9 - Allow 0 as memory usage limit instead of returning
10 LZMA_PROG_ERROR. Now 0 is treated as if 1 byte was specified,
11 which effectively is the same as 0.
13 - Use "noexcept" keyword instead of "throw()" in the public
14 headers when a C++11 (or newer standard) compiler is used.
16 - Added a portability fix for recent Intel C Compilers.
18 - Microsoft Visual Studio build files have been moved under
19 windows/vs2013 and windows/vs2017.
23 - Fix "xz --list --robot missing_or_bad_file.xz" which would
24 try to print an unitialized string and thus produce garbage
25 output. Since the exit status is non-zero, most uses of such
26 a command won't try to interpret the garbage output.
28 - "xz --list foo.xz" could print "Internal error (bug)" in a
29 corner case where a specific memory usage limit had been set.
36 - Always close a file before trying to delete it to avoid
37 problems on some operating system and file system combinations.
39 - Fixed copying of file timestamps on Windows.
41 - Added experimental (disabled by default) sandbox support using
42 Capsicum (FreeBSD >= 10). See --enable-sandbox in INSTALL.
44 * C99/C11 conformance fixes to liblzma. The issues affected at least
45 some builds using link-time optimizations.
47 * Fixed bugs in the rarely-used function lzma_index_dup().
49 * Use of external SHA-256 code is now disabled by default.
50 It can still be enabled by passing --enable-external-sha256
51 to configure. The reasons to disable it by default (see INSTALL
54 - Some OS-specific SHA-256 implementations conflict with
55 OpenSSL and cause problems in programs that link against both
56 liblzma and libcrypto. At least FreeBSD 10 and MINIX 3.3.0
59 - The internal SHA-256 is faster than the SHA-256 code in
60 some operating systems.
62 * Changed CPU core count detection to use sched_getaffinity() on
63 GNU/Linux and GNU/kFreeBSD.
65 * Fixes to the build-system and xz to make xz buildable even when
66 encoders, decoders, or threading have been disabled from libilzma
67 using configure options. These fixes added two new #defines to
68 config.h: HAVE_ENCODERS and HAVE_DECODERS.
73 * Fixed bugs in QNX-specific code.
75 * Omitted the use of pipe2() even if it is available to avoid
76 portability issues with some old Linux and glibc combinations.
78 * Updated German translation.
80 * Added project files to build static and shared liblzma (not the
81 whole XZ Utils) with Visual Studio 2013 update 2 or later.
83 * Documented that threaded decompression hasn't been implemented
84 yet. A 5.2.0 NEWS entry describing multi-threading support had
85 incorrectly said "decompression" when it should have said
91 * Fixed a compression-ratio regression in fast mode of LZMA1 and
92 LZMA2. The bug is present in 5.1.4beta and 5.2.0 releases.
94 * Fixed a portability problem in xz that affected at least OpenBSD.
96 * Fixed xzdiff to be compatible with FreeBSD's mktemp which differs
97 from most other mktemp implementations.
99 * Changed CPU core count detection to use cpuset_getaffinity() on
107 * All fixes from 5.0.8
109 * liblzma: Fixed lzma_stream_encoder_mt_memusage() when a preset
112 * xzdiff: If mktemp isn't installed, mkdir will be used as
113 a fallback to create a temporary directory. Installing mktemp
114 is still recommended.
116 * Updated French, German, Italian, Polish, and Vietnamese
119 Summary of fixes and new features added in the 5.1.x development
124 - Added support for multi-threaded compression. See the
125 lzma_mt structure, lzma_stream_encoder_mt(), and
126 lzma_stream_encoder_mt_memusage() in <lzma/container.h>,
127 lzma_get_progress() in <lzma/base.h>, and lzma_cputhreads()
128 in <lzma/hardware.h> for details.
130 - Made the uses of lzma_allocator const correct.
132 - Added lzma_block_uncomp_encode() to create uncompressed
133 .xz Blocks using LZMA2 uncompressed chunks.
135 - Added support for LZMA_IGNORE_CHECK.
137 - A few speed optimizations were made.
139 - Added support for symbol versioning. It is enabled by default
140 on GNU/Linux, other GNU-based systems, and FreeBSD.
142 - liblzma (not the whole XZ Utils) should now be buildable
143 with MSVC 2013 update 2 or later using windows/config.h.
147 - Fixed a race condition in the signal handling. It was
148 possible that e.g. the first SIGINT didn't make xz exit
149 if reading or writing blocked and one had bad luck. The fix
150 is non-trivial, so as of writing it is unknown if it will be
151 backported to the v5.0 branch.
153 - Multi-threaded compression can be enabled with the
154 --threads (-T) option.
