version 1.6.3beta05 [May 9, 2013]
4541   Calculate our own zlib windowBits when decoding rather than trusting the
4542     CMF bytes in the PNG datastream.
4543   Added an option to force maximum window size for inflating, which was
4544     the behavior of libpng15 and earlier, via a new PNG_MAXIMUM_INFLATE_WINDOW
4545     option for png_set_options().
4546   Added png-fix-itxt and png-fix-too-far-back to the built programs and
4547     removed warnings from the source code and timepng that are revealed as
4548     a result.
4549   Detect wrong libpng versions linked to png-fix-too-far-back, which currently
4550     only works with libpng versions that can be made to reliably fail when
4551     the deflate data contains an out-of-window reference.  This means only
4552     1.6 and later.
4553   Fixed gnu issues: g++ needs a static_cast, gcc 4.4.7 has a broken warning
4554     message which it is easier to work round than ignore.
4555   Updated contrib/pngminus/pnm2png.c (Paul Stewart):
4556     Check for EOF
4557     Ignore "#" delimited comments in input file to pnm2png.c.
4558     Fixed whitespace handling
4559     Added a call to png_set_packing()
4560     Initialize dimension values so if sscanf fails at least we have known
4561       invalid values.
4562   Attempt to detect configuration issues with png-fix-too-far-back, which
4563     requires both the correct libpng and the correct zlib to function
4564     correctly.
4565   Check ZLIB_VERNUM for mismatches, enclose #error in quotes
4566   Added information in the documentation about problems with and fixes for
4567     the bad CRC and bad iTXt chunk situations.
4568 
4569 Version 1.6.3beta06 [May 12, 2013]
4570   Allow contrib/pngminus/pnm2png.c to compile without WRITE_INVERT and
4571     WRITE_PACK supported (writes error message that it can't read P1 or
4572     P4 PBM files).
4573   Improved png-fix-too-far-back usage message, added --suffix option.
4574   Revised contrib/pngminim/*/makefile to generate pnglibconf.h with the
4575     right zlib header files.
4576   Separated CPPFLAGS and CFLAGS in contrib/pngminim/*/makefile
4577 
4578 Version 1.6.3beta07 [June 8, 2013]
4579   Removed a redundant test in png_set_IHDR().
4580   Added set(CMAKE_CONFIGURATION_TYPES ...) to CMakeLists.txt (Andrew Hundt)
4581   Deleted set(CMAKE_BUILD_TYPE) block from CMakeLists.txt
4582   Enclose the prototypes for the simplified write API in
4583     #ifdef PNG_STDIO_SUPPORTED/#endif
4584   Make ARM NEON support work at compile time (not just configure time).
4585     This moves the test on __ARM_NEON__ into pngconf.h to avoid issues when
4586     using a compiler that compiles for multiple architectures at one time.
4587   Removed PNG_FILTER_OPTIMIZATIONS and PNG_ARM_NEON_SUPPORTED from
4588     pnglibconf.h, allowing more of the decisions to be made internally
4589     (pngpriv.h) during the compile.  Without this, symbol prefixing is broken
4590     under certain circumstances on ARM platforms.  Now only the API parts of
4591     the optimizations ('check' vs 'api') are exposed in the public header files
4592     except that the new setting PNG_ARM_NEON_OPT documents how libpng makes the
4593     decision about whether or not to use the optimizations.
4594   Protect symbol prefixing against CC/CPPFLAGS/CFLAGS usage.
4595     Previous iOS/Xcode fixes for the ARM NEON optimizations moved the test
4596     on __ARM_NEON__ from configure time to compile time.  This breaks symbol
4597     prefixing because the definition of the special png_init_filter_functions
4598     call was hidden at configure time if the relevant compiler arguments are
4599     passed in CFLAGS as opposed to CC.  This change attempts to avoid all
4600     the confusion that would result by declaring the init function even when
4601     it is not used, so that it will always get prefixed.
4602 
4603 Version 1.6.3beta08 [June 18, 2013]
4604   Revised libpng.3 so that "doclifter" can process it.
4605 
4606 Version 1.6.3beta09 [June 27, 2013]
4607   Revised example.c to illustrate use of PNG_DEFAULT_sRGB and PNG_GAMMA_MAC_18
4608     as parameters for png_set_gamma().  These have been available since
4609     libpng-1.5.4.
4610   Renamed contrib/tools/png-fix-too-far-back.c to pngfix.c and revised it
4611     to check all compressed chunks known to libpng.
4612 
4613 Version 1.6.3beta10 [July 5, 2013]
4614   Updated documentation to show default behavior of benign errors correctly.
4615   Only compile ARM code when PNG_READ_SUPPORTED is defined.
4616   Fixed undefined behavior in contrib/tools/pngfix.c and added new strip
4617     option. pngfix relied on undefined behavior and even a simple change from
4618     gcc to g++ caused it to fail.  The new strip option 'unsafe' has been
4619     implemented and is the default if --max is given.  Option names have
4620     been clarified, with --strip=transform now stripping the bKGD chunk,
4621     which was stripped previously with --strip=unused.
4622   Added all documented chunk types to pngpriv.h
4623   Unified pngfix.c source with libpng17.
4624 
4625 Version 1.6.3rc01 [July 11, 2013]
4626   No changes.
4627 
4628 Version 1.6.3 [July 18, 2013]
4629   Revised manual about changes in iTXt chunk handling made in libpng-1.6.0.
4630   Added "/* SAFE */" comments in pngrutil.c and pngrtran.c where warnings
4631     may be erroneously issued by code-checking applications.
4632 
4633 Version 1.6.4beta01 [August 21, 2013]
4634   Added information about png_set_options() to the manual.
4635   Delay calling png_init_filter_functions() until a row with nonzero filter
4636     is found.
4637 
4638 Version 1.6.4beta02 [August 30, 2013]
4639   Fixed inconsistent conditional compilation of png_chunk_unknown_handling()
4640     prototype, definition, and usage.  Made it depend on
4641     PNG_HANDLE_AS_UNKNOWN_SUPPORTED everywhere.
4642 
4643 Version 1.6.4rc01 [September 5, 2013]
4644   No changes.
4645 
4646 Version 1.6.4 [September 12, 2013]
4647   No changes.
4648 
4649 Version 1.6.5 [September 14, 2013]
4650   Removed two stray lines of code from arm/arm_init.c.
4651 
4652 Version 1.6.6 [September 16, 2013]
4653   Removed two stray lines of code from arm/arm_init.c, again.
4654 
4655 Version 1.6.7beta01 [September 30, 2013]
4656   Revised unknown chunk code to correct several bugs in the NO_SAVE_/NO_WRITE
4657     combination
4658   Allow HANDLE_AS_UNKNOWN to work when other options are configured off. Also
4659     fixed the pngminim makefiles to work when $(MAKEFLAGS) contains stuff
4660     which terminates the make options (as by default in recent versions of
4661     Gentoo).
4662   Avoid up-cast warnings in pngvalid.c. On ARM the alignment requirements of
4663     png_modifier are greater than that of png_store and as a consequence
4664     compilation of pngvalid.c results in a warning about increased alignment
4665     requirements because of the bare cast to (png_modifier*). The code is safe,
4666     because the pointer is known to point to a stack allocated png_modifier,
4667     but this change avoids the warning.
4668   Fixed default behavior of ARM_NEON_API. If the ARM NEON API option was
4669     compiled without the CHECK option it defaulted to on, not off.
4670   Check user callback behavior in pngunknown.c. Previous versions compiled
4671     if SAVE_UNKNOWN was not available but did nothing since the callback
4672     was never implemented.
4673   Merged pngunknown.c with 1.7 version and back ported 1.7 improvements/fixes
4674 
4675 Version 1.6.7beta02 [October 12, 2013]
4676   Made changes for compatibility with automake 1.14:
4677     1) Added the 'compile' program to the list of programs that must be cleaned
4678        in autogen.sh
4679     2) Added 'subdir-objects' which causes .c files in sub-directories to be
4680        compiled such that the corresponding .o files are also in the
4681        sub-directory.  This is because automake 1.14 warns that the
4682        current behavior of compiling to the top level directory may be removed
4683        in the future.
4684     3) Updated dependencies on pnglibconf.h to match the new .o locations and
4685        added all the files in contrib/libtests and contrib/tools that depend
4686        on pnglibconf.h
4687     4) Added 'BUILD_SOURCES = pnglibconf.h'; this is the automake recommended
4688        way of handling the dependencies of sources that are machine generated;
4689        unfortunately it only works if the user does 'make all' or 'make check',
4690        so the dependencies (3) are still required.
4691   Cleaned up (char*) casts of zlib messages. The latest version of the Intel C
4692     compiler complains about casting a string literal as (char*), so copied the
4693     treatment of z_const from the library code into pngfix.c
4694   Simplified error message code in pngunknown. The simplification has the
4695     useful side effect of avoiding a bogus warning generated by the latest
4696     version of the Intel C compiler (it objects to
4697     condition ? string-literal : string-literal).
4698   Make autogen.sh work with automake 1.13 as well as 1.14. Do this by always
4699     removing the 1.14 'compile' script but never checking for it.
4700 
4701 Version 1.6.7beta03 [October 19, 2013]
4702   Added ARMv8 support (James Yu <james.yu at linaro.org>).  Added file
4703     arm/filter_neon_intrinsics.c; enable with -mfpu=neon.
4704   Revised pngvalid to generate size images with as many filters as it can
4705     manage, limited by the number of rows.
4706   Cleaned up ARM NEON compilation handling. The tests are now in pngpriv.h
4707     and detect the broken GCC compilers.
4708 
4709 Version 1.6.7beta04 [October 26, 2013]
4710   Allow clang derived from older GCC versions to use ARM intrinsics. This
4711     causes all clang builds that use -mfpu=neon to use the intrinsics code,
4712     not the assembler code.  This has only been tested on iOS 7. It may be
4713     necessary to exclude some earlier clang versions but this seems unlikely.
4714   Changed NEON implementation selection mechanism. This allows assembler
4715     or intrinsics to be turned on at compile time during the build by defining
4716     PNG_ARM_NEON_IMPLEMENTATION to the correct value (2 or 1).  This macro
4717     is undefined by default and the build type is selected in pngpriv.h.
4718 
4719 Version 1.6.7rc01 [November 2, 2013]
4720   No changes.
4721 
4722 Version 1.6.7rc02 [November 7, 2013]
4723   Fixed #include in filter_neon_intrinsics.c and ctype macros. The ctype char
4724     checking macros take an unsigned char argument, not a signed char.
4725 
4726 Version 1.6.7 [November 14, 2013]
4727   No changes.
4728 
4729 Version 1.6.8beta01 [November 24, 2013]
4730   Moved prototype for png_handle_unknown() in pngpriv.h outside of
4731     the #ifdef PNG_SET_UNKNOWN_CHUNKS_SUPPORTED/#endif block.
