| blob | b9007671aa5b07f7840b1f750761e92550023234 |
2 XZ Utils Release Notes
3 ======================
5 5.4.4 (2023-08-02)
7 * liblzma and xzdec can now build against WASI SDK when threading
8 support is disabled. xz and tests don't build yet.
10 * CMake:
12 - Fixed a bug preventing other projects from including liblzma
13 multiple times using find_package().
15 - Don't create broken symlinks in Cygwin and MSYS2 unless
16 supported by the environment. This prevented building for the
17 default MSYS2 environment. The problem was introduced in
18 xz 5.4.0.
20 * Documentation:
22 - Small improvements to man pages.
24 - Small improvements and typo fixes for liblzma API
25 documentation.
27 * Tests:
29 - Added a new section to INSTALL to describe basic test usage
30 and address recent questions about building the tests when
31 cross compiling.
33 - Small fixes and improvements to the tests.
35 * Translations:
37 - Fixed a mistake that caused one of the error messages to not
38 be translated. This only affected versions 5.4.2 and 5.4.3.
40 - Updated the Chinese (simplified), Croatian, Esperanto, German,
41 Korean, Polish, Romanian, Spanish, Swedish, Ukrainian, and
42 Vietnamese translations.
44 - Updated the German, Korean, Romanian, and Ukrainian man page
45 translations.
48 5.4.3 (2023-05-04)
50 * All fixes from 5.2.12
52 * Features in the CMake build can now be disabled as CMake cache
53 variables, similar to the Autotools build.
55 * Minor update to the Croatian translation.
58 5.4.2 (2023-03-18)
60 * All fixes from 5.2.11 that were not included in 5.4.1.
62 * If xz is built with support for the Capsicum sandbox but running
63 in an environment that doesn't support Capsicum, xz now runs
64 normally without sandboxing instead of exiting with an error.
66 * liblzma:
68 - Documentation was updated to improve the style, consistency,
69 and completeness of the liblzma API headers.
71 - The Doxygen-generated HTML documentation for the liblzma API
72 header files is now included in the source release and is
73 installed as part of "make install". All JavaScript is
74 removed to simplify license compliance and to reduce the
75 install size.
77 - Fixed a minor bug in lzma_str_from_filters() that produced
78 too many filters in the output string instead of reporting
79 an error if the input array had more than four filters. This
80 bug did not affect xz.
82 * Build systems:
84 - autogen.sh now invokes the doxygen tool via the new wrapper
85 script doxygen/update-doxygen, unless the command line option
86 --no-doxygen is used.
88 - Added microlzma_encoder.c and microlzma_decoder.c to the
89 VS project files for Windows and to the CMake build. These
90 should have been included in 5.3.2alpha.
92 * Tests:
94 - Added a test to the CMake build that was forgotten in the
95 previous release.
97 - Added and refactored a few tests.
99 * Translations:
101 - Updated the Brazilian Portuguese translation.
103 - Added Brazilian Portuguese man page translation.
106 5.4.1 (2023-01-11)
108 * liblzma:
110 - Fixed the return value of lzma_microlzma_encoder() if the
111 LZMA options lc/lp/pb are invalid. Invalid lc/lp/pb options
112 made the function return LZMA_STREAM_END without encoding
113 anything instead of returning LZMA_OPTIONS_ERROR.
115 - Windows / Visual Studio: Workaround a possible compiler bug
116 when targeting 32-bit x86 and compiling the CLMUL version of
117 the CRC64 code. The CLMUL code isn't enabled by the Windows
118 project files but it is in the CMake-based builds.
120 * Build systems:
122 - Windows-specific CMake changes:
124 * Don't try to enable CLMUL CRC64 code if _mm_set_epi64x()
125 isn't available. This fixes CMake-based build with Visual
126 Studio 2013.
128 * Created a workaround for a build failure with windres
129 from GNU binutils. It is used only when the C compiler
130 is GCC (not Clang). The workaround is incompatible
131 with llvm-windres, resulting in "XZx20Utils" instead
132 of "XZ Utils" in the resource file, but without the
133 workaround llvm-windres works correctly. See the
134 comment in CMakeLists.txt for details.
136 * Included the resource files in the xz and xzdec build
137 rules. Building the command line tools is still
138 experimental but possible with MinGW-w64.
140 - Visual Studio: Added stream_decoder_mt.c to the project
141 files. Now the threaded decompressor lzma_stream_decoder_mt()
142 gets built. CMake-based build wasn't affected.
144 - Updated windows/INSTALL-MSVC.txt to mention that CMake-based
145 build is now the preferred method with Visual Studio. The
146 project files will probably be removed after 5.4.x releases.
148 - Changes to #defines in config.h:
150 * HAVE_DECL_CLOCK_MONOTONIC was replaced by
151 HAVE_CLOCK_MONOTONIC. The old macro was always defined
152 in configure-generated config.h to either 0 or 1. The
153 new macro is defined (to 1) only if the declaration of
154 CLOCK_MONOTONIC is available. This matches the way most
155 other config.h macros work and makes things simpler with
156 other build systems.
158 * HAVE_DECL_PROGRAM_INVOCATION_NAME was replaced by
159 HAVE_PROGRAM_INVOCATION_NAME for the same reason.
161 * Tests:
163 - Fixed test script compatibility with ancient /bin/sh
164 versions. Now the five test_compress_* tests should
165 no longer fail on Solaris 10.
167 - Added and refactored a few tests.
169 * Translations:
171 - Updated the Catalan and Esperanto translations.
173 - Added Korean and Ukrainian man page translations.
176 5.4.0 (2022-12-13)
178 This bumps the minor version of liblzma because new features were
179 added. The API and ABI are still backward compatible with liblzma
180 5.2.x and 5.0.x.
182 Since 5.3.5beta:
184 * All fixes from 5.2.10.
186 * The ARM64 filter is now stable. The xz option is now --arm64.
187 Decompression requires XZ Utils 5.4.0. In the future the ARM64
188 filter will be supported by XZ for Java, XZ Embedded (including
189 the version in Linux), LZMA SDK, and 7-Zip.
191 * Translations:
193 - Updated Catalan, Croatian, German, Romanian, and Turkish
194 translations.
196 - Updated German man page translations.
198 - Added Romanian man page translations.
200 Summary of new features added in the 5.3.x development releases:
202 * liblzma:
204 - Added threaded .xz decompressor lzma_stream_decoder_mt().
205 It can use multiple threads with .xz files that have multiple
206 Blocks with size information in Block Headers. The threaded
207 encoder in xz has always created such files.
209 Single-threaded encoder cannot store the size information in
210 Block Headers even if one used LZMA_FULL_FLUSH to create
211 multiple Blocks, so this threaded decoder cannot use multiple
212 threads with such files.
214 If there are multiple Streams (concatenated .xz files), one
215 Stream will be decompressed completely before starting the
216 next Stream.