155 [Fixed: This originally said "decompression".]
157 - New command line options in xz: --single-stream,
158 --block-size=SIZE, --block-list=SIZES,
159 --flush-timeout=TIMEOUT, and --ignore-check.
161 - xz -lvv now shows the minimum xz version that is required to
162 decompress the file. Currently it is 5.0.0 for all supported
163 .xz files except files with empty LZMA2 streams require 5.0.2.
165 * xzdiff and xzgrep now support .lzo files if lzop is installed.
166 The .tzo suffix is also recognized as a shorthand for .tar.lzo.
169 5.1.4beta (2014-09-14)
171 * All fixes from 5.0.6
173 * liblzma: Fixed the use of presets in threaded encoder
176 * xz --block-list and --block-size can now be used together
177 in single-threaded mode. Previously the combination only
178 worked in multi-threaded mode.
180 * Added support for LZMA_IGNORE_CHECK to liblzma and made it
181 available in xz as --ignore-check.
183 * liblzma speed optimizations:
185 - Initialization of a new LZMA1 or LZMA2 encoder has been
186 optimized. (The speed of reinitializing an already-allocated
187 encoder isn't affected.) This helps when compressing many
188 small buffers with lzma_stream_buffer_encode() and other
189 similar situations where an already-allocated encoder state
190 isn't reused. This speed-up is visible in xz too if one
191 compresses many small files one at a time instead running xz
192 once and giving all files as command-line arguments.
194 - Buffer comparisons are now much faster when unaligned access
195 is allowed (configured with --enable-unaligned-access). This
196 speeds up encoding significantly. There is arch-specific code
197 for 32-bit and 64-bit x86 (32-bit needs SSE2 for the best
198 results and there's no run-time CPU detection for now).
199 For other archs there is only generic code which probably
200 isn't as optimal as arch-specific solutions could be.
202 - A few speed optimizations were made to the SHA-256 code.
203 (Note that the builtin SHA-256 code isn't used on all
206 * liblzma can now be built with MSVC 2013 update 2 or later
207 using windows/config.h.
209 * Vietnamese translation was added.
212 5.1.3alpha (2013-10-26)
214 * All fixes from 5.0.5
218 - Fixed a deadlock in the threaded encoder.
220 - Made the uses of lzma_allocator const correct.
222 - Added lzma_block_uncomp_encode() to create uncompressed
223 .xz Blocks using LZMA2 uncompressed chunks.
225 - Added support for native threads on Windows and the ability
226 to detect the number of CPU cores.
230 - Fixed a race condition in the signal handling. It was
231 possible that e.g. the first SIGINT didn't make xz exit
232 if reading or writing blocked and one had bad luck. The fix
233 is non-trivial, so as of writing it is unknown if it will be
234 backported to the v5.0 branch.
236 - Made the progress indicator work correctly in threaded mode.
238 - Threaded encoder now works together with --block-list=SIZES.
240 - Added preliminary support for --flush-timeout=TIMEOUT.
241 It can be useful for (somewhat) real-time streaming. For
242 now the decompression side has to be done with something
243 else than the xz tool due to how xz does buffering, but this
247 5.1.2alpha (2012-07-04)
249 * All fixes from 5.0.3 and 5.0.4
253 - Fixed a deadlock and an invalid free() in the threaded encoder.
255 - Added support for symbol versioning. It is enabled by default
256 on GNU/Linux, other GNU-based systems, and FreeBSD.
258 - Use SHA-256 implementation from the operating system if one is
259 available in libc, libmd, or libutil. liblzma won't use e.g.
260 OpenSSL or libgcrypt to avoid introducing new dependencies.
262 - Fixed liblzma.pc for static linking.
264 - Fixed a few portability bugs.
266 * xz --decompress --single-stream now fixes the input position after
267 successful decompression. Now the following works:
269 echo foo | xz > foo.xz
270 echo bar | xz >> foo.xz
271 ( xz -dc --single-stream ; xz -dc --single-stream ) < foo.xz
273 Note that it doesn't work if the input is not seekable
274 or if there is Stream Padding between the concatenated
277 * xz -lvv now shows the minimum xz version that is required to
278 decompress the file. Currently it is 5.0.0 for all supported .xz
279 files except files with empty LZMA2 streams require 5.0.2.
281 * Added an *incomplete* implementation of --block-list=SIZES to xz.
282 It only works correctly in single-threaded mode and when
283 --block-size isn't used at the same time. --block-list allows
284 specifying the sizes of Blocks which can be useful e.g. when
285 creating files for random-access reading.
288 5.1.1alpha (2011-04-12)
290 * All fixes from 5.0.2
292 * liblzma fixes that will also be included in 5.0.3:
294 - A memory leak was fixed.