4732   Added "-Wall" to CFLAGS in contrib/pngminim/*/makefile
4733   Conditionally compile some unused functions reported by -Wall in
4734     pngminim.
4735   Fixed 'minimal' builds. Various obviously useful minimal configurations
4736     don't build because of missing contrib/libtests test programs and
4737     overly complex dependencies in scripts/pnglibconf.dfa. This change
4738     adds contrib/conftest/*.dfa files that can be used in automatic build
4739     scripts to ensure that these configurations continue to build.
4740   Enabled WRITE_INVERT and WRITE_PACK in contrib/pngminim/encoder.
4741   Fixed pngvalid 'fail' function declaration on the Intel C Compiler.
4742     This reverts to the previous 'static' implementation and works round
4743     the 'unused static function' warning by using PNG_UNUSED().
4744 
4745 Version 1.6.8beta02 [November 30, 2013]
4746   Removed or marked PNG_UNUSED some harmless "dead assignments" reported
4747     by clang scan-build.
4748   Changed tabs to 3 spaces in png_debug macros and changed '"%s"m'
4749     to '"%s" m' to improve portability among compilers.
4750   Changed png_free_default() to free() in pngtest.c
4751 
4752 Version 1.6.8rc01 [December 12, 2013]
4753   Tidied up pngfix inits and fixed pngtest no-write builds.
4754 
4755 Version 1.6.8rc02 [December 14, 2013]
4756   Handle zero-length PLTE chunk or NULL palette with png_error()
4757     instead of png_chunk_report(), which by default issues a warning
4758     rather than an error, leading to later reading from a NULL pointer
4759     (png_ptr->palette) in png_do_expand_palette(). This is CVE-2013-6954
4760     and VU#650142.  Libpng-1.6.1 through 1.6.7 are vulnerable.
4761     Libpng-1.6.0 and earlier do not have this bug.
4762 
4763 Version 1.6.8 [December 19, 2013]
4764   No changes.
4765 
4766 Version 1.6.9beta01 [December 26, 2013]
4767   Bookkeeping: Moved functions around (no changes). Moved transform
4768     function definitions before the place where they are called so that
4769     they can be made static. Move the intrapixel functions and the
4770     grayscale palette builder out of the png?tran.c files. The latter
4771     isn't a transform function and is no longer used internally, and the
4772     former MNG specific functions are better placed in pngread/pngwrite.c
4773   Made transform implementation functions static. This makes the internal
4774     functions called by png_do_{read|write}_transformations static. On an
4775     x86-64 DLL build (Gentoo Linux) this reduces the size of the text
4776     segment of the DLL by 1208 bytes, about 0.6%. It also simplifies
4777     maintenance by removing the declarations from pngpriv.h and allowing
4778     easier changes to the internal interfaces.
4779   Rebuilt configure scripts with automake-1.14.1 and autoconf-2.69
4780     in the tar distributions.
4781 
4782 Version 1.6.9beta02 [January 1, 2014]
4783   Added checks for libpng 1.5 to pngvalid.c.  This supports the use of
4784     this version of pngvalid in libpng 1.5
4785   Merged with pngvalid.c from libpng-1.7 changes to create a single
4786     pngvalid.c
4787   Removed #error macro from contrib/tools/pngfix.c (Thomas Klausner).
4788   Merged pngrio.c, pngtrans.c, pngwio.c, and pngerror.c with libpng-1.7.0
4789   Merged libpng-1.7.0 changes to make no-interlace configurations work
4790     with test programs.
4791   Revised pngvalid.c to support libpng 1.5, which does not support the
4792     PNG_MAXIMUM_INFLATE_WINDOW option, so #define it out when appropriate in
4793     pngvalid.c
4794   Allow unversioned links created on install to be disabled in configure.
4795     In configure builds 'make install' changes/adds links like png.h
4796     and libpng.a to point to the newly installed, versioned, files (e.g.
4797     libpng17/png.h and libpng17.a). Three new configure options and some
4798     rearrangement of Makefile.am allow creation of these links to be disabled.
4799 
4800 Version 1.6.9beta03 [January 10, 2014]
4801   Removed potentially misleading warning from png_check_IHDR().
4802 
4803 Version 1.6.9beta04 [January 20, 2014]
4804   Updated scripts/makefile.* to use CPPFLAGS (Cosmin).
4805   Added clang attribute support (Cosmin).
4806 
4807 Version 1.6.9rc01 [January 28, 2014]
4808   No changes.
4809 
4810 Version 1.6.9rc02 [January 30, 2014]
4811   Quiet an uninitialized memory warning from VC2013 in png_get_png().
4812 
4813 Version 1.6.9 [February 6, 2014]
4814 
4815 Version 1.6.10beta01 [February 9, 2014]
4816   Backported changes from libpng-1.7.0beta30 and beta31:
4817   Fixed a large number of instances where PNGCBAPI was omitted from
4818     function definitions.
4819   Added pngimage test program for png_read_png() and png_write_png()
4820     with two new test scripts.
4821   Removed dependence on !PNG_READ_EXPAND_SUPPORTED for calling
4822     png_set_packing() in png_read_png().
4823   Fixed combination of ~alpha with shift. On read invert alpha, processing
4824     occurred after shift processing, which causes the final values to be
4825     outside the range that should be produced by the shift. Reversing the
4826     order on read makes the two transforms work together correctly and mirrors
4827     the order used on write.
4828   Do not read invalid sBIT chunks. Previously libpng only checked sBIT
4829     values on write, so a malicious PNG writer could therefore cause
4830     the read code to return an invalid sBIT chunk, which might lead to
4831     application errors or crashes.  Such chunks are now skipped (with
4832     chunk_benign_error).
4833   Make png_read_png() and png_write_png() prototypes in png.h depend
4834     upon PNG_READ_SUPPORTED and PNG_WRITE_SUPPORTED.
4835   Support builds with unsupported PNG_TRANSFORM_* values.  All of the
4836     PNG_TRANSFORM_* values are always defined in png.h and, because they
4837     are used for both read and write in some cases, it is not reliable
4838     to #if out ones that are totally unsupported. This change adds error
4839     detection in png_read_image() and png_write_image() to do a
4840     png_app_error() if the app requests something that cannot be done
4841     and it adds corresponding code to pngimage.c to handle such options
4842     by not attempting to test them.
4843 
4844 Version 1.6.10beta02 [February 23, 2014]
4845   Moved redefines of png_error(), png_warning(), png_chunk_error(),
4846     and png_chunk_warning() from pngpriv.h to png.h to make them visible
4847     to libpng-calling applications.
4848   Moved OS dependent code from arm/arm_init.c, to allow the included
4849     implementation of the ARM NEON discovery function to be set at
4850     build-time and provide sample implementations from the current code in the
4851     contrib/arm-neon subdirectory. The __linux__ code has also been changed to
4852     compile and link on Android by using /proc/cpuinfo, and the old linux code
4853     is in contrib/arm-neon/linux-auxv.c.  The new code avoids POSIX and Linux
4854     dependencies apart from opening /proc/cpuinfo and is C90 compliant.
4855   Check for info_ptr == NULL early in png_read_end() so we don't need to
4856     run all the png_handle_*() and depend on them to return if info_ptr == NULL.
4857     This improves the performance of png_read_end(png_ptr, NULL) and makes
4858     it more robust against future programming errors.
4859   Check for __has_extension before using it in pngconf.h, to
4860     support older Clang versions (Jeremy Sequoia).
4861   Treat CRC error handling with png_set_crc_action(), instead of with
4862     png_set_benign_errors(), which has been the case since libpng-1.6.0beta18.
4863   Use a user warning handler in contrib/gregbook/readpng2.c instead of default,
4864     so warnings will be put on stderr even if libpng has CONSOLE_IO disabled.
4865   Added png_ptr->process_mode = PNG_READ_IDAT_MODE in png_push_read_chunk
4866     after recognizing the IDAT chunk, which avoids an infinite loop while
4867     reading a datastream whose first IDAT chunk is of zero-length.
4868     This fixes CERT VU#684412 and CVE-2014-0333.
4869   Don't recognize known sRGB profiles as sRGB if they have been hacked,
4870     but don't reject them and don't issue a copyright violation warning.
4871 
4872 Version 1.6.10beta03 [February 25, 2014]
4873   Moved some documentation from png.h to libpng.3 and libpng-manual.txt
4874   Minor editing of contrib/arm-neon/README and contrib/examples/*.c
4875 
4876 Version 1.6.10rc01 [February 27, 2014]
4877   Fixed typos in the manual and in scripts/pnglibconf.dfa (CFLAGS -> CPPFLAGS
4878     and PNG_USR_CONFIG -> PNG_USER_CONFIG).
4879 
4880 Version 1.6.10rc02 [February 28, 2014]
4881   Removed unreachable return statement after png_chunk_error()
4882     in pngrutil.c
4883 
4884 Version 1.6.10rc03 [March 4, 2014]
4885   Un-deprecated png_data_freer().
4886 
4887 Version 1.6.10 [March 6, 2014]
4888   No changes.
4889 
4890 Version 1.6.11beta01 [March 17, 2014]
4891   Use "if (value != 0)" instead of "if (value)" consistently.
4892   Changed ZlibSrcDir from 1.2.5 to 1.2.8 in projects/vstudio.
4893   Moved configuration information from the manual to the INSTALL file.
4894 
4895 Version 1.6.11beta02 [April 6, 2014]
4896   Removed #if/#else/#endif from inside two pow() calls in pngvalid.c because
4897     they were handled improperly by Portland Group's PGI-14.1 - PGI-14.3
4898     when using its "__builtin_pow()" function.
4899   Silence 'unused parameter' build warnings (Cosmin Truta).
4900   $(CP) is now used alongside $(RM_F).  Also, use 'copy' instead of 'cp'
4901     where applicable, and applied other minor makefile changes (Cosmin).
4902   Don't warn about invalid dimensions exceeding user limits (Cosmin).
4903   Allow an easy replacement of the default pre-built configuration
4904     header with a custom header, via the make PNGLIBCONF_H_PREBUILT
4905     macro (Cosmin).
4906 
4907 Version 1.6.11beta03 [April 6, 2014]
4908   Fixed a typo in pngrutil.c, introduced in libpng-1.5.6, that interferes
4909     with "blocky" expansion of sub-8-bit interlaced PNG files (Eric Huss).
4910   Optionally use  __builtin_bswap16() in png_do_swap().
4911 
4912 Version 1.6.11beta04 [April 19, 2014]
4913   Made progressive reading of interlaced images consistent with the
4914     behavior of the sequential reader and consistent with the manual, by
4915     moving some code out of the PNG_READ_INTERLACING_SUPPORTED blocks. The
4916     row_callback now receives the proper pass number and unexpanded rows, when
4917     png_combine_row() isn't built or used, and png_set_interlace_handling()
4918     is not called.
