7 * Fixed bugs in QNX-specific code.
9 * Omitted the use of pipe2() even if it is available to avoid
10 portability issues with some old Linux and glibc combinations.
12 * Updated German translation.
14 * Added project files to build static and shared liblzma (not the
15 whole XZ Utils) with Visual Studio 2013 update 2 or later.
17 * Documented that threaded decompression hasn't been implemented
18 yet. A 5.2.0 NEWS entry describing multi-threading support had
19 incorrectly said "decompression" when it should have said
25 * Fixed a compression-ratio regression in fast mode of LZMA1 and
26 LZMA2. The bug is present in 5.1.4beta and 5.2.0 releases.
28 * Fixed a portability problem in xz that affected at least OpenBSD.
30 * Fixed xzdiff to be compatible with FreeBSD's mktemp which differs
31 from most other mktemp implementations.
33 * Changed CPU core count detection to use cpuset_getaffinity() on
41 * All fixes from 5.0.8
43 * liblzma: Fixed lzma_stream_encoder_mt_memusage() when a preset
46 * xzdiff: If mktemp isn't installed, mkdir will be used as
47 a fallback to create a temporary directory. Installing mktemp
50 * Updated French, German, Italian, Polish, and Vietnamese
53 Summary of fixes and new features added in the 5.1.x development
58 - Added support for multi-threaded compression. See the
59 lzma_mt structure, lzma_stream_encoder_mt(), and
60 lzma_stream_encoder_mt_memusage() in <lzma/container.h>,
61 lzma_get_progress() in <lzma/base.h>, and lzma_cputhreads()
62 in <lzma/hardware.h> for details.
64 - Made the uses of lzma_allocator const correct.
66 - Added lzma_block_uncomp_encode() to create uncompressed
67 .xz Blocks using LZMA2 uncompressed chunks.
69 - Added support for LZMA_IGNORE_CHECK.
71 - A few speed optimizations were made.
73 - Added support for symbol versioning. It is enabled by default
74 on GNU/Linux, other GNU-based systems, and FreeBSD.
76 - liblzma (not the whole XZ Utils) should now be buildable
77 with MSVC 2013 update 2 or later using windows/config.h.
81 - Fixed a race condition in the signal handling. It was
82 possible that e.g. the first SIGINT didn't make xz exit
83 if reading or writing blocked and one had bad luck. The fix
84 is non-trivial, so as of writing it is unknown if it will be
85 backported to the v5.0 branch.
87 - Multi-threaded compression can be enabled with the
88 --threads (-T) option.
89 [Fixed: This originally said "decompression".]
91 - New command line options in xz: --single-stream,
92 --block-size=SIZE, --block-list=SIZES,
93 --flush-timeout=TIMEOUT, and --ignore-check.
95 - xz -lvv now shows the minimum xz version that is required to
96 decompress the file. Currently it is 5.0.0 for all supported
97 .xz files except files with empty LZMA2 streams require 5.0.2.
99 * xzdiff and xzgrep now support .lzo files if lzop is installed.
100 The .tzo suffix is also recognized as a shorthand for .tar.lzo.
103 5.1.4beta (2014-09-14)
105 * All fixes from 5.0.6
107 * liblzma: Fixed the use of presets in threaded encoder
110 * xz --block-list and --block-size can now be used together
111 in single-threaded mode. Previously the combination only
112 worked in multi-threaded mode.
114 * Added support for LZMA_IGNORE_CHECK to liblzma and made it
115 available in xz as --ignore-check.
117 * liblzma speed optimizations:
119 - Initialization of a new LZMA1 or LZMA2 encoder has been
120 optimized. (The speed of reinitializing an already-allocated
121 encoder isn't affected.) This helps when compressing many
122 small buffers with lzma_stream_buffer_encode() and other
123 similar situations where an already-allocated encoder state
124 isn't reused. This speed-up is visible in xz too if one
125 compresses many small files one at a time instead running xz
126 once and giving all files as command-line arguments.
128 - Buffer comparisons are now much faster when unaligned access
129 is allowed (configured with --enable-unaligned-access). This
130 speeds up encoding significantly. There is arch-specific code
131 for 32-bit and 64-bit x86 (32-bit needs SSE2 for the best
132 results and there's no run-time CPU detection for now).
133 For other archs there is only generic code which probably
134 isn't as optimal as arch-specific solutions could be.
136 - A few speed optimizations were made to the SHA-256 code.
137 (Note that the builtin SHA-256 code isn't used on all
140 * liblzma can now be built with MSVC 2013 update 2 or later
141 using windows/config.h.
143 * Vietnamese translation was added.
146 5.1.3alpha (2013-10-26)
148 * All fixes from 5.0.5
152 - Fixed a deadlock in the threaded encoder.
154 - Made the uses of lzma_allocator const correct.
156 - Added lzma_block_uncomp_encode() to create uncompressed
157 .xz Blocks using LZMA2 uncompressed chunks.
159 - Added support for native threads on Windows and the ability
160 to detect the number of CPU cores.
164 - Fixed a race condition in the signal handling. It was
165 possible that e.g. the first SIGINT didn't make xz exit
166 if reading or writing blocked and one had bad luck. The fix
167 is non-trivial, so as of writing it is unknown if it will be
168 backported to the v5.0 branch.
170 - Made the progress indicator work correctly in threaded mode.
172 - Threaded encoder now works together with --block-list=SIZES.
174 - Added preliminary support for --flush-timeout=TIMEOUT.
175 It can be useful for (somewhat) real-time streaming. For
176 now the decompression side has to be done with something
177 else than the xz tool due to how xz does buffering, but this
181 5.1.2alpha (2012-07-04)
183 * All fixes from 5.0.3 and 5.0.4
187 - Fixed a deadlock and an invalid free() in the threaded encoder.
189 - Added support for symbol versioning. It is enabled by default
190 on GNU/Linux, other GNU-based systems, and FreeBSD.
192 - Use SHA-256 implementation from the operating system if one is
193 available in libc, libmd, or libutil. liblzma won't use e.g.
194 OpenSSL or libgcrypt to avoid introducing new dependencies.
196 - Fixed liblzma.pc for static linking.
198 - Fixed a few portability bugs.
200 * xz --decompress --single-stream now fixes the input position after
201 successful decompression. Now the following works:
203 echo foo | xz > foo.xz
204 echo bar | xz >> foo.xz
205 ( xz -dc --single-stream ; xz -dc --single-stream ) < foo.xz
207 Note that it doesn't work if the input is not seekable
208 or if there is Stream Padding between the concatenated
211 * xz -lvv now shows the minimum xz version that is required to
212 decompress the file. Currently it is 5.0.0 for all supported .xz
213 files except files with empty LZMA2 streams require 5.0.2.
215 * Added an *incomplete* implementation of --block-list=SIZES to xz.
216 It only works correctly in single-threaded mode and when
217 --block-size isn't used at the same time. --block-list allows
218 specifying the sizes of Blocks which can be useful e.g. when
219 creating files for random-access reading.
222 5.1.1alpha (2011-04-12)
224 * All fixes from 5.0.2
226 * liblzma fixes that will also be included in 5.0.3:
228 - A memory leak was fixed.
230 - lzma_stream_buffer_encode() no longer creates an empty .xz
231 Block if encoding an empty buffer. Such an empty Block with
232 LZMA2 data would trigger a bug in 5.0.1 and older (see the
233 first bullet point in 5.0.2 notes). When releasing 5.0.2,
234 I thought that no encoder creates this kind of files but
237 - Validate function arguments better in a few functions. Most
238 importantly, specifying an unsupported integrity check to
239 lzma_stream_buffer_encode() no longer creates a corrupt .xz
240 file. Probably no application tries to do that, so this
241 shouldn't be a big problem in practice.
243 - Document that lzma_block_buffer_encode(),
244 lzma_easy_buffer_encode(), lzma_stream_encoder(), and
245 lzma_stream_buffer_encode() may return LZMA_UNSUPPORTED_CHECK.
247 - The return values of the _memusage() functions are now
250 * Support for multithreaded compression was added using the simplest
251 method, which splits the input data into blocks and compresses
252 them independently. Other methods will be added in the future.
253 The current method has room for improvement, e.g. it is possible
254 to reduce the memory usage.
256 * Added the options --single-stream and --block-size=SIZE to xz.
258 * xzdiff and xzgrep now support .lzo files if lzop is installed.
259 The .tzo suffix is also recognized as a shorthand for .tar.lzo.
261 * Support for short 8.3 filenames under DOS was added to xz. It is
262 experimental and may change before it gets into a stable release.
267 * Fixed an old bug in xzgrep that affected OpenBSD and probably
268 a few other operating systems too.
270 * Updated French and German translations.
272 * Added support for detecting the amount of RAM on AmigaOS/AROS.
274 * Minor build system updates.
279 * Fix regressions introduced in 5.0.6:
281 - Fix building with non-GNU make.
283 - Fix invalid Libs.private value in liblzma.pc which broke
284 static linking against liblzma if the linker flags were
285 taken from pkg-config.
290 * xzgrep now exits with status 0 if at least one file matched.
292 * A few minor portability and build system fixes
297 * lzmadec and liblzma's lzma_alone_decoder(): Support decompressing
298 .lzma files that have less common settings in the headers
299 (dictionary size other than 2^n or 2^n + 2^(n-1), or uncompressed
300 size greater than 256 GiB). The limitations existed to avoid false
301 positives when detecting .lzma files. The lc + lp <= 4 limitation
302 still remains since liblzma's LZMA decoder has that limitation.
304 NOTE: xz's .lzma support or liblzma's lzma_auto_decoder() are NOT
305 affected by this change. They still consider uncommon .lzma headers
306 as not being in the .lzma format. Changing this would give way too
307 many false positives.
311 - Interaction of preset and custom filter chain options was
312 made less illogical. This affects only certain less typical
313 uses cases so few people are expected to notice this change.
315 Now when a custom filter chain option (e.g. --lzma2) is
316 specified, all preset options (-0 ... -9, -e) earlier are on
317 the command line are completely forgotten. Similarly, when
318 a preset option is specified, all custom filter chain options
319 earlier on the command line are completely forgotten.
321 Example 1: "xz -9 --lzma2=preset=5 -e" is equivalent to "xz -e"
322 which is equivalent to "xz -6e". Earlier -e didn't put xz back
323 into preset mode and thus the example command was equivalent
324 to "xz --lzma2=preset=5".
326 Example 2: "xz -9e --lzma2=preset=5 -7" is equivalent to
327 "xz -7". Earlier a custom filter chain option didn't make
328 xz forget the -e option so the example was equivalent to
331 - Fixes and improvements to error handling.
333 - Various fixes to the man page.
335 * xzless: Fixed to work with "less" versions 448 and later.
337 * xzgrep: Made -h an alias for --no-filename.
339 * Include the previously missing debug/translation.bash which can
340 be useful for translators.
342 * Include a build script for Mac OS X. This has been in the Git
343 repository since 2010 but due to a mistake in Makefile.am the
344 script hasn't been included in a release tarball before.
351 - Fix lzma_index_init(). It could crash if memory allocation
354 - Fix the possibility of an incorrect LZMA_BUF_ERROR when a BCJ
355 filter is used and the application only provides exactly as
356 much output space as is the uncompressed size of the file.
358 - Fix a bug in doc/examples_old/xz_pipe_decompress.c. It didn't
359 check if the last call to lzma_code() really returned
360 LZMA_STREAM_END, which made the program think that truncated
363 - New example programs in doc/examples (old programs are now in
364 doc/examples_old). These have more comments and more detailed
367 * Fix "xz -lvv foo.xz". It could crash on some corrupted files.
369 * Fix output of "xz --robot -lv" and "xz --robot -lvv" which
370 incorrectly printed the filename also in the "foo (x/x)" format.
372 * Fix exit status of "xzdiff foo.xz bar.xz".
374 * Fix exit status of "xzgrep foo binary_file".
376 * Fix portability to EBCDIC systems.
378 * Fix a configure issue on AIX with the XL C compiler. See INSTALL
381 * Update French, German, Italian, and Polish translations.
388 - A memory leak was fixed.
390 - lzma_stream_buffer_encode() no longer creates an empty .xz
391 Block if encoding an empty buffer. Such an empty Block with
392 LZMA2 data would trigger a bug in 5.0.1 and older (see the
393 first bullet point in 5.0.2 notes). When releasing 5.0.2,
394 I thought that no encoder creates this kind of files but
397 - Validate function arguments better in a few functions. Most
398 importantly, specifying an unsupported integrity check to
399 lzma_stream_buffer_encode() no longer creates a corrupt .xz
400 file. Probably no application tries to do that, so this
401 shouldn't be a big problem in practice.
403 - Document that lzma_block_buffer_encode(),
404 lzma_easy_buffer_encode(), lzma_stream_encoder(), and
405 lzma_stream_buffer_encode() may return LZMA_UNSUPPORTED_CHECK.
407 - The return values of the _memusage() functions are now
410 * Fix command name detection in xzgrep. xzegrep and xzfgrep now
411 correctly use egrep and fgrep instead of grep.
413 * French translation was added.
418 * LZMA2 decompressor now correctly accepts LZMA2 streams with no
419 uncompressed data. Previously it considered them corrupt. The
420 bug can affect applications that use raw LZMA2 streams. It is
421 very unlikely to affect .xz files because no compressor creates
422 .xz files with empty LZMA2 streams. (Empty .xz files are a
423 different thing than empty LZMA2 streams.)
425 * "xz --suffix=.foo filename.foo" now refuses to compress the
426 file due to it already having the suffix .foo. It was already
427 documented on the man page, but the code lacked the test.
429 * "xzgrep -l foo bar.xz" works now.
431 * Polish translation was added.
436 * xz --force now (de)compresses files that have setuid, setgid,
437 or sticky bit set and files that have multiple hard links.
438 The man page had it documented this way already, but the code
441 * gzip and bzip2 support in xzdiff was fixed.
445 * Minor fix to Czech translation
450 Only the most important changes compared to 4.999.9beta are listed
451 here. One change is especially important:
453 * The memory usage limit is now disabled by default. Some scripts
454 written before this change may have used --memory=max on xz command
455 line or in XZ_OPT. THESE USES OF --memory=max SHOULD BE REMOVED
456 NOW, because they interfere with user's ability to set the memory
457 usage limit himself. If user-specified limit causes problems to
458 your script, blame the user.
460 Other significant changes:
462 * Added support for XZ_DEFAULTS environment variable. This variable
463 allows users to set default options for xz, e.g. default memory
464 usage limit or default compression level. Scripts that use xz
465 must never set or unset XZ_DEFAULTS. Scripts should use XZ_OPT
466 instead if they need a way to pass options to xz via an
467 environment variable.
469 * The compression settings associated with the preset levels
470 -0 ... -9 have been changed. --extreme was changed a little too.
471 It is now less likely to make compression worse, but with some
472 files the new --extreme may compress slightly worse than the old
475 * If a preset level (-0 ... -9) is specified after a custom filter
476 chain options have been used (e.g. --lzma2), the custom filter
477 chain will be forgotten. Earlier the preset options were
478 completely ignored after custom filter chain options had been
481 * xz will create sparse files when decompressing if the uncompressed
482 data contains long sequences of binary zeros. This is done even
483 when writing to standard output that is connected to a regular
484 file and certain additional conditions are met to make it safe.
486 * Support for "xz --list" was added. Combine with --verbose or
487 --verbose --verbose (-vv) for detailed output.
489 * I had hoped that liblzma API would have been stable after
490 4.999.9beta, but there have been a couple of changes in the
491 advanced features, which don't affect most applications:
493 - Index handling code was revised. If you were using the old
494 API, you will get a compiler error (so it's easy to notice).
496 - A subtle but important change was made to the Block handling
497 API. lzma_block.version has to be initialized even for
498 lzma_block_header_decode(). Code that doesn't do it will work
499 for now, but might break in the future, which makes this API
502 * The major soname has been bumped to 5.0.0. liblzma API and ABI
503 are now stable, so the need to recompile programs linking against
504 liblzma shouldn't arise soon.