218 - A new decoder flag LZMA_FAIL_FAST was added. It makes the
219 threaded decompressor report errors soon instead of first
220 flushing all pending data before the error location.
222 - New Filter IDs:
223 * LZMA_FILTER_ARM64 is for ARM64 binaries.
224 * LZMA_FILTER_LZMA1EXT is for raw LZMA1 streams that don't
225 necessarily use the end marker.
227 - Added lzma_str_to_filters(), lzma_str_from_filters(), and
228 lzma_str_list_filters() to convert a preset or a filter chain
229 string to a lzma_filter[] and vice versa. These should make
230 it easier to write applications that allow users to specify
231 custom compression options.
233 - Added lzma_filters_free() which can be convenient for freeing
234 the filter options in a filter chain (an array of lzma_filter
235 structures).
237 - lzma_file_info_decoder() to makes it a little easier to get
238 the Index field from .xz files. This helps in getting the
239 uncompressed file size but an easy-to-use random access
240 API is still missing which has existed in XZ for Java for
241 a long time.
243 - Added lzma_microlzma_encoder() and lzma_microlzma_decoder().
244 It is used by erofs-utils and may be used by others too.
246 The MicroLZMA format is a raw LZMA stream (without end marker)
247 whose first byte (always 0x00) has been replaced with
248 bitwise-negation of the LZMA properties (lc/lp/pb). It was
249 created for use in EROFS but may be used in other contexts
250 as well where it is important to avoid wasting bytes for
251 stream headers or footers. The format is also supported by
252 XZ Embedded (the XZ Embedded version in Linux got MicroLZMA
253 support in Linux 5.16).
255 The MicroLZMA encoder API in liblzma can compress into a
256 fixed-sized output buffer so that as much data is compressed
257 as can be fit into the buffer while still creating a valid
258 MicroLZMA stream. This is needed for EROFS.
260 - Added lzma_lzip_decoder() to decompress the .lz (lzip) file
261 format version 0 and the original unextended version 1 files.
262 Also lzma_auto_decoder() supports .lz files.
264 - lzma_filters_update() can now be used with the multi-threaded
265 encoder (lzma_stream_encoder_mt()) to change the filter chain
266 after LZMA_FULL_BARRIER or LZMA_FULL_FLUSH.
268 - In lzma_options_lzma, allow nice_len = 2 and 3 with the match
269 finders that require at least 3 or 4. Now it is internally
270 rounded up if needed.
272 - CLMUL-based CRC64 on x86-64 and E2K with runtime processor
273 detection. On 32-bit x86 it currently isn't available unless
274 --disable-assembler is used which can make the non-CLMUL
275 CRC64 slower; this might be fixed in the future.
277 - Building with --disable-threads --enable-small
278 is now thread-safe if the compiler supports
279 __attribute__((__constructor__)).
281 * xz:
283 - Using -T0 (--threads=0) will now use multi-threaded encoder
284 even on a single-core system. This is to ensure that output
285 from the same xz binary is identical on both single-core and
286 multi-core systems.
288 - --threads=+1 or -T+1 is now a way to put xz into
289 multi-threaded mode while using only one worker thread.
290 The + is ignored if the number is not 1.
292 - A default soft memory usage limit is now used for compression
293 when -T0 is used and no explicit limit has been specified.
294 This soft limit is used to restrict the number of threads
295 but if the limit is exceeded with even one thread then xz
296 will continue with one thread using the multi-threaded
297 encoder and this limit is ignored. If the number of threads
298 is specified manually then no default limit will be used;
299 this affects only -T0.
301 This change helps on systems that have very many cores and
302 using all of them for xz makes no sense. Previously xz -T0
303 could run out of memory on such systems because it attempted
304 to reserve memory for too many threads.
306 This also helps with 32-bit builds which don't have a large
307 amount of address space that would be required for many
308 threads. The default soft limit for -T0 is at most 1400 MiB
309 on all 32-bit platforms.
311 - Previously a low value in --memlimit-compress wouldn't cause
312 xz to switch from multi-threaded mode to single-threaded mode
313 if the limit cannot otherwise be met; xz failed instead. Now
314 xz can switch to single-threaded mode and then, if needed,
315 scale down the LZMA2 dictionary size too just like it already
316 did when it was started in single-threaded mode.
318 - The option --no-adjust no longer prevents xz from scaling down
319 the number of threads as that doesn't affect the compressed
320 output (only performance). Now --no-adjust only prevents
321 adjustments that affect compressed output, that is, with
322 --no-adjust xz won't switch from multi-threaded mode to
323 single-threaded mode and won't scale down the LZMA2
324 dictionary size.
326 - Added a new option --memlimit-mt-decompress=LIMIT. This is
327 used to limit the number of decompressor threads (possibly
328 falling back to single-threaded mode) but it will never make
329 xz refuse to decompress a file. This has a system-specific
330 default value because without any limit xz could end up
331 allocating memory for the whole compressed input file, the
332 whole uncompressed output file, multiple thread-specific
333 decompressor instances and so on. Basically xz could
334 attempt to use an insane amount of memory even with fairly
335 common files. The system-specific default value is currently
336 the same as the one used for compression with -T0.
338 The new option works together with the existing option
339 --memlimit-decompress=LIMIT. The old option sets a hard limit
340 that must not be exceeded (xz will refuse to decompress)
341 while the new option only restricts the number of threads.
342 If the limit set with --memlimit-mt-decompress is greater
343 than the limit set with --memlimit-compress, then the latter
344 value is used also for --memlimit-mt-decompress.
346 - Added new information to the output of xz --info-memory and
347 new fields to the output of xz --robot --info-memory.
349 - In --lzma2=nice=NUMBER allow 2 and 3 with all match finders
350 now that liblzma handles it.
352 - Don't mention endianness for ARM and ARM-Thumb filters in
353 --long-help. The filters only work for little endian
354 instruction encoding but modern ARM processors using
355 big endian data access still use little endian
356 instruction encoding. So the help text was misleading.
357 In contrast, the PowerPC filter is only for big endian
358 32/64-bit PowerPC code. Little endian PowerPC would need
359 a separate filter.
361 - Added decompression support for the .lz (lzip) file format
362 version 0 and the original unextended version 1. It is
363 autodetected by default. See also the option --format on
364 the xz man page.
366 - Sandboxing enabled by default:
367 * Capsicum (FreeBSD)
368 * pledge(2) (OpenBSD)
370 * Scripts now support the .lz format using xz.
372 * A few new tests were added.
374 * The liblzma-specific tests are now supported in CMake-based
375 builds too ("make test").
378 5.3.5beta (2022-12-01)
380 * All fixes from 5.2.9.
382 * liblzma:
384 - Added new LZMA_FILTER_LZMA1EXT for raw encoder and decoder to
385 handle raw LZMA1 streams that don't have end of payload marker
386 (EOPM) alias end of stream (EOS) marker. It can be used in
387 filter chains, for example, with the x86 BCJ filter.