296 - lzma_stream_buffer_encode() no longer creates an empty .xz
297 Block if encoding an empty buffer. Such an empty Block with
298 LZMA2 data would trigger a bug in 5.0.1 and older (see the
299 first bullet point in 5.0.2 notes). When releasing 5.0.2,
300 I thought that no encoder creates this kind of files but
303 - Validate function arguments better in a few functions. Most
304 importantly, specifying an unsupported integrity check to
305 lzma_stream_buffer_encode() no longer creates a corrupt .xz
306 file. Probably no application tries to do that, so this
307 shouldn't be a big problem in practice.
309 - Document that lzma_block_buffer_encode(),
310 lzma_easy_buffer_encode(), lzma_stream_encoder(), and
311 lzma_stream_buffer_encode() may return LZMA_UNSUPPORTED_CHECK.
313 - The return values of the _memusage() functions are now
316 * Support for multithreaded compression was added using the simplest
317 method, which splits the input data into blocks and compresses
318 them independently. Other methods will be added in the future.
319 The current method has room for improvement, e.g. it is possible
320 to reduce the memory usage.
322 * Added the options --single-stream and --block-size=SIZE to xz.
324 * xzdiff and xzgrep now support .lzo files if lzop is installed.
325 The .tzo suffix is also recognized as a shorthand for .tar.lzo.
327 * Support for short 8.3 filenames under DOS was added to xz. It is
328 experimental and may change before it gets into a stable release.
333 * Fixed an old bug in xzgrep that affected OpenBSD and probably
334 a few other operating systems too.
336 * Updated French and German translations.
338 * Added support for detecting the amount of RAM on AmigaOS/AROS.
340 * Minor build system updates.
345 * Fix regressions introduced in 5.0.6:
347 - Fix building with non-GNU make.
349 - Fix invalid Libs.private value in liblzma.pc which broke
350 static linking against liblzma if the linker flags were
351 taken from pkg-config.
356 * xzgrep now exits with status 0 if at least one file matched.
358 * A few minor portability and build system fixes
363 * lzmadec and liblzma's lzma_alone_decoder(): Support decompressing
364 .lzma files that have less common settings in the headers
365 (dictionary size other than 2^n or 2^n + 2^(n-1), or uncompressed
366 size greater than 256 GiB). The limitations existed to avoid false
367 positives when detecting .lzma files. The lc + lp <= 4 limitation
368 still remains since liblzma's LZMA decoder has that limitation.
370 NOTE: xz's .lzma support or liblzma's lzma_auto_decoder() are NOT
371 affected by this change. They still consider uncommon .lzma headers
372 as not being in the .lzma format. Changing this would give way too
373 many false positives.
377 - Interaction of preset and custom filter chain options was
378 made less illogical. This affects only certain less typical
379 uses cases so few people are expected to notice this change.
381 Now when a custom filter chain option (e.g. --lzma2) is
382 specified, all preset options (-0 ... -9, -e) earlier are on
383 the command line are completely forgotten. Similarly, when
384 a preset option is specified, all custom filter chain options
385 earlier on the command line are completely forgotten.
387 Example 1: "xz -9 --lzma2=preset=5 -e" is equivalent to "xz -e"
388 which is equivalent to "xz -6e". Earlier -e didn't put xz back
389 into preset mode and thus the example command was equivalent
390 to "xz --lzma2=preset=5".
392 Example 2: "xz -9e --lzma2=preset=5 -7" is equivalent to
393 "xz -7". Earlier a custom filter chain option didn't make
394 xz forget the -e option so the example was equivalent to
397 - Fixes and improvements to error handling.
399 - Various fixes to the man page.
401 * xzless: Fixed to work with "less" versions 448 and later.
403 * xzgrep: Made -h an alias for --no-filename.
405 * Include the previously missing debug/translation.bash which can
406 be useful for translators.
408 * Include a build script for Mac OS X. This has been in the Git
409 repository since 2010 but due to a mistake in Makefile.am the
410 script hasn't been included in a release tarball before.
417 - Fix lzma_index_init(). It could crash if memory allocation
420 - Fix the possibility of an incorrect LZMA_BUF_ERROR when a BCJ
421 filter is used and the application only provides exactly as
422 much output space as is the uncompressed size of the file.
424 - Fix a bug in doc/examples_old/xz_pipe_decompress.c. It didn't
425 check if the last call to lzma_code() really returned
426 LZMA_STREAM_END, which made the program think that truncated
429 - New example programs in doc/examples (old programs are now in
430 doc/examples_old). These have more comments and more detailed
433 * Fix "xz -lvv foo.xz". It could crash on some corrupted files.