4919   Allow PNG_sRGB_PROFILE_CHECKING = (-1) to mean no sRGB profile checking.
4920 
4921 Version 1.6.11beta05 [April 26, 2014]
4922   Do not reject ICC V2 profiles that lack padding (Kai-Uwe Behrmann).
4923   Relocated closing bracket of the sRGB profile test loop to avoid getting
4924     "Not recognizing known sRGB profile that has been edited" warning for
4925     ICC V2 profiles that lack the MD5 signature in the profile header.
4926 
4927 Version 1.6.11beta06 [May 19, 2014]
4928   Added PNG_SKIP_sRGB_CHECK_PROFILE choice for png_set_option().
4929 
4930 Version 1.6.11rc01 [May 27, 2014]
4931   No changes.
4932 
4933 Version 1.6.11rc02 [June 3, 2014]
4934   Test ZLIB_VERNUM instead of PNG_ZLIB_VERNUM in contrib/tools/pngfix.c
4935 
4936 Version 1.6.11 [June 5, 2014]
4937   No changes.
4938 
4939 Version 1.6.12rc01 [June 6, 2014]
4940   Relocated new code from 1.6.11beta06 in png.c to a point after the
4941     declarations (Max Stepin).
4942 
4943 Version 1.6.12rc02 [June 7, 2014]
4944   Changed file permissions of contrib/tools/intgamma.sh,
4945     test-driver, and compile from 0644 to 0755 (Cosmin).
4946 
4947 Version 1.6.12rc03 [June 8, 2014]
4948   Ensure "__has_attribute()" macro exists before trying to use it with
4949     old clang compilers (MacPorts Ticket #43939).
4950 
4951 Version 1.6.12 [June 12, 2014]
4952   No changes.
4953 
4954 Version 1.6.13beta01 [July 4, 2014]
4955   Quieted -Wsign-compare and -Wclobber compiler warnings in
4956     contrib/pngminus/*.c
4957   Added "(void) png_ptr;" where needed in contrib/gregbook to quiet
4958     compiler complaints about unused pointers.
4959   Split a long output string in contrib/gregbook/rpng2-x.c.
4960   Added "PNG_SET_OPTION" requirement for sRGB chunk support to pnglibconf.dfa,
4961     Needed for write-only support (John Bowler).
4962   Changed "if defined(__ARM_NEON__)" to
4963     "if (defined(__ARM_NEON__) || defined(__ARM_NEON))" (James Wu).
4964   Fixed clang no-warning builds: png_digit was defined but never used.
4965     
4966 Version 1.6.13beta02 [July 21, 2014]
4967   Fixed an incorrect separator ("/" should be "\") in scripts/makefile.vcwin32
4968     (bug report from Wolfgang S. Kechel).  Bug was introduced in libpng-1.6.11.
4969     Also fixed makefile.bc32, makefile.bor, makefile.msc, makefile.intel, and
4970     makefile.tc3 similarly.
4971 
4972 Version 1.6.13beta03 [August 3, 2014]
4973   Removed scripts/makefile.elf. It has not worked since libpng-1.5.0beta14
4974     due to elimination of the PNG_FUNCTION_EXPORT and PNG_DATA_EXPORT
4975     definitions from pngconf.h.
4976   Ensure that CMakeLists.txt makes the target "lib" directory before making
4977     symbolic link into it (SourceForge bug report #226 by Rolf Timmermans).
4978 
4979 Version 1.6.13beta04 [August 8, 2014]
4980   Added opinion that the ECCN (Export Control Classification Number) for
4981     libpng is EAR99 to the README file.
4982   Eliminated use of "$<" in makefile explicit rules, when copying
4983     $PNGLIBCONF_H_PREBUILT.  This does not work on some versions of make;
4984     bug introduced in libpng version 1.6.11.
4985 
4986 Version 1.6.13rc01 [August 14, 2014]
4987   Made "ccopts" agree with "CFLAGS" in scripts/makefile.hp* and makefile.*sunu
4988 
4989 Version 1.6.13 [August 21, 2014]
4990   No changes.
4991 
4992 Version 1.6.14beta01 [September 14, 2014]
4993   Guard usage of png_ptr->options with #ifdef PNG_SET_OPTION_SUPPORTED.
4994   Do not build contrib/tools/pngfix.c when PNG_SETJMP_NOT_SUPPORTED,
4995     to allow "make" to complete without setjmp support (bug report by
4996     Claudio Fontana)
4997   Add "#include <setjmp.h>" to contrib/tools/pngfix.c (John Bowler)
4998 
4999 Version 1.6.14beta02 [September 18, 2014]
5000   Use nanosleep() instead of usleep() in contrib/gregbook/rpng2-x.c
5001     because usleep() is deprecated.
5002   Define usleep() in contrib/gregbook/rpng2-x.c if not already defined
5003     in unistd.h and nanosleep() is not available; fixes error introduced
5004     in libpng-1.6.13.
5005   Disable floating point exception handling in pngvalid.c when
5006     PNG_FLOATING_ARITHMETIC is not supported (bug report by "zootus
5007     at users.sourceforge.net").
5008 
5009 Version 1.6.14beta03 [September 19, 2014]
5010   Define FE_DIVBYZERO, FE_INVALID, and FE_OVERFLOW in pngvalid.c if not
5011     already defined.  Revert floating point exception handling in pngvalid.c
5012     to version 1.6.14beta01 behavior.
5013 
5014 Version 1.6.14beta04 [September 27, 2014]
5015   Fixed incorrect handling of the iTXt compression flag in pngrutil.c
5016     (bug report by Shunsaku Hirata).  Bug was introduced in libpng-1.6.0.
5017 
5018 Version 1.6.14beta05 [October 1, 2014]
5019   Added "option READ_iCCP enables READ_COMPRESSED_TEXT" to pnglibconf.dfa
5020 
5021 Version 1.6.14beta06 [October 5, 2014]
5022   Removed unused "text_len" parameter from private function png_write_zTXt().
5023   Conditionally compile some code in png_deflate_claim(), when
5024     PNG_WARNINGS_SUPPORTED and PNG_ERROR_TEXT_SUPPORTED are disabled.
5025   Replaced repeated code in pngpread.c with PNG_PUSH_SAVE_BUFFER_IF_FULL.
5026   Added "chunk iTXt enables TEXT" and "chunk zTXt enables TEXT"
5027     to pnglibconf.dfa.
5028   Removed "option READ_COMPRESSED_TEXT enables READ_TEXT" from pnglibconf.dfa,
5029     to make it possible to configure a libpng that supports iCCP but not TEXT.
5030 
5031 Version 1.6.14beta07 [October 7, 2014]
5032   Removed "option WRITE_COMPRESSED_TEXT enables WRITE_TEXT" from pnglibconf.dfa
5033   Only mark text chunks as written after successfully writing them.
5034 
5035 Version 1.6.14rc01 [October 15, 2014]
5036   Fixed some typos in comments.
5037 
5038 Version 1.6.14rc02 [October 17, 2014]
5039   Changed png_convert_to_rfc_1123() to png_convert_to_rfc_1123_buffer()
5040     in the manual, to reflect the change made in libpng-1.6.0.
5041   Updated README file to explain that direct access to the png_struct
5042     and info_struct members has not been permitted since libpng-1.5.0.
5043 
5044 Version 1.6.14 [October 23, 2014]
5045   No changes.
5046 
5047 Version 1.6.15beta01 [October 29, 2014]
5048   Changed "if (!x)" to "if (x == 0)" and "if (x)" to "if (x != 0)"
5049   Simplified png_free_data().
5050   Added missing "ptr = NULL" after some instances of png_free().
5051 
5052 Version 1.6.15beta02 [November 1, 2014]
5053   Changed remaining "if (!x)" to "if (x == 0)" and "if (x)" to "if (x != 0)"
5054 
5055 Version 1.6.15beta03 [November 3, 2014]
5056   Added PNG_USE_ARM_NEON configuration flag (Marcin Juszkiewicz).
5057 
5058 Version 1.6.15beta04 [November 4, 2014]
5059   Removed new PNG_USE_ARM_NEON configuration flag and made a one-line
5060     revision to configure.ac to support ARM on aarch64 instead (John Bowler).
5061 
5062 Version 1.6.15beta05 [November 5, 2014]
5063   Use png_get_libpng_ver(NULL) instead of PNG_LIBPNG_VER_STRING in
5064     example.c, pngtest.c, and applications in the contrib directory.
5065   Fixed an out-of-range read in png_user_version_check() (Bug report from
5066     Qixue Xiao, CVE-2015-8540).
5067   Simplified and future-proofed png_user_version_check().
5068   Fixed GCC unsigned int->float warnings. Various versions of GCC
5069     seem to generate warnings when an unsigned value is implicitly
5070     converted to double. This is probably a GCC bug but this change
5071     avoids the issue by explicitly converting to (int) where safe.
5072   Free all allocated memory in pngimage. The file buffer cache was left
5073     allocated at the end of the program, harmless but it causes memory
5074     leak reports from clang.
5075   Fixed array size calculations to avoid warnings. At various points
5076     in the code the number of elements in an array is calculated using
5077     sizeof.  This generates a compile time constant of type (size_t) which
5078     is then typically assigned to an (unsigned int) or (int). Some versions
5079     of GCC on 64-bit systems warn about the apparent narrowing, even though
5080     the same compiler does apparently generate the correct, in-range,
5081     numeric constant.  This adds appropriate, safe, casts to make the
5082     warnings go away.
5083 
5084 Version 1.6.15beta06 [November 6, 2014]
5085   Reverted use png_get_libpng_ver(NULL) instead of PNG_LIBPNG_VER_STRING
5086     in the manual, example.c, pngtest.c, and applications in the contrib
5087     directory.  It was incorrect advice.
5088 
5089 Version 1.6.15beta07 [November 7, 2014]
5090   Removed #ifdef PNG_16BIT_SUPPORTED/#endif around png_product2(); it is
5091     needed by png_reciprocal2().
5092   Added #ifdef PNG_16BIT_SUPPORTED/#endif around png_log16bit() and
5093     png_do_swap().
5094   Changed all "#endif /* PNG_FEATURE_SUPPORTED */" to "#endif /* FEATURE */"
5095 
5096 Version 1.6.15beta08 [November 8, 2014]
5097   More housecleaning in *.h
5098 
5099 Version 1.6.15rc01 [November 13, 2014]
5100 
5101 Version 1.6.15rc02 [November 14, 2014]
5102   The macros passed in the command line to Borland make were ignored if
5103     similarly-named macros were already defined in makefiles. This behavior
5104     is different from POSIX make and other make programs.  Surround the
5105     macro definitions with ifndef guards (Cosmin).
5106 
5107 Version 1.6.15rc03 [November 16, 2014]
5108   Added "-D_CRT_SECURE_NO_WARNINGS" to CFLAGS in scripts/makefile.vcwin32.
5109   Removed the obsolete $ARCH variable from scripts/makefile.darwin.
5110 
5111 Version 1.6.15 [November 20, 2014]
5112   No changes.
5113 
5114 Version 1.6.16beta01 [December 14, 2014]
5115   Added ".align 2" to arm/filter_neon.S to support old GAS assemblers that
5116     don't do alignment correctly.