389 - Added lzma_str_to_filters(), lzma_str_from_filters(), and
390 lzma_str_list_filters() to make it easier for applications
391 to get custom compression options from a user and convert
392 it to an array of lzma_filter structures.
394 - Added lzma_filters_free().
396 - lzma_filters_update() can now be used with the multi-threaded
397 encoder (lzma_stream_encoder_mt()) to change the filter chain
398 after LZMA_FULL_BARRIER or LZMA_FULL_FLUSH.
400 - In lzma_options_lzma, allow nice_len = 2 and 3 with the match
401 finders that require at least 3 or 4. Now it is internally
402 rounded up if needed.
404 - ARM64 filter was modified. It is still experimental.
406 - Fixed LTO build with Clang if -fgnuc-version=10 or similar
407 was used to make Clang look like GCC >= 10. Now it uses
408 __has_attribute(__symver__) which should be reliable.
410 * xz:
412 - --threads=+1 or -T+1 is now a way to put xz into multi-threaded
413 mode while using only one worker thread.
415 - In --lzma2=nice=NUMBER allow 2 and 3 with all match finders
416 now that liblzma handles it.
418 * Updated translations: Chinese (simplified), Korean, and Turkish.
421 5.3.4alpha (2022-11-15)
423 * All fixes from 5.2.7 and 5.2.8.
425 * liblzma:
427 - Minor improvements to the threaded decoder.
429 - Added CRC64 implementation that uses SSSE3, SSE4.1, and CLMUL
430 instructions on 32/64-bit x86 and E2K. On 32-bit x86 it's
431 not enabled unless --disable-assembler is used but then
432 the non-CLMUL code might be slower. Processor support is
433 detected at runtime so this is built by default on x86-64
434 and E2K. On these platforms, if compiler flags indicate
435 unconditional CLMUL support (-msse4.1 -mpclmul) then the
436 generic version is not built, making liblzma 8-9 KiB smaller
437 compared to having both versions included.
439 With extremely compressible files this can make decompression
440 up to twice as fast but with typical files 5 % improvement
441 is a more realistic expectation.
443 The CLMUL version is slower than the generic version with
444 tiny inputs (especially at 1-8 bytes per call, but up to
445 16 bytes). In normal use in xz this doesn't matter at all.
447 - Added an experimental ARM64 filter. This is *not* the final
448 version! Files created with this experimental version won't
449 be supported in the future versions! The filter design is
450 a compromise where improving one use case makes some other
451 cases worse.
453 - Added decompression support for the .lz (lzip) file format
454 version 0 and the original unextended version 1. See the
455 API docs of lzma_lzip_decoder() for details. Also
456 lzma_auto_decoder() supports .lz files.
458 - Building with --disable-threads --enable-small
459 is now thread-safe if the compiler supports
460 __attribute__((__constructor__))
462 * xz:
464 - Added support for OpenBSD's pledge(2) as a sandboxing method.
466 - Don't mention endianness for ARM and ARM-Thumb filters in
467 --long-help. The filters only work for little endian
468 instruction encoding but modern ARM processors using
469 big endian data access still use little endian
470 instruction encoding. So the help text was misleading.
471 In contrast, the PowerPC filter is only for big endian
472 32/64-bit PowerPC code. Little endian PowerPC would need
473 a separate filter.
475 - Added --experimental-arm64. This will be renamed once the
476 filter is finished. Files created with this experimental
477 filter will not be supported in the future!
479 - Added new fields to the output of xz --robot --info-memory.
481 - Added decompression support for the .lz (lzip) file format
482 version 0 and the original unextended version 1. It is
483 autodetected by default. See also the option --format on
484 the xz man page.
486 * Scripts now support the .lz format using xz.
488 * Build systems:
490 - New #defines in config.h: HAVE_ENCODER_ARM64,
491 HAVE_DECODER_ARM64, HAVE_LZIP_DECODER, HAVE_CPUID_H,
492 HAVE_FUNC_ATTRIBUTE_CONSTRUCTOR, HAVE_USABLE_CLMUL
494 - New configure options: --disable-clmul-crc,
495 --disable-microlzma, --disable-lzip-decoder, and
496 'pledge' is now an option in --enable-sandbox (but
497 it's autodetected by default anyway).
499 - INSTALL was updated to document the new configure options.
501 - PACKAGERS now lists also --disable-microlzma and
502 --disable-lzip-decoder as configure options that must
503 not be used in builds for non-embedded use.
505 * Tests:
507 - Fix some of the tests so that they skip instead of fail if
508 certain features have been disabled with configure options.
509 It's still not perfect.
511 - Other improvements to tests.
513 * Updated translations: Croatian, Finnish, Hungarian, Polish,
514 Romanian, Spanish, Swedish, and Ukrainian.
517 5.3.3alpha (2022-08-22)
519 * All fixes from 5.2.6.
521 * liblzma:
523 - Fixed 32-bit build.
525 - Added threaded .xz decompressor lzma_stream_decoder_mt().
526 It can use multiple threads with .xz files that have multiple
527 Blocks with size information in Block Headers. The threaded
528 encoder in xz has always created such files.
530 Single-threaded encoder cannot store the size information in
531 Block Headers even if one used LZMA_FULL_FLUSH to create
532 multiple Blocks, so this threaded decoder cannot use multiple
533 threads with such files.
535 If there are multiple Streams (concatenated .xz files), one
536 Stream will be decompressed completely before starting the
537 next Stream.
539 - A new decoder flag LZMA_FAIL_FAST was added. It makes the
540 threaded decompressor report errors soon instead of first
541 flushing all pending data before the error location.
543 * xz:
545 - Using -T0 (--threads=0) will now use multi-threaded encoder
546 even on a single-core system. This is to ensure that output
547 from the same xz binary is identical on both single-core and
548 multi-core systems.
550 - A default soft memory usage limit is now used for compression
551 when -T0 is used and no explicit limit has been specified.
552 This soft limit is used to restrict the number of threads
553 but if the limit is exceeded with even one thread then xz
554 will continue with one thread using the multi-threaded
555 encoder and this limit is ignored. If the number of threads
556 is specified manually then no default limit will be used;
557 this affects only -T0.
559 This change helps on systems that have very many cores and
560 using all of them for xz makes no sense. Previously xz -T0
561 could run out of memory on such systems because it attempted
562 to reserve memory for too many threads.
564 This also helps with 32-bit builds which don't have a large
565 amount of address space that would be required for many
566 threads. The default limit is 1400 MiB on all 32-bit
567 platforms with -T0.
569 Now xz -T0 should just work. It might use too few threads
570 in some cases but at least it shouldn't easily run out of
571 memory. It's possible that this will be tweaked before 5.4.0.