435 * Fix output of "xz --robot -lv" and "xz --robot -lvv" which
436 incorrectly printed the filename also in the "foo (x/x)" format.
438 * Fix exit status of "xzdiff foo.xz bar.xz".
440 * Fix exit status of "xzgrep foo binary_file".
442 * Fix portability to EBCDIC systems.
444 * Fix a configure issue on AIX with the XL C compiler. See INSTALL
447 * Update French, German, Italian, and Polish translations.
454 - A memory leak was fixed.
456 - lzma_stream_buffer_encode() no longer creates an empty .xz
457 Block if encoding an empty buffer. Such an empty Block with
458 LZMA2 data would trigger a bug in 5.0.1 and older (see the
459 first bullet point in 5.0.2 notes). When releasing 5.0.2,
460 I thought that no encoder creates this kind of files but
463 - Validate function arguments better in a few functions. Most
464 importantly, specifying an unsupported integrity check to
465 lzma_stream_buffer_encode() no longer creates a corrupt .xz
466 file. Probably no application tries to do that, so this
467 shouldn't be a big problem in practice.
469 - Document that lzma_block_buffer_encode(),
470 lzma_easy_buffer_encode(), lzma_stream_encoder(), and
471 lzma_stream_buffer_encode() may return LZMA_UNSUPPORTED_CHECK.
473 - The return values of the _memusage() functions are now
476 * Fix command name detection in xzgrep. xzegrep and xzfgrep now
477 correctly use egrep and fgrep instead of grep.
479 * French translation was added.
484 * LZMA2 decompressor now correctly accepts LZMA2 streams with no
485 uncompressed data. Previously it considered them corrupt. The
486 bug can affect applications that use raw LZMA2 streams. It is
487 very unlikely to affect .xz files because no compressor creates
488 .xz files with empty LZMA2 streams. (Empty .xz files are a
489 different thing than empty LZMA2 streams.)
491 * "xz --suffix=.foo filename.foo" now refuses to compress the
492 file due to it already having the suffix .foo. It was already
493 documented on the man page, but the code lacked the test.
495 * "xzgrep -l foo bar.xz" works now.
497 * Polish translation was added.
502 * xz --force now (de)compresses files that have setuid, setgid,
503 or sticky bit set and files that have multiple hard links.
504 The man page had it documented this way already, but the code
507 * gzip and bzip2 support in xzdiff was fixed.
511 * Minor fix to Czech translation
516 Only the most important changes compared to 4.999.9beta are listed
517 here. One change is especially important:
519 * The memory usage limit is now disabled by default. Some scripts
520 written before this change may have used --memory=max on xz command
521 line or in XZ_OPT. THESE USES OF --memory=max SHOULD BE REMOVED
522 NOW, because they interfere with user's ability to set the memory
523 usage limit himself. If user-specified limit causes problems to
524 your script, blame the user.
526 Other significant changes:
528 * Added support for XZ_DEFAULTS environment variable. This variable
529 allows users to set default options for xz, e.g. default memory
530 usage limit or default compression level. Scripts that use xz
531 must never set or unset XZ_DEFAULTS. Scripts should use XZ_OPT
532 instead if they need a way to pass options to xz via an
533 environment variable.
535 * The compression settings associated with the preset levels
536 -0 ... -9 have been changed. --extreme was changed a little too.
537 It is now less likely to make compression worse, but with some
538 files the new --extreme may compress slightly worse than the old
541 * If a preset level (-0 ... -9) is specified after a custom filter
542 chain options have been used (e.g. --lzma2), the custom filter
543 chain will be forgotten. Earlier the preset options were
544 completely ignored after custom filter chain options had been
547 * xz will create sparse files when decompressing if the uncompressed
548 data contains long sequences of binary zeros. This is done even
549 when writing to standard output that is connected to a regular
550 file and certain additional conditions are met to make it safe.
552 * Support for "xz --list" was added. Combine with --verbose or
553 --verbose --verbose (-vv) for detailed output.
555 * I had hoped that liblzma API would have been stable after
556 4.999.9beta, but there have been a couple of changes in the
557 advanced features, which don't affect most applications:
559 - Index handling code was revised. If you were using the old
560 API, you will get a compiler error (so it's easy to notice).
562 - A subtle but important change was made to the Block handling
563 API. lzma_block.version has to be initialized even for
564 lzma_block_header_decode(). Code that doesn't do it will work
565 for now, but might break in the future, which makes this API
568 * The major soname has been bumped to 5.0.0. liblzma API and ABI
569 are now stable, so the need to recompile programs linking against
570 liblzma shouldn't arise soon.