5117   Revised Makefile.am and scripts/symbols.dfn to work with MinGW/MSYS
5118     (Bob Friesenhahn).
5119 
5120 Version 1.6.16beta02 [December 15, 2014]
5121   Revised Makefile.am and scripts/*.dfn again to work with MinGW/MSYS;
5122     renamed scripts/*.dfn to scripts/*.c (John Bowler).
5123 
5124 Version 1.6.16beta03 [December 21, 2014]
5125   Quiet a "comparison always true" warning in pngstest.c (John Bowler).
5126 
5127 Version 1.6.16rc01 [December 21, 2014]
5128   Restored a test on width that was removed from png.c at libpng-1.6.9
5129     (Bug report by Alex Eubanks, CVE-2015-0973).
5130 
5131 Version 1.6.16rc02 [December 21, 2014]
5132   Undid the update to pngrutil.c in 1.6.16rc01.
5133 
5134 Version 1.6.16rc03 [December 21, 2014]
5135   Fixed an overflow in png_combine_row() with very wide interlaced images
5136     (Bug report and fix by John Bowler, CVE-2014-9495).
5137 
5138 Version 1.6.16 [December 22, 2014]
5139   No changes.
5140 
5141 Version 1.6.17beta01 [January 29, 2015]
5142   Removed duplicate PNG_SAFE_LIMITS_SUPPORTED handling from pngconf.h
5143   Corrected the width limit calculation in png_check_IHDR().
5144   Removed user limits from pngfix. Also pass NULL pointers to
5145     png_read_row to skip the unnecessary row de-interlace stuff.
5146   Added testing of png_set_packing() to pngvalid.c
5147   Regenerated configure scripts in the *.tar distributions with libtool-2.4.4
5148   Implement previously untested cases of libpng transforms in pngvalid.c
5149   Fixed byte order in png_do_read_filler() with 16-bit input. Previously
5150     the high and low bytes of the filler, from png_set_filler() or from
5151     png_set_add_alpha(), were read in the wrong order.
5152   Made the check for out-of-range values in png_set_tRNS() detect
5153     values that are exactly 2^bit_depth, and work on 16-bit platforms.
5154   Merged some parts of libpng-1.6.17beta01 and libpng-1.7.0beta47.
5155   Added #ifndef __COVERITY__ where needed in png.c, pngrutil.c and
5156     pngset.c to avoid warnings about dead code.
5157   Added "& 0xff" to many instances of expressions that are typecast
5158     to (png_byte), to avoid Coverity warnings.
5159 
5160 Version 1.6.17beta02 [February 7, 2015]
5161   Work around one more Coverity-scan dead-code warning.
5162   Do not build png_product2() when it is unused.
5163 
5164 Version 1.6.17beta03 [February 17, 2015]
5165   Display user limits in the output from pngtest.
5166   Eliminated the PNG_SAFE_LIMITS macro and restored the 1-million-column
5167     and 1-million-row default limits in pnglibconf.dfa, that can be reset
5168     by the user at build time or run time.  This provides a more robust
5169     defense against DOS and as-yet undiscovered overflows.
5170 
5171 Version 1.6.17beta04 [February 21, 2015]
5172   Added PNG_WRITE_CUSTOMIZE_COMPRESSION_SUPPORTED macro, on by default.
5173   Allow user to call png_get_IHDR() with NULL arguments (Reuben Hawkins).
5174   Rebuilt configure scripts with automake-1.15 and libtool-2.4.6
5175 
5176 Version 1.6.17beta05 [February 25, 2015]
5177   Restored compiling of png_reciprocal2 with PNG_NO_16BIT.
5178 
5179 Version 1.6.17beta06 [February 27, 2015]
5180   Moved png_set_filter() prototype into a PNG_WRITE_SUPPORTED block
5181     of png.h.
5182   Avoid runtime checks when converting integer to png_byte with
5183     Visual Studio (Sergey Kosarevsky)
5184 
5185 Version 1.6.17rc01 [March 4, 2015]
5186   No changes.
5187 
5188 Version 1.6.17rc02 [March 9, 2015]
5189   Removed some comments that the configure script did not handle
5190     properly from scripts/pnglibconf.dfa and pnglibconf.h.prebuilt.
5191   Free the unknown_chunks structure even when it contains no data.
5192 
5193 Version 1.6.17rc03 [March 12, 2015]
5194   Updated CMakeLists.txt to add OSX framework, change YES/NO to ON/OFF
5195     for consistency, and remove some useless tests (Alexey Petruchik).
5196 
5197 Version 1.6.17rc04 [March 16, 2015]
5198   Remove pnglibconf.h, pnglibconf.c, and pnglibconf.out instead of
5199     pnglibconf.* in "make clean" (Cosmin).
5200   Fix bug in calculation of maxbits, in png_write_sBIT, introduced
5201     in libpng-1.6.17beta01 (John Bowler).
5202 
5203 Version 1.6.17rc05 [March 21, 2015]
5204   Define PNG_FILTER_* and PNG_FILTER_VALUE_* in png.h even when WRITE
5205     is not supported (John Bowler).  This fixes an error introduced in
5206     libpng-1.6.17beta06.
5207   Reverted "& 0xff" additions of version 1.6.17beta01. Libpng passes
5208     the Coverity scan without them.
5209 
5210 Version 1.6.17rc06 [March 23, 2015]
5211   Remove pnglibconf.dfn and pnglibconf.pre with "make clean".
5212   Reformatted some "&0xff" instances to "& 0xff".
5213   Fixed simplified 8-bit-linear to sRGB alpha. The calculated alpha
5214     value was wrong.  It's not clear if this affected the final stored
5215     value; in the obvious code path the upper and lower 8-bits of the
5216     alpha value were identical and the alpha was truncated to 8-bits
5217     rather than dividing by 257 (John Bowler).
5218 
5219 Version 1.6.17 [March 26, 2015]
5220   No changes.
5221 
5222 Version 1.6.18beta01 [April 1, 2015]
5223   Removed PNG_SET_CHUNK_[CACHE|MALLOC]_LIMIT_SUPPORTED macros.  They
5224     have been combined with PNG_SET_USER_LIMITS_SUPPORTED (resolves
5225     bug report by Andrew Church).
5226   Fixed rgb_to_gray checks and added tRNS checks to pngvalid.c.  This
5227     fixes some arithmetic errors that caused some tests to fail on
5228     some 32-bit platforms (Bug reports by Peter Breitenlohner [i686]
5229     and Petr Gajdos [i586]).
5230 
5231 Version 1.6.18beta02 [April 26, 2015]
5232   Suppressed some warnings from the Borland C++ 5.5.1/5.82 compiler
5233     (Bug report by Viktor Szakats).
5234 
5235 Version 1.6.18beta03 [May 6, 2015]
5236   Replaced "unexpected" with an integer (0xabadca11) in pngset.c
5237     where a long was expected, to avoid a compiler warning when PNG_DEBUG > 1.
5238   Added contrib/examples/simpleover.c, to demonstrate how to handle
5239     alpha compositing of multiple images, using the "simplified API"
5240     and an example PNG generation tool, contrib/examples/genpng.c
5241     (John Bowler).
5242 
5243 Version 1.6.18beta04 [May 20, 2015]
5244   PNG_RELEASE_BUILD replaces tests where the code depended on the build base
5245     type and can be defined on the command line, allowing testing in beta
5246     builds (John Bowler).
5247   Avoid Coverity issue 80858 (REVERSE NULL) in pngtest.c PNG_DEBUG builds.
5248   Avoid a harmless potential integer overflow in png_XYZ_from_xy() (Bug
5249     report from Christopher Ferris).
5250 
5251 Version 1.6.18beta05 [May 31, 2015]
5252   Backport filter selection code from libpng-1.7.0beta51, to combine
5253     sub_row, up_row, avg_row, and paeth_row into try_row and tst_row.
5254   Changed png_voidcast(), etc., to voidcast(), etc., in contrib/tools/pngfix.c
5255     to avoid confusion with the libpng private macros.
5256   Fixed old cut&paste bug in the weighted filter selection code in
5257     pngwutil.c, introduced in libpng-0.95, March 1997.
5258 
5259 Version 1.6.18beta06 [June 1, 2015]
5260   Removed WRITE_WEIGHTED_FILTERED code, to save a few kbytes of the
5261     compiled library size. It never worked properly and as far as we can
5262     tell, no one uses it. The png_set_filter_heuristics() and
5263     png_set_filter_heuristics_fixed() APIs are retained but deprecated
5264     and do nothing.
5265 
5266 Version 1.6.18beta07 [June 6, 2015]
5267   Removed non-working progressive reader 'skip' function. This
5268     function has apparently never been used. It was implemented
5269     to support back-door modification of png_struct in libpng-1.4.x
5270     but (because it does nothing and cannot do anything) was apparently
5271     never tested (John Bowler).
5272   Fixed cexcept.h in which GCC 5 now reports that one of the auto
5273     variables in the Try macro needs to be volatile to prevent value
5274     being lost over the setjmp (John Bowler).
5275   Fixed NO_WRITE_FILTER and -Wconversion build breaks (John Bowler).
5276   Fix g++ build breaks (John Bowler).
5277   Quieted some Coverity issues in pngfix.c, png-fix-itxt.c, pngvalid.c,
5278     pngstest.c, and pngimage.c. Most seem harmless, but png-fix-itxt
5279     would only work with iTXt chunks with length 255 or less.
5280   Added #ifdef's to contrib/examples programs so people don't try
5281     to compile them without the minimum required support enabled
5282     (suggested by Flavio Medeiros).
5283 
5284 Version 1.6.18beta08 [June 30, 2015]
5285   Eliminated the final two Coverity defects (insecure temporary file
5286     handling in contrib/libtests/pngstest.c; possible overflow of
5287     unsigned char in contrib/tools/png-fix-itxt.c). To use the "secure"
5288     file handling, define PNG_USE_MKSTEMP, otherwise "tmpfile()" will
5289     be used.
5290   Removed some unused WEIGHTED_FILTER macros from png.h and pngstruct.h
5291 
5292 Version 1.6.18beta09 [July 5, 2015]
5293   Removed some useless typecasts from contrib/tools/png-fix-itxt.c
5294   Fixed a new signed-unsigned comparison in pngrtran.c (Max Stepin).
5295   Replaced arbitrary use of 'extern' with #define PNG_LINKAGE_*.  To
5296     preserve API compatibility, the new defines all default to "extern"
5297     (requested by Jan Nijtmans).
5298 
5299 Version 1.6.18rc01 [July 9, 2015]
5300   Belatedly added Mans Rullgard and James Yu to the list of Contributing
5301     Authors.
5302 
5303 Version 1.6.18rc02 [July 12, 2015]
5304   Restored unused FILTER_HEURISTIC macros removed at libpng-1.6.18beta08
5305     to png.h to avoid compatibility warnings.
5306 
5307 Version 1.6.18rc03 [July 15, 2015]
5308   Minor changes to the man page
5309 
5310 Version 1.6.18 [July 23, 2015]
5311   No changes.
5312 
5313 Version 1.6.19beta01 [July 30, 2015]
5314   Updated obsolete information about the simplified API macros in the
5315     manual pages (Bug report by Arc Riley).
5316   Avoid potentially dereferencing NULL info_ptr in png_info_init_3().
5317   Rearranged png.h to put the major sections in the same order as
5318     in libpng17.
5319   Eliminated unused PNG_COST_SHIFT, PNG_WEIGHT_SHIFT, PNG_COST_FACTOR, and
5320     PNG_WEIGHT_FACTOR macros.
5321   Suppressed some warnings from the Borland C++ 5.5.1/5.82 compiler
5322     (Bug report by Viktor Szakats).  Several warnings remain and are
5323     unavoidable, where we test for overflow.
5324   Fixed potential leak of png_pixels in contrib/pngminus/pnm2png.c
5325   Fixed uninitialized variable in contrib/gregbook/rpng2-x.c
5326 
5327 Version 1.6.19beta02 [August 19, 2015]
5328   Moved config.h.in~ from the "libpng_autotools_files" list to the
5329     "libpng_autotools_extra" list in autogen.sh because it was causing a
5330     false positive for missing files (bug report by Robert C. Seacord).
5331   Removed unreachable "break" statements in png.c, pngread.c, and pngrtran.c
5332     to suppress clang warnings (Bug report by Viktor Szakats).
5333   Fixed some bad links in the man page.
5334   Changed "n bit" to "n-bit" in comments.
5335   Added signed/unsigned 16-bit safety net. This removes the dubious
5336     0x8000 flag definitions on 16-bit systems. They aren't supported
5337     yet the defs *probably* work, however it seems much safer to do this
5338     and be advised if anyone, contrary to advice, is building libpng 1.6
5339     on a 16-bit system. It also adds back various switch default clauses
5340     for GCC; GCC errors out if they are not present (with an appropriately
5341     high level of warnings).
5342   Safely convert num_bytes to a png_byte in png_set_sig_bytes() (Robert
5343     Seacord).
5344   Fixed the recently reported 1's complement security issue by replacing
5345     the value that is illegal in the PNG spec, in both signed and unsigned
5346     values, with 0. Illegal unsigned values (anything greater than or equal
5347     to  0x80000000) can still pass through, but since these are not illegal
5348     in ANSI-C (unlike 0x80000000 in the signed case) the checking that
5349     occurs later can catch them (John Bowler).
5350 
5351 Version 1.6.19beta03 [September 26, 2015]
5352   Fixed png_save_int_32 when int is not 2's complement (John Bowler).
5353   Updated libpng16 with all the recent test changes from libpng17,
5354     including changes to pngvalid.c to ensure that the original,
5355     distributed, version of contrib/visupng/cexcept.h can be used
5356     (John Bowler).
5357   pngvalid contains the correction to the use of SAVE/STORE_
5358     UNKNOWN_CHUNKS; a bug revealed by changes in libpng 1.7. More
5359     tests contain the --strict option to detect warnings and the
5360     pngvalid-standard test has been corrected so that it does not
5361     turn on progressive-read. There is a separate test which does
5362     that. (John Bowler)
5363   Also made some signed/unsigned fixes.
5364   Make pngstest error limits version specific. Splitting the machine
5365     generated error structs out to a file allows the values to be updated
5366     without changing pngstest.c itself. Since libpng 1.6 and 1.7 have
5367     slightly different error limits this simplifies maintenance. The
5368     makepngs.sh script has also been updated to more accurately reflect
5369     current problems in libpng 1.7 (John Bowler).
5370   Incorporated new test PNG files into make check.  tests/pngstest-*
5371     are changed so that the new test files are divided into 8 groups by
5372     gamma and alpha channel.  These tests have considerably better code
5373     and pixel-value coverage than contrib/pngsuite; however,coverage is
5374     still incomplete (John Bowler).
5375   Removed the '--strict' in 1.6 because of the double-gamma-correction
5376     warning, updated pngstest-errors.h for the errors detected with the
5377     new contrib/testspngs PNG test files (John Bowler).
5378 
5379 Version 1.6.19beta04 [October 15, 2015]
5380   Worked around rgb-to-gray issues in libpng 1.6.  The previous
5381     attempts to ignore the errors in the code aren't quite enough to
5382     deal with the 'channel selection' encoding added to libpng 1.7; abort.
5383     pngvalid.c is changed to drop this encoding in prior versions.
5384   Fixed 'pow' macros in pngvalid.c. It is legal for 'pow' to be a
5385     macro, therefore the argument list cannot contain preprocessing
5386     directives.  Make sure pow is a function where this happens. This is
5387     a minimal safe fix, the issue only arises in non-performance-critical
5388     code (bug report by Curtis Leach, fix by John Bowler).
5389   Added sPLT support to pngtest.c
5390 
5391 Version 1.6.19rc01 [October 23, 2015]
5392   No changes.
5393 
5394 Version 1.6.19rc02 [October 31, 2015]
5395   Prevent setting or writing over-length PLTE chunk (Cosmin Truta).
5396   Silently truncate over-length PLTE chunk while reading.
5397   Libpng incorrectly calculated the output rowbytes when the application
5398     decreased either the number of channels or the bit depth (or both) in
5399     a user transform.  This was safe; libpng overallocated buffer space
5400    (potentially by quite a lot; up to 4 times the amount required) but,
5401    from 1.5.4 on, resulted in a png_error (John Bowler).
5402 
5403 Version 1.6.19rc03 [November 3, 2015]
5404   Fixed some inconsequential cut-and-paste typos in png_set_cHRM_XYZ_fixed().
5405   Clarified COPYRIGHT information to state explicitly that versions
5406     are derived from previous versions.
5407   Removed much of the long list of previous versions from png.h and
5408     libpng.3.
5409 
5410 Version 1.6.19rc04 [November 5, 2015]
5411   Fixed new bug with CRC error after reading an over-length palette
5412     (bug report by Cosmin Truta) (CVE-2015-8126).
5413 
5414 Version 1.6.19 [November 12, 2015]
5415   Cleaned up coding style in png_handle_PLTE().
5416 
5417 Version 1.6.20beta01 [November 20, 2015]
5418   Avoid potential pointer overflow/underflow in png_handle_sPLT() and
5419     png_handle_pCAL() (Bug report by John Regehr).
5420 
5421 Version 1.6.20beta02 [November 23, 2015]
5422   Fixed incorrect implementation of png_set_PLTE() that uses png_ptr
5423     not info_ptr, that left png_set_PLTE() open to the CVE-2015-8126
5424     vulnerability.  Fixes CVE-2015-8472.
5425 
5426 Version 1.6.20beta03 [November 24, 2015]
5427   Backported tests from libpng-1.7.0beta69.
5428 
5429 Version 1.6.20rc01 [November 26, 2015]
5430   Fixed an error in handling of bad zlib CMINFO field in pngfix, found by
5431     American Fuzzy Lop, reported by Brian Carpenter.  inflate() doesn't
5432     immediately fault a bad CMINFO field; instead a 'too far back' error
5433     happens later (at least some times).  pngfix failed to limit CMINFO to
5434     the allowed values but then assumed that window_bits was in range,
5435     triggering an assert. The bug is mostly harmless; the PNG file cannot
5436     be fixed.
5437 
5438 Version 1.6.20rc02 [November 29, 2015]
5439   In libpng 1.6 zlib initialization was changed to use the window size
5440     in the zlib stream, not a fixed value. This causes some invalid images,
5441     where CINFO is too large, to display 'correctly' if the rest of the
5442     data is valid.  This provides a workaround for zlib versions where the
5443     error arises (ones that support the API change to use the window size
5444     in the stream).
5445 
5446 Version 1.6.20 [December 3, 2015]
5447   No changes.
5448 
5449 Version 1.6.21beta01 [December 11, 2015]
5450   Fixed syntax "$(command)" in tests/pngstest that some shells other than
5451     bash could not parse (Bug report by Nelson Beebe). Use `command` instead.
5452 
5453 Version 1.6.21beta02 [December 14, 2015]
5454   Moved png_check_keyword() from pngwutil.c to pngset.c
5455   Removed LE/BE dependencies in pngvalid, to 'fix' the current problem
5456     in the BigEndian tests by not testing it, making the BE code the same 
5457     as the LE version.
5458   Fixes to pngvalid for various reduced build configurations (eliminate unused
5459     statics) and a fix for the case in rgb_to_gray when the digitize option
5460     reduces graylo to 0, producing a large error.
5461 
5462 Version 1.6.21beta03 [December 18, 2015]
5463   Widened the 'limit' check on the internally calculated error limits in
5464     the 'DIGITIZE' case (the code used prior to 1.7 for rgb_to_gray error
5465     checks) and changed the check to only operate in non-release builds
5466     (base build type not RC or RELEASE.)
5467   Fixed undefined behavior in pngvalid.c, undefined because
5468     (png_byte) << shift is undefined if it changes the signed bit
5469     (because png_byte is promoted to int). The libpng exported functions
5470     png_get_uint_32 and png_get_uint_16 handle this. (Bug reported by
5471     David Drysdale as a result of reports from UBSAN in clang 3.8).
5472   This changes pngvalid to use BE random numbers; this used to produce
5473     errors but these should not be fixed as a result of the previous changes.
5474 
5475 Version 1.6.21rc01 [January 4, 2016]
5476   In projects/vstudio, combined readme.txt and WARNING into README.txt
5477 
5478 Version 1.6.21rc02 [January 7, 2016]
5479   Relocated assert() in contrib/tools/pngfix.c, bug found by American
5480     Fuzzy Lop, reported by Brian Carpenter.
5481   Marked 'limit' UNUSED in transform_range_check().  This only affects
5482     release builds.
5483 
5484 Version 1.6.21 [January 15, 2016]
5485   Worked around a false-positive Coverity issue in pngvalid.c.
5486 
5487 Version 1.6.22beta01 [January 23, 2016]
5488   Changed PNG_USE_MKSTEMP to __COVERITY__ to select alternate
5489     "tmpfile()" implementation in contrib/libtests/pngstest.c
5490   Fixed NO_STDIO build of pngunknown.c to skip calling png_init_io()
5491     if there is no stdio.h support.
5492   Added a png_image_write_to_memory() API and a number of assist macros
5493     to allow an application that uses the simplified API write to bypass
5494     stdio and write directly to memory.
5495   Added some warnings (png.h) and some check code to detect *possible*
5496     overflow in the ROW_STRIDE and simplified image SIZE macros.  This
5497     disallows image width/height/format that *might* overflow.  This is
5498     a quiet API change that limits in-memory image size (uncompressed) to
5499     less than 4GByte and image row size (stride) to less than 2GByte.
5500   Revised workaround for false-positive Coverity issue in pngvalid.c.
5501 
5502 Version 1.6.22beta02 [February 8, 2016]
5503   Only use exit(77) in configure builds.
5504   Corrected error in PNG_IMAGE_PNG_SIZE_MAX. This new macro underreported
5505     the palette size because it failed to take into account that the memory
5506     palette has to be expanded to full RGB when it is written to PNG.
5507   Updated CMakeLists.txt, added supporting scripts/gen*.cmake.in
5508     and test.cmake.in (Roger Leigh).
5509   Relaxed limit checks on gamma values in pngrtran.c. As suggested in
5510     the comments gamma values outside the range currently permitted
5511     by png_set_alpha_mode are useful for HDR data encoding.  These values
5512     are already permitted by png_set_gamma so it is reasonable caution to
5513     extend the png_set_alpha_mode range as HDR imaging systems are starting
5514     to emerge.
5515 
5516 Version 1.6.22beta03 [March 9, 2016]
5517   Added a common-law trademark notice and export control information
5518     to the LICENSE file, png.h, and the man page.
5519   Restored "& 0xff" in png_save_uint_16() and png_save_uint_32() that
5520     were accidentally removed from libpng-1.6.17. 
5521   Changed PNG_INFO_cHNK and PNG_FREE_cHNK from 0xnnnn to 0xnnnnU in png.h
5522     (Robert C. Seacord).
5523   Removed dubious "#if INT_MAX" test from png.h that was added to
5524     libpng-1.6.19beta02 (John Bowler).
5525   Add ${INCLUDES} in scripts/genout.cmake.in (Bug report by Nixon Kwok).
5526   Updated LICENSE to say files in the contrib directory are not
5527     necessarily under the libpng license, and that some makefiles have
5528     other copyright owners.
5529   Added INTEL-SSE2 support (Mike Klein and Matt Sarett, Google, Inc.).
5530   Made contrib/libtests/timepng more robust.  The code no longer gives
5531     up/fails on invalid PNG data, it just skips it (with error messages).
5532     The code no longer fails on PNG files with data beyond IEND.  Options
5533     exist to use png_read_png (reading the whole image, not by row) and, in
5534     that case, to apply any of the supported transforms.  This makes for
5535     more realistic testing; the decoded data actually gets used in a
5536     meaningful fashion (John Bowler).
5537   Fixed some misleading indentation (Krishnaraj Bhat).
5538 
5539 Version 1.6.22beta04 [April 5, 2016]
5540   Force GCC compilation to C89 if needed (Dagobert Michelsen).
5541   SSE filter speed improvements for bpp=3:
5542     memcpy-free implementations of load3() / store3().
5543     call load3() only when needed at the end of a scanline.
5544 
5545 Version 1.6.22beta05 [April 27, 2016]
5546   Added PNG_FAST_FILTERS macro (defined as
5547     PNG_FILTER_NONE|PNG_FILTER_SUB|PNG_FILTER_UP).
5548   Various fixes for contrib/libtests/timepng.c
5549   Moved INTEL-SSE code from pngpriv.h into contrib/intel/intel_sse.patch.
5550   Fixed typo (missing underscore) in #define PNG_READ_16_TO_8_SUPPORTED
5551     (Bug report by Y.Ohashik).
5552 
5553 Version 1.6.22beta06 [May 5, 2016]
5554   Rebased contrib/intel_sse.patch.
5555   Quieted two Coverity issues in contrib/libtests/timepng.c.
5556   Fixed issues with scripts/genout.cmake.in (David Capello, Nixon Kwok):
5557     Added support to use multiple directories in ZLIBINCDIR variable,
5558     Fixed CMAKE_C_FLAGS with multiple values when genout is compiled on MSVC,
5559     Fixed pnglibconf.c compilation on OS X including the sysroot path.
5560 
5561 Version 1.6.22rc01 [May 14, 2016]
5562   No changes.
5563 
5564 Version 1.6.22rc02 [May 16, 2016]
5565   Removed contrib/timepng from default build; it does not build on platforms
5566     that don't supply clock_gettime().
5567 
5568 Version 1.6.22rc03 [May 17, 2016]
5569   Restored contrib/timepng to default build but check for the presence
5570     of clock_gettime() in configure.ac and Makefile.am.
5571 
5572 Version 1.6.22 [May 26, 2016]
5573   No changes.
5574 
5575 Version 1.6.23beta01 [May 29, 2016]
5576   Stop a potential memory leak in png_set_tRNS() (Bug report by Ted Ying).
5577   Fixed the progressive reader to handle empty first IDAT chunk properly
5578     (patch by Timothy Nikkel).  This bug was introduced in libpng-1.6.0 and
5579     only affected the libpng16 branch.
5580   Added tests in pngvalid.c to check zero-length IDAT chunks in various
5581     positions.  Fixed the sequential reader to handle these more robustly
5582     (John Bowler).
5583 
5584 Version 1.6.23rc01 [June 2, 2016]
5585   Corrected progressive read input buffer in pngvalid.c. The previous version
5586     the code invariably passed just one byte at a time to libpng.  The intent
5587     was to pass a random number of bytes in the range 0..511.
5588   Moved sse2 prototype from pngpriv.h to contrib/intel/intel_sse.patch.
5589   Added missing ")" in pngerror.c (Matt Sarrett).
5590 
5591 Version 1.6.23rc02 [June 4, 2016]
5592   Fixed undefined behavior in png_push_save_buffer(). Do not call
5593     memcpy() with a null source, even if count is zero (Leon Scroggins III).
5594 
5595 Version 1.6.23 [June 9, 2016]
5596   Fixed bad link to RFC2083 in png.5 (Nikola Forro).
5597 
5598 Version 1.6.24beta01 [June 11, 2016]
5599   Avoid potential overflow of the PNG_IMAGE_SIZE macro.  This macro
5600     is not used within libpng, but is used in some of the examples.
5601 
5602 Version 1.6.24beta02 [June 23, 2016]
5603   Correct filter heuristic overflow handling. This was broken when the
5604     write filter code was moved out-of-line; if there is a single filter and
5605     the heuristic sum overflows the calculation of the filtered line is not
5606     completed.  In versions prior to 1.6 the code was duplicated in-line
5607     and the check not performed, so the filter operation completed; however,
5608     in the multi-filter case where the sum is performed the 'none' filter would
5609     be selected if all the sums overflowed, even if it wasn't in the filter
5610     list.  The fix to the first problem is simply to provide PNG_SIZE_MAX as
5611     the current lmins sum value; this means the sum can never exceed it and
5612     overflows silently.  A reasonable compiler that does choose to inline
5613     the code will simply eliminate the sum check.
5614   The fix to the second problem is to use high precision arithmetic (this is
5615     implemented in 1.7), however a simple safe fix here is to chose the lowest
5616     numbered filter in the list from png_set_filter (this only works if the
5617     first problem is also fixed) (John Bowler).
5618   Use a more efficient absolute value calculation on SSE2 (Matthieu Darbois).
5619   Fixed the case where PNG_IMAGE_BUFFER_SIZE can overflow in the application
5620     as a result of the application using an increased 'row_stride'; previously
5621     png_image_finish_read only checked for overflow on the base calculation of
5622     components.  (I.e. it checked for overflow of a 32-bit number on the total
5623     number of pixel components in the output format, not the possibly padded row
5624     length and not the number of bytes, which for linear formats is twice the
5625     number of components.)
5626   MSVC does not like '-(unsigned)', so replaced it with 0U-(unsigned)
5627   MSVC does not like (uInt) = -(unsigned) (i.e. as an initializer), unless
5628     the conversion is explicitly invoked by a cast.
5629   Put the SKIP definition in the correct place. It needs to come after the
5630     png.h include (see all the other .c files in contrib/libtests) because it
5631     depends on PNG_LIBPNG_VER.
5632   Removed the three compile warning options from the individual project
5633     files into the zlib.props globals.  It increases the warning level from 4
5634     to All and adds a list of the warnings that need to be turned off.  This is
5635     semi-documentary; the intent is to tell libpng users which warnings have
5636     been examined and judged non-fixable at present.  The warning about
5637     structure padding is fixable, but it would be a significant change (moving
5638     structure members around).
5639 
5640 Version 1.6.24beta03 [July 4, 2016]
5641   Optimized absolute value calculation in filter selection, similar to
5642     code in the PAETH decoder in pngrutil.c. Build with PNG_USE_ABS to
5643     use this.
5644   Added pngcp to the build together with a pngcp.dfa configuration test.
5645   Added high resolution timing to pngcp.
5646   Added "Common linking failures" section to INSTALL.
5647   Relocated misplaced #endif in png.c sRGB profile checking.
5648   Fixed two Coverity issues in pngcp.c.
5649 
5650 Version 1.6.24beta04 [July 8, 2016]
5651   Avoid filter-selection heuristic sum calculations in cases where only one
5652     filter is a candidate for selection. This trades off code size (added
5653     private png_setup_*_row_only() functions) for speed.
5654 
5655 Version 1.6.24beta05 [July 13, 2016]
5656   Fixed some indentation to comply with our coding style.
5657   Added contrib/tools/reindent.
5658 
5659 Version 1.6.24beta06 [July 18, 2016]
5660   Fixed more indentation to comply with our coding style.
5661   Eliminated unnecessary tests of boolean png_isaligned() vs 0.
5662 
5663 Version 1.6.24rc01 [July 25, 2016]
5664   No changes.
5665 
5666 Version 1.6.24rc02 [August 1, 2016]
5667   Conditionally compile SSE2 headers in contrib/intel/intel_sse.patch
5668   Conditionally compile png_decompress_chunk().
5669 
5670 Version 1.6.24rc03 [August 2, 2016]
5671   Conditionally compile ARM_NEON headers in pngpriv.h
5672   Updated contrib/intel/intel_sse.patch
5673 
5674 Version 1.6.24[August 4, 2016]
5675   No changes.
5676 
5677 Version 1.6.25beta01 [August 12, 2016]
5678   Reject oversized iCCP profile immediately.
5679   Cleaned up PNG_DEBUG compile of pngtest.c.
5680   Conditionally compile png_inflate().
5681 
5682 Version 1.6.25beta02 [August 18, 2016]
5683   Don't install pngcp; it conflicts with pngcp in the pngtools package.
5684   Minor editing of INSTALL, (whitespace, added copyright line)
5685 
5686 Version 1.6.25rc01 [August 24, 2016]
5687   No changes.
5688 
5689 Version 1.6.25rc02 [August 29, 2016]
5690   Added MIPS support (Mandar Sahastrabuddhe <Mandar.Sahastrabuddhe@imgtec.com>).
5691   Only the UP filter is currently implemented.
5692 
5693 Version 1.6.25rc03 [August 29, 2016]
5694   Rebased contrib/intel/intel_sse.patch after the MIPS implementation.
5695 
5696 Version 1.6.25rc04 [August 30, 2016]
5697   Added MIPS support for SUB, AVG, and PAETH filters (Mandar Sahastrabuddhe).
5698 
5699 Version 1.6.25rc05 [August 30, 2016]
5700   Rebased contrib/intel/intel_sse.patch after the MIPS implementation update..
5701 
5702 Version 1.6.25 [September 1, 2016]
5703   No changes.
5704 
5705 Version 1.6.26beta01 [September 26, 2016]
5706   Fixed handling zero length IDAT in pngfix (bug report by Agostino Sarubbo,
5707     bugfix by John Bowler).
5708   Do not issue a png_error() on read in png_set_pCAL() because png_handle_pCAL
5709     has allocated memory that libpng needs to free.
5710   Conditionally compile png_set_benign_errors() in pngread.c and pngtest.c
5711   Issue a png_benign_error instead of a png_error on ADLER32 mismatch
5712     while decoding compressed data chunks.
5713   Changed PNG_ZLIB_VERNUM to ZLIB_VERNUM in pngpriv.h, pngstruct.h, and
5714     pngrutil.c.
5715   If CRC handling of critical chunks has been set to PNG_CRC_QUIET_USE,
5716     ignore the ADLER32 checksum in the IDAT chunk as well as the chunk CRCs.
5717   Issue png_benign_error() on ADLER32 checksum mismatch instead of png_error().
5718   Add tests/badcrc.png and tests/badadler.png to tests/pngtest.
5719   Merged pngtest.c with libpng-1.7.0beta84/pngtest.c
5720 
5721 Version 1.6.26beta02 [October 1, 2016]
5722   Updated the documentation about CRC and ADLER32 handling.
5723   Quieted 117 warnings from clang-3.8 in pngtrans.c, pngread.c,
5724      pngwrite.c, pngunknown.c, and pngvalid.c.
5725   Quieted 58 (out of 144) -Wconversion compiler warnings by changing
5726     flag definitions in pngpriv.h from 0xnnnn to 0xnnnnU and trivial changes
5727     in png.c, pngread.c, and pngwutil.c.
5728 
5729 Version 1.6.26beta03 [October 2, 2016]
5730   Removed contrib/libtests/*.orig and *.rej that slipped into the tarballs.
5731   Quieted the 86 remaining -Wconversion compiler warnings by
5732     revising the png_isaligned() macro and trivial changes in png.c,
5733     pngerror.c, pngget.c, pngmem.c, pngset.c, pngrtran.c, pngrutil.c,
5734     pngwtran.c, pngwrite.c, and pngwutil.c.
5735 
5736 Version 1.6.26beta04 [October 3, 2016]
5737   Quieted (bogus?) clang warnings about "absolute value has no effect"
5738     when PNG_USE_ABS is defined.
5739   Fixed offsets in contrib/intel/intel_sse.patch
5740 
5741 Version 1.6.26beta05 [October 6, 2016]
5742   Changed integer constant 4294967294 to unsigned 4294967294U in pngconf.h
5743     to avoid a signed/unsigned compare in the preprocessor.
5744 
5745 Version 1.6.26beta06 [October 7, 2016]
5746   Use zlib-1.2.8.1 inflateValidate() instead of inflateReset2() to
5747     optionally avoid ADLER32 evaluation.
5748 
5749 Version 1.6.26rc01 [October 12, 2016]
5750   No changes.
5751 
5752 Version 1.6.26 [October 20, 2016]
5753   Cosmetic change, "ptr != 0" to "ptr != NULL" in png.c and pngrutil.c
5754   Despammed email addresses (replaced "@" with " at ").
5755 
5756 Version 1.6.27beta01 [November 2, 2016]
5757   Restrict the new ADLER32-skipping to IDAT chunks.  It broke iCCP chunk
5758     handling: an erroneous iCCP chunk would throw a png_error and reject the
5759     entire PNG image instead of rejecting just the iCCP chunk with a warning,
5760     if built with zlib-1.2.8.1.
5761 
5762 Version 1.6.27rc01 [December 27, 2016]
5763   Control ADLER32 checking with new PNG_IGNORE_ADLER32 option. Fixes
5764     an endless loop when handling erroneous ADLER32 checksums; bug
5765     introduced in libpng-1.6.26.
5766   Removed the use of a macro containing the pre-processor 'defined'
5767     operator.  It is unclear whether this is valid; a macro that
5768     "generates" 'defined' is not permitted, but the use of the word
5769     "generates" within the C90 standard seems to imply more than simple
5770     substitution of an expression itself containing a well-formed defined
5771     operation.
5772   Added ARM support to CMakeLists.txt (Andreas Franek).
5773 
5774 Version 1.6.27 [December 29, 2016]
5775   Fixed a potential null pointer dereference in png_set_text_2() (bug report
5776     and patch by Patrick Keshishian, CVE-2016-10087).
5777 
5778 Version 1.6.28rc01 [January 3, 2017]
5779   Fixed arm/aarch64 detection in CMakeLists.txt (Gianfranco Costamagna).
5780   Added option to Cmake build allowing a custom location of zlib to be
5781     specified in a scenario where libpng is being built as a subproject
5782     alongside zlib by another project (Sam Serrels).
5783   Changed png_ptr->options from a png_byte to png_uint_32, to accommodate
5784     up to 16 options.
5785 
5786 Version 1.6.28rc02 [January 4, 2017]
5787   Added "include(GNUInstallDirs)" to CMakeLists.txt (Gianfranco Costamagna).
5788   Moved SSE2 optimization code into the main libpng source directory.
5789     Configure libpng with "configure --enable-intel-sse" or compile
5790     libpng with "-DPNG_INTEL_SSE" in CPPFLAGS to enable it.
5791 
5792 Version 1.6.28rc03 [January 4, 2017]
5793   Backed out the SSE optimization and last CMakeLists.txt to allow time for QA.
5794 
5795 Version 1.6.28 [January 5, 2017]
5796   No changes.
5797 
5798 Version 1.6.29beta01 [January 12, 2017]
5799   Readded "include(GNUInstallDirs)" to CMakeLists.txt (Gianfranco Costamagna).
5800   Moved SSE2 optimization code into the main libpng source directory.
5801     Configure libpng with "configure --enable-intel-sse" or compile
5802     libpng with "-DPNG_INTEL_SSE" in CPPFLAGS to enable it.
5803   Simplified conditional compilation in pngvalid.c, for AIX (Michael Felt).
5804 
5805 Version 1.6.29beta02 [February 22, 2017]
5806   Avoid conditional directives that break statements in pngrutil.c (Romero
5807     Malaquias)
5808   The contrib/examples/pngtopng.c recovery code was in the wrong "if"
5809     branches; the comments were correct.
5810   Added code for PowerPC VSX optimisation (Vadim Barkov).
5811 
5812 Version 1.6.29beta03 [March 1, 2017]
5813   Avoid potential overflow of shift operations in png_do_expand() (Aaron Boxer).
5814   Change test ZLIB_VERNUM >= 0x1281 to ZLIB_VERNUM >= 0x1290 in pngrutil.c
5815     because Solaris 11 distributes zlib-1.2.8.f that is older than 1.2.8.1,
5816     as suggested in zlib FAQ, item 24.
5817   Suppress clang warnings about implicit sign changes in png.c
5818 
5819 Version 1.6.29 [March 16, 2017]
5820   No changes.
5821 
5822 Version 1.6.30beta01 [April 1, 2017]
5823   Added missing "$(CPPFLAGS)" to the compile line for c.pic.o in
5824     makefile.linux and makefile.solaris-x86 (Cosmin).
5825   Revised documentation of png_get_error_ptr() in the libpng manual.
5826   Silence clang -Wcomma and const drop warnings (Viktor Szakats).
5827   Update Sourceforge URLs in documentation (https instead of http).
5828 
5829 Version 1.6.30beta02 [April 22, 2017]
5830   Document need to check for integer overflow when allocating a pixel
5831     buffer for multiple rows in contrib/gregbook, contrib/pngminus,
5832     example.c, and in the manual (suggested by Jaeseung Choi). This
5833     is similar to the bug reported against pngquant in CVE-2016-5735.
5834   Removed reference to the obsolete PNG_SAFE_LIMITS macro in the documentation.
5835 
5836 Version 1.6.30beta03 [May 22, 2017]
5837   Check for integer overflow in contrib/visupng and contrib/tools/genpng.
5838   Do not double evaluate CMAKE_SYSTEM_PROCESSOR in CMakeLists.txt.
5839   Test CMAKE_HOST_WIN32 instead of WIN32 in CMakeLists.txt.
5840   Fix some URL in documentation.
5841 
5842 Version 1.6.30beta04 [June 7, 2017]
5843   Avoid writing an empty IDAT when the last IDAT exactly fills the
5844     compression buffer (bug report by Brian Baird). This bug was
5845     introduced in libpng-1.6.0.
5846 
5847 Version 1.6.30rc01 [June 14, 2017]
5848   No changes.
5849 
5850 Version 1.6.30rc02 [June 25, 2017]
5851   Update copyright year in pnglibconf.h, make ltmain.sh executable.
5852   Add a reference to the libpng.download site in README.
5853 
5854 Version 1.6.30 [June 28, 2017]
5855   No changes.
5856 
5857 Version 1.6.31beta01 [July 5, 2017]
5858   Guard the definition of _POSIX_SOURCE in pngpriv.h (AIX already defines it;
5859     bug report by Michael Felt).
5860   Revised pngpriv.h to work around failure to compile arm/filter_neon.S
5861     ("typedef" directive is unrecognized by the assembler). The problem
5862     was introduced in libpng-1.6.30beta01.
5863   Added "Requires: zlib" to libpng.pc.in (Pieter Neerincx).
5864   Added special case for FreeBSD in arm/filter_neon.S (Maya Rashish).
5865 
5866 Version 1.6.31beta02 [July 8, 2017]
5867   Added instructions for disabling hardware optimizations in INSTALL.
5868   Added "--enable-hardware-optimizations" configuration flag to enable
5869     or disable all hardware optimizations with one flag.
5870 
5871 Version 1.6.31beta03 [July 9, 2017]
5872   Updated CMakeLists.txt to add INTEL_SSE and MIPS_MSA platforms.
5873   Changed "int" to "png_size_t" in intel/filter_sse2.c to prevent
5874     possible integer overflow (Bug report by John Bowler).
5875   Quieted "declaration after statement" warnings in intel/filter_sse2.c.
5876   Added scripts/makefile-linux-opt, which has hardware optimizations enabled.
5877 
5878 Version 1.6.31beta04 [July 11, 2017]
5879   Removed one of the GCC-7.1.0 'strict-overflow' warnings that result when
5880     integers appear on both sides of a compare.  Worked around the others by
5881     forcing the strict-overflow setting in the relevant functions to a level
5882     where they are not reported (John Bowler).
5883   Changed "FALL THROUGH" comments to "FALLTHROUGH" because GCC doesn't like
5884     the space.
5885   Worked around some C-style casts from (void*) because g++ 5.4.0 objects
5886     to them.
5887   Increased the buffer size for 'sprint' to pass the gcc 7.1.0 'sprint
5888     overflow' check that is on by default with -Wall -Wextra.
5889 
5890 Version 1.6.31beta05 [July 13, 2017]
5891   Added eXIf chunk support.
5892 
5893 Version 1.6.31beta06 [July 17, 2017]
5894   Added a minimal eXIf chunk (with Orientation and FocalLengthIn35mmFilm
5895     tags) to pngtest.png.
5896 
5897 Version 1.6.31beta07 [July 18, 2017]
5898   Revised the eXIf chunk in pngtest.png to fix "Bad IFD1 Directory" warning.
5899 
5900 Version 1.6.31rc01 [July 19, 2017]
5901   No changes.
5902 
5903 Version 1.6.31rc02 [July 25, 2017]
5904   Fixed typo in example.c (png_free_image should be png_image_free) (Bug
5905     report by John Smith)
5906 
5907 Version 1.6.31 [July 27, 2017]
5908   No changes.
5909 
5910 Version 1.6.32beta01 [July 31, 2017]
5911   Avoid possible NULL dereference in png_handle_eXIf when benign_errors
5912     are allowed. Avoid leaking the input buffer "eXIf_buf".
5913   Eliminated png_ptr->num_exif member from pngstruct.h and added num_exif
5914     to arguments for png_get_eXIf() and png_set_eXIf().
5915   Added calls to png_handle_eXIf(() in pngread.c and png_write_eXIf() in
5916     pngwrite.c, and made various other fixes to png_write_eXIf().
5917   Changed name of png_get_eXIF and png_set_eXIf() to png_get_eXIf_1() and
5918     png_set_eXIf_1(), respectively, to avoid breaking API compatibility
5919     with libpng-1.6.31.
5920 
5921 Version 1.6.32beta02 [August 1, 2017]
5922   Updated contrib/libtests/pngunknown.c with eXIf chunk.
5923 
5924 Version 1.6.32beta03 [August 2, 2017]
5925   Initialized btoa[] in pngstest.c
5926   Stop memory leak when returning from png_handle_eXIf() with an error
5927     (Bug report from the OSS-fuzz project).
5928 
5929 Version 1.6.32beta04 [August 2, 2017]
5930   Replaced local eXIf_buf with info_ptr-eXIf_buf in png_handle_eXIf().
5931   Update libpng.3 and libpng-manual.txt about eXIf functions.
5932 
5933 Version 1.6.32beta05 [August 2, 2017]
5934   Restored png_get_eXIf() and png_set_eXIf() to maintain API compatibility.
5935 
5936 Version 1.6.32beta06 [August 2, 2017]
5937   Removed png_get_eXIf_1() and png_set_eXIf_1().
5938 
5939 Version 1.6.32beta07 [August 3, 2017]
5940   Check length of all chunks except IDAT against user limit to fix an
5941     OSS-fuzz issue (Fixes CVE-2017-12652).
5942 
5943 Version 1.6.32beta08 [August 3, 2017]
5944   Check length of IDAT against maximum possible IDAT size, accounting
5945     for height, rowbytes, interlacing and zlib/deflate overhead.
5946   Restored png_get_eXIf_1() and png_set_eXIf_1(), because strlen(eXIf_buf)
5947     does not work (the eXIf chunk data can contain zeroes).
5948 
5949 Version 1.6.32beta09 [August 3, 2017]
5950   Require cmake-2.8.8 in CMakeLists.txt. Revised symlink creation,
5951     no longer using deprecated cmake LOCATION feature (Clifford Yapp).
5952   Fixed five-byte error in the calculation of IDAT maximum possible size.
5953   
5954 Version 1.6.32beta10 [August 5, 2017]
5955   Moved chunk-length check into a png_check_chunk_length() private
5956     function (Suggested by Max Stepin).
5957   Moved bad pngs from tests to contrib/libtests/crashers
5958   Moved testing of bad pngs into a separate tests/pngtest-badpngs script
5959   Added the --xfail (expected FAIL) option to pngtest.c. It writes XFAIL
5960     in the output but PASS for the libpng test.
5961   Require cmake-3.0.2 in CMakeLists.txt (Clifford Yapp).
5962   Fix "const" declaration info_ptr argument to png_get_eXIf_1() and the
5963     num_exif argument to png_get_eXIf_1() (Github Issue 171).
5964 
5965 Version 1.6.32beta11 [August 7, 2017]
5966   Added "eXIf" to "chunks_to_ignore[]" in png_set_keep_unknown_chunks().
5967   Added huge_IDAT.png and empty_ancillary_chunks.png to testpngs/crashers.
5968   Make pngtest --strict, --relax, --xfail options imply -m (multiple).
5969   Removed unused chunk_name parameter from png_check_chunk_length().
5970   Relocated setting free_me for eXIf data, to stop an OSS-fuzz leak.
5971   Initialize profile_header[] in png_handle_iCCP() to fix OSS-fuzz issue.
5972   Initialize png_ptr->row_buf[0] to 255 in png_read_row() to fix OSS-fuzz UMR.
5973   Attempt to fix a UMR in png_set_text_2() to fix OSS-fuzz issue.
5974   Increase minimum zlib stream from 9 to 14 in png_handle_iCCP(), to account
5975     for the minimum 'deflate' stream, and relocate the test to a point
5976     after the keyword has been read.
5977   Check that the eXIf chunk has at least 2 bytes and begins with "II" or "MM".
5978 
5979 Version 1.6.32rc01 [August 18, 2017]
5980   Added a set of "huge_xxxx_chunk.png" files to contrib/testpngs/crashers,
5981     one for each known chunk type, with length = 2GB-1.
5982   Check for 0 return from png_get_rowbytes() and added some (size_t) typecasts
5983     in contrib/pngminus/*.c to stop some Coverity issues (162705, 162706,
5984     and 162707).
5985   Renamed chunks in contrib/testpngs/crashers to avoid having files whose
5986     names differ only in case; this causes problems with some platforms
5987     (github issue #172).
5988 
5989 Version 1.6.32rc02 [August 22, 2017]
5990   Added contrib/oss-fuzz directory which contains files used by the oss-fuzz
5991     project (https://github.com/google/oss-fuzz/tree/master/projects/libpng).
5992 
5993 Version 1.6.32 [August 24, 2017]
5994   No changes.
5995 
5996 Version 1.6.33beta01 [August 28, 2017]
5997   Added PNGMINUS_UNUSED macro to contrib/pngminus/p*.c and added missing
5998     parenthesis in contrib/pngminus/pnm2png.c (bug report by Christian Hesse).
5999   Fixed off-by-one error in png_do_check_palette_indexes() (Bug report
6000     by Mick P., Source Forge Issue #269).
6001 
6002 Version 1.6.33beta02 [September 3, 2017]
6003   Initialize png_handler.row_ptr in contrib/oss-fuzz/libpng_read_fuzzer.cc
6004     to fix shortlived oss-fuzz issue 3234.
6005   Compute a larger limit on IDAT because some applications write a deflate
6006     buffer for each row (Bug report by Andrew Church).
6007   Use current date (DATE) instead of release-date (RDATE) in last
6008     changed date of contrib/oss-fuzz files.
6009   Enabled ARM support in CMakeLists.txt (Bernd Kuhls).
6010 
6011 Version 1.6.33beta03 [September 14, 2017]
6012   Fixed incorrect typecast of some arguments to png_malloc() and
6013     png_calloc() that were png_uint_32 instead of png_alloc_size_t
6014     (Bug report by "irwir" in Github libpng issue #175).
6015   Use pnglibconf.h.prebuilt when building for ANDROID with cmake (Github
6016     issue 162, by rcdailey).
6017 
6018 Version 1.6.33rc01 [September 20, 2017]
6019   Initialize memory allocated by png_inflate to zero, using memset, to
6020     stop an oss-fuzz "use of uninitialized value" detection in png_set_text_2()
6021     due to truncated iTXt or zTXt chunk.
6022   Initialize memory allocated by png_read_buffer to zero, using memset, to
6023     stop an oss-fuzz "use of uninitialized value" detection in
6024     png_icc_check_tag_table() due to truncated iCCP chunk.
6025   Removed a redundant test (suggested by "irwir" in Github issue #180).
6026 
6027 Version 1.6.33rc02 [September 23, 2017]
6028   Added an interlaced version of each file in contrib/pngsuite.
6029   Relocate new memset() call in pngrutil.c.
6030   Removed more redundant tests (suggested by "irwir" in Github issue #180).
6031   Add support for loading images with associated alpha in the Simplified
6032     API (Samuel Williams).
6033 
6034 Version 1.6.33 [September 28, 2017]
6035   Revert contrib/oss-fuzz/libpng_read_fuzzer.cc to libpng-1.6.32 state.
6036   Initialize png_handler.row_ptr in contrib/oss-fuzz/libpng_read_fuzzer.cc
6037   Add end_info structure and png_read_end() to the libpng fuzzer.
6038 
6039 Version 1.6.34 [September 29, 2017]
6040   Removed contrib/pngsuite/i*.png; some of them caused test failures.
6041 
6042 Version 1.6.35beta01 [March 6, 2018]
6043   Restored 21 of the contrib/pngsuite/i*.png, which do not cause test
6044     failures. Placed the remainder in contrib/pngsuite/interlaced/i*.png.
6045   Added calls to png_set_*() transforms commonly used by browsers to
6046     the fuzzer.
6047   Removed some unnecessary brackets in pngrtran.c
6048   Fixed miscellaneous typos (Patch by github user "luzpaz").
6049   Change "ASM C" to "C ASM" in CMakeLists.txt
6050   Fixed incorrect handling of bKGD chunk in sub-8-bit files (Cosmin)
6051   Added hardware optimization directories to zip and 7z distributions.
6052   Fixed incorrect bitmask for options.
6053   Fixed many spelling typos.
6054 
6055 Version 1.6.35beta02 [March 28, 2018]
6056   Make png_get_iCCP consistent with man page (allow compression-type argument
6057   to be NULL, bug report by Lenard Szolnoki).
6058 
6059 Version 1.6.35 [July 15, 2018]
6060   Replaced the remaining uses of png_size_t with size_t (Cosmin)
6061   Fixed the calculation of row_factor in png_check_chunk_length
6062     (reported by Thuan Pham in SourceForge issue #278)
6063   Added missing parentheses to a macro definition
6064     (suggested by "irwir" in GitHub issue #216)
6065 
6066 Send comments/corrections/commendations to png-mng-implement at lists.sf.net
6067 (subscription required; visit
6068 https://lists.sourceforge.net/lists/listinfo/png-mng-implement
6069 to subscribe).
6070 
6071 Glenn R-P