573 - Changes to --memlimit-compress and --no-adjust:
575 In single-threaded mode, --memlimit-compress can make xz
576 scale down the LZMA2 dictionary size to meet the memory usage
577 limit. This obviously affects the compressed output. However,
578 if xz was in threaded mode, --memlimit-compress could make xz
579 reduce the number of threads but it wouldn't make xz switch
580 from multi-threaded mode to single-threaded mode or scale
581 down the LZMA2 dictionary size. This seemed illogical.
583 Now --memlimit-compress can make xz switch to single-threaded
584 mode if one thread in multi-threaded mode uses too much
585 memory. If memory usage is still too high, then the LZMA2
586 dictionary size can be scaled down too.
588 The option --no-adjust was also changed so that it no longer
589 prevents xz from scaling down the number of threads as that
590 doesn't affect compressed output (only performance). After
591 this commit --no-adjust only prevents adjustments that affect
592 compressed output, that is, with --no-adjust xz won't switch
593 from multithreaded mode to single-threaded mode and won't
594 scale down the LZMA2 dictionary size.
596 - Added a new option --memlimit-mt-decompress=LIMIT. This is
597 used to limit the number of decompressor threads (possibly
598 falling back to single-threaded mode) but it will never make
599 xz refuse to decompress a file. This has a system-specific
600 default value because without any limit xz could end up
601 allocating memory for the whole compressed input file, the
602 whole uncompressed output file, multiple thread-specific
603 decompressor instances and so on. Basically xz could
604 attempt to use an insane amount of memory even with fairly
605 common files.
607 The new option works together with the existing option
608 --memlimit-decompress=LIMIT. The old option sets a hard limit
609 that must not be exceeded (xz will refuse to decompress)
610 while the new option only restricts the number of threads.
611 If the limit set with --memlimit-mt-decompress is greater
612 than the limit set with --memlimit-compress, then the latter
613 value is used also for --memlimit-mt-decompress.
615 * Tests:
617 - Added a few more tests.
619 - Added tests/code_coverage.sh to create a code coverage report
620 of the tests.
622 * Build systems:
624 - Automake's parallel test harness is now used to make tests
625 finish faster.
627 - Added the CMake files to the distribution tarball. These were
628 supposed to be in 5.2.5 already.
630 - Added liblzma tests to the CMake build.
632 - Windows: Fix building of liblzma.dll with the included
633 Visual Studio project files.
636 5.3.2alpha (2021-10-28)
638 This release was made on short notice so that recent erofs-utils can
639 be built with LZMA support without needing a snapshot from xz.git.
640 Thus many pending things were not included, not even updated
641 translations (which would need to be updated for the new --list
642 strings anyway).
644 * All fixes from 5.2.5.
646 * xz:
648 - When copying metadata from the source file to the destination
649 file, don't try to set the group (GID) if it is already set
650 correctly. This avoids a failure on OpenBSD (and possibly on
651 a few other OSes) where files may get created so that their
652 group doesn't belong to the user, and fchown(2) can fail even
653 if it needs to do nothing.
655 - The --keep option now accepts symlinks, hardlinks, and
656 setuid, setgid, and sticky files. Previously this required
657 using --force.
659 - Split the long strings used in --list and --info-memory modes
660 to make them much easier for translators.
662 - If built with sandbox support and enabling the sandbox fails,
663 xz will now immediately exit with exit status of 1. Previously
664 it would only display a warning if -vv was used.
666 - Cap --memlimit-compress to 2000 MiB on MIPS32 because on
667 MIPS32 userspace processes are limited to 2 GiB of address
668 space.
670 * liblzma:
672 - Added lzma_microlzma_encoder() and lzma_microlzma_decoder().
673 The API is in lzma/container.h.
675 The MicroLZMA format is a raw LZMA stream (without end marker)
676 whose first byte (always 0x00) has been replaced with
677 bitwise-negation of the LZMA properties (lc/lp/pb). It was
678 created for use in EROFS but may be used in other contexts
679 as well where it is important to avoid wasting bytes for
680 stream headers or footers. The format is also supported by
681 XZ Embedded.
683 The MicroLZMA encoder API in liblzma can compress into a
684 fixed-sized output buffer so that as much data is compressed
685 as can be fit into the buffer while still creating a valid
686 MicroLZMA stream. This is needed for EROFS.
688 - Added fuzzing support.
690 - Support Intel Control-flow Enforcement Technology (CET) in
691 32-bit x86 assembly files.
693 - Visual Studio: Use non-standard _MSVC_LANG to detect C++
694 standard version in the lzma.h API header. It's used to
695 detect when "noexcept" can be used.
697 * Scripts:
699 - Fix exit status of xzdiff/xzcmp. Exit status could be 2 when
700 the correct value is 1.
702 - Fix exit status of xzgrep.
704 - Detect corrupt .bz2 files in xzgrep.
706 - Add zstd support to xzgrep and xzdiff/xzcmp.
708 - Fix less(1) version detection in xzless. It failed if the
709 version number from "less -V" contained a dot.
711 * Fix typos and technical issues in man pages.
713 * Build systems:
715 - Windows: Fix building of resource files when config.h isn't
716 used. CMake + Visual Studio can now build liblzma.dll.
718 - Various fixes to the CMake support. It might still need a few
719 more fixes even for liblzma-only builds.
722 5.3.1alpha (2018-04-29)
724 * All fixes from 5.2.4.
726 * Add lzma_file_info_decoder() into liblzma and use it in xz to
727 implement the --list feature.
729 * Capsicum sandbox support is enabled by default where available
730 (FreeBSD >= 10).
733 5.2.12 (2023-05-04)
735 * Fixed a build system bug that prevented building liblzma as a
736 shared library when configured with --disable-threads. This bug
737 affected releases 5.2.6 to 5.2.11 and 5.4.0 to 5.4.2.
739 * Include <intrin.h> for Windows intrinsic functions where they are
740 needed. This fixed a bug that prevented building liblzma using
741 clang-cl on Windows.
743 * Minor update to the Croatian translation. The small change
744 applies to a string in both 5.2 and 5.4 branches.
747 5.2.11 (2023-03-18)
749 * Removed all possible cases of null pointer + 0. It is undefined
750 behavior in C99 and C17. This was detected by a sanitizer and had
751 not caused any known issues.
753 * Build systems:
755 - Added a workaround for building with GCC on MicroBlaze Linux.
756 GCC 12 on MicroBlaze doesn't support the __symver__ attribute
757 even though __has_attribute(__symver__) returns true. The
758 build is now done without the extra RHEL/CentOS 7 symbols
759 that were added in XZ Utils 5.2.7. The workaround only
760 applies to the Autotools build (not CMake).
762 - CMake: Ensure that the C compiler language is set to C99 or
763 a newer standard.
765 - CMake changes from XZ Utils 5.4.1:
767 * Added a workaround for a build failure with
768 windres from GNU binutils.
770 * Included the Windows resource files in the xz
771 and xzdec build rules.
774 5.2.10 (2022-12-13)
776 * xz: Don't modify argv[] when parsing the --memlimit* and
777 --block-list command line options. This fixes confusing
778 arguments in process listing (like "ps auxf").
780 * GNU/Linux only: Use __has_attribute(__symver__) to detect if
781 that attribute is supported. This fixes build on Mandriva where
782 Clang is patched to define __GNUC__ to 11 by default (instead
783 of 4 as used by Clang upstream).
786 5.2.9 (2022-11-30)
788 * liblzma:
790 - Fixed an infinite loop in LZMA encoder initialization
791 if dict_size >= 2 GiB. (The encoder only supports up
792 to 1536 MiB.)
794 - Fixed two cases of invalid free() that can happen if
795 a tiny allocation fails in encoder re-initialization
796 or in lzma_filters_update(). These bugs had some
797 similarities with the bug fixed in 5.2.7.
799 - Fixed lzma_block_encoder() not allowing the use of
800 LZMA_SYNC_FLUSH with lzma_code() even though it was
801 documented to be supported. The sync-flush code in
802 the Block encoder was already used internally via
803 lzma_stream_encoder(), so this was just a missing flag
804 in the lzma_block_encoder() API function.
806 - GNU/Linux only: Don't put symbol versions into static
807 liblzma as it breaks things in some cases (and even if
808 it didn't break anything, symbol versions in static
809 libraries are useless anyway). The downside of the fix
810 is that if the configure options --with-pic or --without-pic
811 are used then it's not possible to build both shared and
812 static liblzma at the same time on GNU/Linux anymore;
813 with those options --disable-static or --disable-shared
814 must be used too.
816 * New email address for bug reports is <xz@tukaani.org> which
817 forwards messages to Lasse Collin and Jia Tan.
820 5.2.8 (2022-11-13)
822 * xz:
824 - If xz cannot remove an input file when it should, this
825 is now treated as a warning (exit status 2) instead of
826 an error (exit status 1). This matches GNU gzip and it
827 is more logical as at that point the output file has
828 already been successfully closed.
830 - Fix handling of .xz files with an unsupported check type.
831 Previously such printed a warning message but then xz
832 behaved as if an error had occurred (didn't decompress,
833 exit status 1). Now a warning is printed, decompression
834 is done anyway, and exit status is 2. This used to work
835 slightly before 5.0.0. In practice this bug matters only
836 if xz has been built with some check types disabled. As
837 instructed in PACKAGERS, such builds should be done in
838 special situations only.
840 - Fix "xz -dc --single-stream tests/files/good-0-empty.xz"
841 which failed with "Internal error (bug)". That is,
842 --single-stream was broken if the first .xz stream in
843 the input file didn't contain any uncompressed data.
845 - Fix displaying file sizes in the progress indicator when
846 working in passthru mode and there are multiple input files.
847 Just like "gzip -cdf", "xz -cdf" works like "cat" when the
848 input file isn't a supported compressed file format. In
849 this case the file size counters weren't reset between
850 files so with multiple input files the progress indicator
851 displayed an incorrect (too large) value.
853 * liblzma:
855 - API docs in lzma/container.h:
856 * Update the list of decoder flags in the decoder
857 function docs.
858 * Explain LZMA_CONCATENATED behavior with .lzma files
859 in lzma_auto_decoder() docs.
861 - OpenBSD: Use HW_NCPUONLINE to detect the number of
862 available hardware threads in lzma_physmem().
864 - Fix use of wrong macro to detect x86 SSE2 support.
865 __SSE2_MATH__ was used with GCC/Clang but the correct
866 one is __SSE2__. The first one means that SSE2 is used
867 for floating point math which is irrelevant here.
868 The affected SSE2 code isn't used on x86-64 so this affects
869 only 32-bit x86 builds that use -msse2 without -mfpmath=sse
870 (there is no runtime detection for SSE2). It improves LZMA
871 compression speed (not decompression).
873 - Fix the build with Intel C compiler 2021 (ICC, not ICX)
874 on Linux. It defines __GNUC__ to 10 but doesn't support
875 the __symver__ attribute introduced in GCC 10.
877 * Scripts: Ignore warnings from xz by using --quiet --no-warn.
878 This is needed if the input .xz files use an unsupported
879 check type.
881 * Translations:
883 - Updated Croatian and Turkish translations.
885 - One new translations wasn't included because it needed
886 technical fixes. It will be in upcoming 5.4.0. No new
887 translations will be added to the 5.2.x branch anymore.
889 - Renamed the French man page translation file from
890 fr_FR.po to fr.po and thus also its install directory
891 (like /usr/share/man/fr_FR -> .../fr).
893 - Man page translations for upcoming 5.4.0 are now handled
894 in the Translation Project.
896 * Update doc/faq.txt a little so it's less out-of-date.
899 5.2.7 (2022-09-30)
901 * liblzma:
903 - Made lzma_filters_copy() to never modify the destination
904 array if an error occurs. lzma_stream_encoder() and
905 lzma_stream_encoder_mt() already assumed this. Before this
906 change, if a tiny memory allocation in lzma_filters_copy()
907 failed it would lead to a crash (invalid free() or invalid
908 memory reads) in the cleanup paths of these two encoder
909 initialization functions.
911 - Added missing integer overflow check to lzma_index_append().
912 This affects xz --list and other applications that decode
913 the Index field from .xz files using lzma_index_decoder().
914 Normal decompression of .xz files doesn't call this code
915 and thus most applications using liblzma aren't affected
916 by this bug.
918 - Single-threaded .xz decoder (lzma_stream_decoder()): If
919 lzma_code() returns LZMA_MEMLIMIT_ERROR it is now possible
920 to use lzma_memlimit_set() to increase the limit and continue
921 decoding. This was supposed to work from the beginning
922 but there was a bug. With other decoders (.lzma or
923 threaded .xz decoder) this already worked correctly.
925 - Fixed accumulation of integrity check type statistics in
926 lzma_index_cat(). This bug made lzma_index_checks() return
927 only the type of the integrity check of the last Stream
928 when multiple lzma_indexes were concatenated. Most
929 applications don't use these APIs but in xz it made
930 xz --list not list all check types from concatenated .xz
931 files. In xz --list --verbose only the per-file "Check:"
932 lines were affected and in xz --robot --list only the "file"
933 line was affected.
935 - Added ABI compatibility with executables that were linked
936 against liblzma in RHEL/CentOS 7 or other liblzma builds
937 that had copied the problematic patch from RHEL/CentOS 7
938 (xz-5.2.2-compat-libs.patch). For the details, see the
939 comment at the top of src/liblzma/validate_map.sh.
941 WARNING: This uses __symver__ attribute with GCC >= 10.
942 In other cases the traditional __asm__(".symver ...")
943 is used. Using link-time optimization (LTO, -flto) with
944 GCC versions older than 10 can silently result in
945 broken liblzma.so.5 (incorrect symbol versions)! If you
946 want to use -flto with GCC, you must use GCC >= 10.
947 LTO with Clang seems to work even with the traditional
948 __asm__(".symver ...") method.
950 * xzgrep: Fixed compatibility with old shells that break if
951 comments inside command substitutions have apostrophes (').
952 This problem was introduced in 5.2.6.
954 * Build systems:
956 - New #define in config.h: HAVE_SYMBOL_VERSIONS_LINUX
958 - Windows: Fixed liblzma.dll build with Visual Studio project
959 files. It broke in 5.2.6 due to a change that was made to
960 improve CMake support.
962 - Windows: Building liblzma with UNICODE defined should now
963 work.
965 - CMake files are now actually included in the release tarball.
966 They should have been in 5.2.5 already.
968 - Minor CMake fixes and improvements.
970 * Added a new translation: Turkish
973 5.2.6 (2022-08-12)
975 * xz:
977 - The --keep option now accepts symlinks, hardlinks, and
978 setuid, setgid, and sticky files. Previously this required
979 using --force.
981 - When copying metadata from the source file to the destination
982 file, don't try to set the group (GID) if it is already set
983 correctly. This avoids a failure on OpenBSD (and possibly on
984 a few other OSes) where files may get created so that their
985 group doesn't belong to the user, and fchown(2) can fail even
986 if it needs to do nothing.
988 - Cap --memlimit-compress to 2000 MiB instead of 4020 MiB on
989 MIPS32 because on MIPS32 userspace processes are limited
990 to 2 GiB of address space.
992 * liblzma:
994 - Fixed a missing error-check in the threaded encoder. If a
995 small memory allocation fails, a .xz file with an invalid
996 Index field would be created. Decompressing such a file would
997 produce the correct output but result in an error at the end.
998 Thus this is a "mild" data corruption bug. Note that while
999 a failed memory allocation can trigger the bug, it cannot
1000 cause invalid memory access.
1002 - The decoder for .lzma files now supports files that have
1003 uncompressed size stored in the header and still use the
1004 end of payload marker (end of stream marker) at the end
1005 of the LZMA stream. Such files are rare but, according to
1006 the documentation in LZMA SDK, they are valid.
1007 doc/lzma-file-format.txt was updated too.
1009 - Improved 32-bit x86 assembly files:
1010 * Support Intel Control-flow Enforcement Technology (CET)
1011 * Use non-executable stack on FreeBSD.
1013 - Visual Studio: Use non-standard _MSVC_LANG to detect C++
1014 standard version in the lzma.h API header. It's used to
1015 detect when "noexcept" can be used.
1017 * xzgrep:
1019 - Fixed arbitrary command injection via a malicious filename
1020 (CVE-2022-1271, ZDI-CAN-16587). A standalone patch for
1021 this was released to the public on 2022-04-07. A slight
1022 robustness improvement has been made since then and, if
1023 using GNU or *BSD grep, a new faster method is now used
1024 that doesn't use the old sed-based construct at all. This
1025 also fixes bad output with GNU grep >= 3.5 (2020-09-27)
1026 when xzgrepping binary files.
1028 This vulnerability was discovered by:
1029 cleemy desu wayo working with Trend Micro Zero Day Initiative
1031 - Fixed detection of corrupt .bz2 files.
1033 - Improved error handling to fix exit status in some situations
1034 and to fix handling of signals: in some situations a signal
1035 didn't make xzgrep exit when it clearly should have. It's
1036 possible that the signal handling still isn't quite perfect
1037 but hopefully it's good enough.
1039 - Documented exit statuses on the man page.
1041 - xzegrep and xzfgrep now use "grep -E" and "grep -F" instead
1042 of the deprecated egrep and fgrep commands.
1044 - Fixed parsing of the options -E, -F, -G, -P, and -X. The
1045 problem occurred when multiple options were specified in
1046 a single argument, for example,
1048 echo foo | xzgrep -Fe foo
1050 treated foo as a filename because -Fe wasn't correctly
1051 split into -F -e.
1053 - Added zstd support.
1055 * xzdiff/xzcmp:
1057 - Fixed wrong exit status. Exit status could be 2 when the
1058 correct value is 1.
1060 - Documented on the man page that exit status of 2 is used
1061 for decompression errors.
1063 - Added zstd support.
1065 * xzless:
1067 - Fix less(1) version detection. It failed if the version number
1068 from "less -V" contained a dot.
1070 * Translations:
1072 - Added new translations: Catalan, Croatian, Esperanto,
1073 Korean, Portuguese, Romanian, Serbian, Spanish, Swedish,
1074 and Ukrainian
1076 - Updated the Brazilian Portuguese translation.
1078 - Added French man page translation. This and the existing
1079 German translation aren't complete anymore because the
1080 English man pages got a few updates and the translators
1081 weren't reached so that they could update their work.
1083 * Build systems:
1085 - Windows: Fix building of resource files when config.h isn't
1086 used. CMake + Visual Studio can now build liblzma.dll.
1088 - Various fixes to the CMake support. Building static or shared
1089 liblzma should work fine in most cases. In contrast, building
1090 the command line tools with CMake is still clearly incomplete
1091 and experimental and should be used for testing only.
1094 5.2.5 (2020-03-17)
1096 * liblzma:
1098 - Fixed several C99/C11 conformance bugs. Now the code is clean
1099 under gcc/clang -fsanitize=undefined. Some of these changes
1100 might have a negative effect on performance with old GCC
1101 versions or compilers other than GCC and Clang. The configure
1102 option --enable-unsafe-type-punning can be used to (mostly)
1103 restore the old behavior but it shouldn't normally be used.
1105 - Improved API documentation of lzma_properties_decode().
1107 - Added a very minor encoder speed optimization.
1109 * xz:
1111 - Fixed a crash in "xz -dcfv not_an_xz_file". All four options
1112 were required to trigger it. The crash occurred in the
1113 progress indicator code when xz was in passthru mode where
1114 xz works like "cat".
1116 - Fixed an integer overflow with 32-bit off_t. It could happen
1117 when decompressing a file that has a long run of zero bytes
1118 which xz would try to write as a sparse file. Since the build
1119 system enables large file support by default, off_t is
1120 normally 64-bit even on 32-bit systems.
1122 - Fixes for --flush-timeout:
1123 * Fix semi-busy-waiting.
1124 * Avoid unneeded flushes when no new input has arrived
1125 since the previous flush was completed.
1127 - Added a special case for 32-bit xz: If --memlimit-compress is
1128 used to specify a limit that exceeds 4020 MiB, the limit will
1129 be set to 4020 MiB. The values "0" and "max" aren't affected
1130 by this and neither is decompression. This hack can be
1131 helpful when a 32-bit xz has access to 4 GiB address space
1132 but the specified memlimit exceeds 4 GiB. This can happen
1133 e.g. with some scripts.
1135 - Capsicum sandbox is now enabled by default where available
1136 (FreeBSD >= 10). The sandbox debug messages (xz -vv) were
1137 removed since they seemed to be more annoying than useful.
1139 - DOS build now requires DJGPP 2.05 instead of 2.04beta.
1140 A workaround for a locale problem with DJGPP 2.05 was added.
1142 * xzgrep and other scripts:
1144 - Added a configure option --enable-path-for-scripts=PREFIX.
1145 It is disabled by default except on Solaris where the default
1146 is /usr/xpg4/bin. See INSTALL for details.
1148 - Added a workaround for a POSIX shell detection problem on
1149 Solaris.
1151 * Build systems:
1153 - Added preliminary build instructions for z/OS. See INSTALL
1154 section 1.2.9.
1156 - Experimental CMake support was added. It should work to build
1157 static liblzma on a few operating systems. It may or may not
1158 work to build shared liblzma. On some platforms it can build
1159 xz and xzdec too but those are only for testing. See the
1160 comment in the beginning of CMakeLists.txt for details.
1162 - Visual Studio project files were updated.
1163 WindowsTargetPlatformVersion was removed from VS2017 files
1164 and set to "10.0" in the added VS2019 files. In the future
1165 the VS project files will be removed when CMake support is
1166 good enough.
1168 - New #defines in config.h: HAVE___BUILTIN_ASSUME_ALIGNED,
1169 HAVE___BUILTIN_BSWAPXX, and TUKLIB_USE_UNSAFE_TYPE_PUNNING.
1171 - autogen.sh has a new optional dependency on po4a and a new
1172 option --no-po4a to skip that step. This matters only if one
1173 wants to remake the build files. po4a is used to update the
1174 translated man pages but as long as the man pages haven't
1175 been modified, there's nothing to update and one can use
1176 --no-po4a to avoid the dependency on po4a.
1178 * Translations:
1180 - XZ Utils translations are now handled by the Translation
1181 Project: https://translationproject.org/domain/xz.html
1183 - All man pages are now included in German too.
1185 - New xz translations: Brazilian Portuguese, Finnish,
1186 Hungarian, Chinese (simplified), Chinese (traditional),
1187 and Danish (partial translation)
1189 - Updated xz translations: French, German, Italian, and Polish
1191 - Unfortunately a few new xz translations weren't included due
1192 to technical problems like too long lines in --help output or
1193 misaligned column headings in tables. In the future, many of
1194 these strings will be split and e.g. the table column
1195 alignment will be handled in software. This should make the
1196 strings easier to translate.
1199 5.2.4 (2018-04-29)
1201 * liblzma:
1203 - Allow 0 as memory usage limit instead of returning
1204 LZMA_PROG_ERROR. Now 0 is treated as if 1 byte was specified,
1205 which effectively is the same as 0.
1207 - Use "noexcept" keyword instead of "throw()" in the public
1208 headers when a C++11 (or newer standard) compiler is used.
1210 - Added a portability fix for recent Intel C Compilers.
1212 - Microsoft Visual Studio build files have been moved under
1213 windows/vs2013 and windows/vs2017.
1215 * xz:
1217 - Fix "xz --list --robot missing_or_bad_file.xz" which would
1218 try to print an uninitialized string and thus produce garbage
1219 output. Since the exit status is non-zero, most uses of such
1220 a command won't try to interpret the garbage output.
1222 - "xz --list foo.xz" could print "Internal error (bug)" in a
1223 corner case where a specific memory usage limit had been set.
1226 5.2.3 (2016-12-30)
1228 * xz:
1230 - Always close a file before trying to delete it to avoid
1231 problems on some operating system and file system combinations.
1233 - Fixed copying of file timestamps on Windows.
1235 - Added experimental (disabled by default) sandbox support using
1236 Capsicum (FreeBSD >= 10). See --enable-sandbox in INSTALL.
1238 * C99/C11 conformance fixes to liblzma. The issues affected at least
1239 some builds using link-time optimizations.
1241 * Fixed bugs in the rarely-used function lzma_index_dup().
1243 * Use of external SHA-256 code is now disabled by default.
1244 It can still be enabled by passing --enable-external-sha256
1245 to configure. The reasons to disable it by default (see INSTALL
1246 for more details):
1248 - Some OS-specific SHA-256 implementations conflict with
1249 OpenSSL and cause problems in programs that link against both
1250 liblzma and libcrypto. At least FreeBSD 10 and MINIX 3.3.0
1251 are affected.
1253 - The internal SHA-256 is faster than the SHA-256 code in
1254 some operating systems.
1256 * Changed CPU core count detection to use sched_getaffinity() on
1257 GNU/Linux and GNU/kFreeBSD.
1259 * Fixes to the build-system and xz to make xz buildable even when
1260 encoders, decoders, or threading have been disabled from libilzma
1261 using configure options. These fixes added two new #defines to
1262 config.h: HAVE_ENCODERS and HAVE_DECODERS.
1265 5.2.2 (2015-09-29)
1267 * Fixed bugs in QNX-specific code.
1269 * Omitted the use of pipe2() even if it is available to avoid
1270 portability issues with some old Linux and glibc combinations.
1272 * Updated German translation.
1274 * Added project files to build static and shared liblzma (not the
1275 whole XZ Utils) with Visual Studio 2013 update 2 or later.
1277 * Documented that threaded decompression hasn't been implemented
1278 yet. A 5.2.0 NEWS entry describing multi-threading support had
1279 incorrectly said "decompression" when it should have said
1280 "compression".
1283 5.2.1 (2015-02-26)
1285 * Fixed a compression-ratio regression in fast mode of LZMA1 and
1286 LZMA2. The bug is present in 5.1.4beta and 5.2.0 releases.
1288 * Fixed a portability problem in xz that affected at least OpenBSD.
1290 * Fixed xzdiff to be compatible with FreeBSD's mktemp which differs
1291 from most other mktemp implementations.
1293 * Changed CPU core count detection to use cpuset_getaffinity() on
1294 FreeBSD.
1297 5.2.0 (2014-12-21)
1299 Since 5.1.4beta:
1301 * All fixes from 5.0.8
1303 * liblzma: Fixed lzma_stream_encoder_mt_memusage() when a preset
1304 was used.
1306 * xzdiff: If mktemp isn't installed, mkdir will be used as
1307 a fallback to create a temporary directory. Installing mktemp
1308 is still recommended.
1310 * Updated French, German, Italian, Polish, and Vietnamese
1311 translations.
1313 Summary of fixes and new features added in the 5.1.x development
1314 releases:
1316 * liblzma:
1318 - Added support for multi-threaded compression. See the
1319 lzma_mt structure, lzma_stream_encoder_mt(), and
1320 lzma_stream_encoder_mt_memusage() in <lzma/container.h>,
1321 lzma_get_progress() in <lzma/base.h>, and lzma_cputhreads()
1322 in <lzma/hardware.h> for details.
1324 - Made the uses of lzma_allocator const correct.
1326 - Added lzma_block_uncomp_encode() to create uncompressed
1327 .xz Blocks using LZMA2 uncompressed chunks.
1329 - Added support for LZMA_IGNORE_CHECK.
1331 - A few speed optimizations were made.
1333 - Added support for symbol versioning. It is enabled by default
1334 on GNU/Linux, other GNU-based systems, and FreeBSD.
1336 - liblzma (not the whole XZ Utils) should now be buildable
1337 with MSVC 2013 update 2 or later using windows/config.h.
1339 * xz:
1341 - Fixed a race condition in the signal handling. It was
1342 possible that e.g. the first SIGINT didn't make xz exit
1343 if reading or writing blocked and one had bad luck. The fix
1344 is non-trivial, so as of writing it is unknown if it will be
1345 backported to the v5.0 branch.
1347 - Multi-threaded compression can be enabled with the
1348 --threads (-T) option.
1349 [Fixed: This originally said "decompression".]
1351 - New command line options in xz: --single-stream,
1352 --block-size=SIZE, --block-list=SIZES,
1353 --flush-timeout=TIMEOUT, and --ignore-check.
1355 - xz -lvv now shows the minimum xz version that is required to
1356 decompress the file. Currently it is 5.0.0 for all supported
1357 .xz files except files with empty LZMA2 streams require 5.0.2.
1359 * xzdiff and xzgrep now support .lzo files if lzop is installed.
1360 The .tzo suffix is also recognized as a shorthand for .tar.lzo.
1363 5.1.4beta (2014-09-14)
1365 * All fixes from 5.0.6
1367 * liblzma: Fixed the use of presets in threaded encoder
1368 initialization.
1370 * xz --block-list and --block-size can now be used together
1371 in single-threaded mode. Previously the combination only
1372 worked in multi-threaded mode.
1374 * Added support for LZMA_IGNORE_CHECK to liblzma and made it
1375 available in xz as --ignore-check.
1377 * liblzma speed optimizations:
1379 - Initialization of a new LZMA1 or LZMA2 encoder has been
1380 optimized. (The speed of reinitializing an already-allocated
1381 encoder isn't affected.) This helps when compressing many
1382 small buffers with lzma_stream_buffer_encode() and other
1383 similar situations where an already-allocated encoder state
1384 isn't reused. This speed-up is visible in xz too if one
1385 compresses many small files one at a time instead running xz
1386 once and giving all files as command-line arguments.
1388 - Buffer comparisons are now much faster when unaligned access
1389 is allowed (configured with --enable-unaligned-access). This
1390 speeds up encoding significantly. There is arch-specific code
1391 for 32-bit and 64-bit x86 (32-bit needs SSE2 for the best
1392 results and there's no run-time CPU detection for now).
1393 For other archs there is only generic code which probably
1394 isn't as optimal as arch-specific solutions could be.
1396 - A few speed optimizations were made to the SHA-256 code.
1397 (Note that the builtin SHA-256 code isn't used on all
1398 operating systems.)
1400 * liblzma can now be built with MSVC 2013 update 2 or later
1401 using windows/config.h.
1403 * Vietnamese translation was added.
1406 5.1.3alpha (2013-10-26)
1408 * All fixes from 5.0.5
1410 * liblzma:
1412 - Fixed a deadlock in the threaded encoder.
1414 - Made the uses of lzma_allocator const correct.
1416 - Added lzma_block_uncomp_encode() to create uncompressed
1417 .xz Blocks using LZMA2 uncompressed chunks.
1419 - Added support for native threads on Windows and the ability
1420 to detect the number of CPU cores.
1422 * xz:
1424 - Fixed a race condition in the signal handling. It was
1425 possible that e.g. the first SIGINT didn't make xz exit
1426 if reading or writing blocked and one had bad luck. The fix
1427 is non-trivial, so as of writing it is unknown if it will be
1428 backported to the v5.0 branch.
1430 - Made the progress indicator work correctly in threaded mode.
1432 - Threaded encoder now works together with --block-list=SIZES.
1434 - Added preliminary support for --flush-timeout=TIMEOUT.
1435 It can be useful for (somewhat) real-time streaming. For
1436 now the decompression side has to be done with something
1437 else than the xz tool due to how xz does buffering, but this
1438 should be fixed.
1441 5.1.2alpha (2012-07-04)
1443 * All fixes from 5.0.3 and 5.0.4
1445 * liblzma:
1447 - Fixed a deadlock and an invalid free() in the threaded encoder.
1449 - Added support for symbol versioning. It is enabled by default
1450 on GNU/Linux, other GNU-based systems, and FreeBSD.
1452 - Use SHA-256 implementation from the operating system if one is
1453 available in libc, libmd, or libutil. liblzma won't use e.g.
1454 OpenSSL or libgcrypt to avoid introducing new dependencies.
1456 - Fixed liblzma.pc for static linking.
1458 - Fixed a few portability bugs.
1460 * xz --decompress --single-stream now fixes the input position after
1461 successful decompression. Now the following works:
1463 echo foo | xz > foo.xz
1464 echo bar | xz >> foo.xz
1465 ( xz -dc --single-stream ; xz -dc --single-stream ) < foo.xz
1467 Note that it doesn't work if the input is not seekable
1468 or if there is Stream Padding between the concatenated
1469 .xz Streams.
1471 * xz -lvv now shows the minimum xz version that is required to
1472 decompress the file. Currently it is 5.0.0 for all supported .xz
1473 files except files with empty LZMA2 streams require 5.0.2.
1475 * Added an *incomplete* implementation of --block-list=SIZES to xz.
1476 It only works correctly in single-threaded mode and when
1477 --block-size isn't used at the same time. --block-list allows
1478 specifying the sizes of Blocks which can be useful e.g. when
1479 creating files for random-access reading.
1482 5.1.1alpha (2011-04-12)
1484 * All fixes from 5.0.2
1486 * liblzma fixes that will also be included in 5.0.3:
1488 - A memory leak was fixed.
1490 - lzma_stream_buffer_encode() no longer creates an empty .xz
1491 Block if encoding an empty buffer. Such an empty Block with
1492 LZMA2 data would trigger a bug in 5.0.1 and older (see the
1493 first bullet point in 5.0.2 notes). When releasing 5.0.2,
1494 I thought that no encoder creates this kind of files but
1495 I was wrong.
1497 - Validate function arguments better in a few functions. Most
1498 importantly, specifying an unsupported integrity check to
1499 lzma_stream_buffer_encode() no longer creates a corrupt .xz
1500 file. