TIFF CHANGE INFORMATION
This document describes the changes made to the software between the
previous and current versions (see above).
If you don't find something listed here, then it was not done in this
timeframe, or it was not considered important enough to be mentioned.
The following information is located here:
CHANGES IN THE SOFTWARE CONFIGURATION:
- libtiff/libtiff.def: Brent Roman submitted new version adding
serveral missing entry points. Also add a few other entry points
later.
- configure, Makefile.in, etc: added support for OPTIMIZER being
set from config.site.
- config.guess: updated wholesale to an FSF version apparently
from 1998 (as opposed to 1994). This is mainly inspired by
providing for MacOS X support.
- configure/config.site: modified to check if -lm is needed for
MACHDEPLIBS if not supplied by config.site. Needed for Darwin.
- libtiff/tiff.h: Applied hac to try and resolve the problem
with the inttypes.h include file on AIX. (Bug 39)
- configure, *Makefile.in: Various changes to improve configuration
for HP/UX specifically, and also in general. (Bug 40) They include:
- Try to handle /usr/bin/sh instead of /bin/sh where necessary.
- Upgrade to HP/UX 10.x+ compiler, linker and dso options.
- Fixed mmap() test to avoid MMAP_FIXED ... it isn't available on HP
- Use -${MAKEFLAGS} in sub makes from makefiles.
- Fixed SCRIPT_SH/SHELL handling.
- configure: Changes for DSO generation on AIX provided by
John Marquart .
- configure, libtiff/Makefile.in: Modified to build DSOs properly
on Darwin thanks to Robert Krajewski ([email protected]) and
Keisuke Fujii ([email protected]).
- configure, libtiff/Makefile.in: applied OpenBSD patches as per bug 61.
- Makefile.in: added DESTDIR support as per bug 60.
- libtiff/tif_jpeg.c: Define HAVE_BOOLEAN on windows if RPCNDR.H
has been included.
- man/Makefile.in: add TIFFClientOpen link as per debian submitted
bug 66.
- libtiff/Makefile.in: Fixed @DSOSUB_VERSION to be @DSOSUF_VERSION@
in two places.
CHANGES IN LIBTIFF:
- tif_fax3.c: keep rw_mode flag internal to fax3 state to remember
whether we are encoding or decoding. This is to ensure graceful
recovery if TIFFClientOpen() discovers an attempt to open a compressed
file for "r+" access, and subsequently close it, as it resets the
tif_mode flag to O_RDONLY in this case to avoid writes, confusing the
compressor's concept of whether it is in encode or decode mode.
- tif_luv.c/tiff.h/tiffio.h:
New version of TIFF LogLuv (SGILOG) modules contributed by Greg Ward
([email protected]). He writes:
- I improved the gamut-mapping function in tif_luv.c for imaginary
colors, because some images were being super-saturated on the input
side and this resulted in some strange color shifts in the output.
- I added a psuedotag in tiff.h to control random dithering during
LogLuv encoding. This is turned off by default for 32-bit LogLuv and
on for 24-bit LogLuv output. Dithering improves the average color
accuracy over the image.
- I added a #define for LOG_LUV_PUBLIC, which is enabled by default in
tiffio.h, to expose internal routines for converting between LogLuv and
XYZ coordinates. This is helpful for writing more efficient,
specialized conversion routines, especially for reading LogLuv files.
- libtiff/tif_dirinfo.c: don't declare tiffFieldInfo static on VMS.
- Added TIFFTAG_COPYRIGHT support.
- tif_getimage.c: Added support for 16bit minisblack/miniswhite
images in RGBA interface.
- libtiff/tif_dirinfo.c: removed duplicate TIFFTAG_PHOTOSHOP as per
bug 44.
- libtiff/tif_dirwrite.c: Added support for TIFF_VARIABLE2 in the
case of writing TIFF_BYTE/TIFF_SBYTE fields as per bug 43.
- libtiff/tif_dirinfo.c: Modified the TIFF_BYTE definition for
TIFFTAG_PHOTOSHOP to use a writecount of TIFF_VARIABLE2 (-3) to
force use of uint32 counts instead of short counts.
- libtiff/tif_dirinfo.c: moved pixar and copyright flags to
ensure everything is in order.
- Integrated experimental OJPEG support from Scott Marovich of HP.
- libtiff/tif_open.c: Seek back to zero after failed read,
before writing header.
- libtiff/tiff.h, libtiff/tif_fax3.c: added check for __LP64__
when checking for 64 bit architectures as per bugzilla bug 67.
- libtiff/tif_getimage.c: Use memmove() instead of TIFFmemcpy()
in TIFFReadRGBATile() to avoid issues in cases of overlapping
buffers. See Bug 69 in Bugzilla.
- libtiff/tif_getimage.c: Don't complain for CMYK (separated)
images with more than four samples per pixel as per bug 73.
- libtiff/tif_getimage.c: relax handling of contig case where
there are extra samples that are supposed to be ignored as per bug 75. This
should now work for 8bit greyscale or palletted images.
- libtiff/tif_packbits.c: fixed memory overrun error as per bug 77.
- libtiff/tif_getimage.c: Fixed problem with reading strips or
tiles that don't start on a tile boundary. Fix contributed by
Josep Vallverdu (from HP), and further described in bug 47.
- libtif/tif_fax3.c: Removed #ifdef PURIFY logic, and modified to
always use the "safe" version, even if there is a very slight
cost in performance as per bug 54.
- libtiff/tif_lzw.c: added dummy LZWSetupEncode() to report an
error about LZW not being available.
- libtiff/tif_dir.c: propagate failure to initialize compression
back from TIFFSetField() as an error status, so applications can
detect failure.
- libtiff/tif_lzw.c: Avoid MS VC++ 5.0 optimization bug as per bug 78.
- libtiff/tif_dirwrite.c: added TIFFRewriteDirectory() function.
Updated TIFFWriteDirectory man page to include TIFFRewriteDirectory.
- libtiff/tiff.h: I have created COMPRESSION_CCITT_T4,
COMPRESSION_CCITT_T6, TIFFTAG_T4OPTIONS and TIFFTAG_T6OPTIONS aliases
in keeping with TIFF 6.0 standard in tiff.h as per bug 83.
- Added PHOTOMETRIC_ITULAB as per bug 90.
CHANGES IN THE TOOLS:
- Brent Roman contributed updated tiffcp utility (and tiffcp.1)
with support for extracting subimages with the ,n syntax, and also
adding the -b bias removal flag.
- tiff2ps.c/tiff2ps.1: Substantial changes to tiff2ps by
Bruce A. Mallett, including a faster encoder, fixes for level
2 PostScript, and support for the imagemask operator.
- fax2ps.c: Helge (libtiff at oldach.net) submitted fix
that corrects behaviour for non-Letter paper
sizes. (Bug 35) It fixes two problems:
Without scaling (-S) the fax is now centered on the page size specified
with -H and/or -W. Before, fax2ps was using an obscure and practially
useless algorithm to allocate the image relative to Letter sized paper
which sometime sled to useless whitespace on the paper, while at the
same time cutting of the faxes printable area at the opposite border.
Second, scaling now preserves aspect ratio, which makes unusual faxes
(in particular short ones) print properly.
- thumbnail.c: changed default output compression
to packbits from LZW since LZW isn't generally available.
- tiff2rgba.c: added -n flag to avoid emitting alpha component. Also added
a man page for tiff2rgba.
- tiffcmp.c: Fixed multi samples per pixel support for ContigCompare
as per bug 53.
Updated bug section of tiffcmp.1 to note tiled file issues.
- libtiff/tif_getimage.c: Fixed so that failure is properly
reported by gtTileContig, gtStripContig, gtTileSeparate and
gtStripSeparate as per bug 51.
CHANGES IN THE LZW COMPRESSION KIT:
- Rewrote lzw patching process so that is required to enable full
LZW support is to drop the tif_lzw.c from the
libtiff-lzw-compression-kit over the one in the libtiff directory.
- Some changes were made to make recovery from failure to
initialize the LZW compressor more graceful.
- Note that as distributed libtiff support LZW decompression, but
not LZW compression.
CHANGES IN THE CONTRIB AREA:
- Fixed distribution to include contrib/addtiffo/tif_ovrcache.{c,h}.
TIFF home page.
Last updated $Date: 2016-09-25 20:05:45 $.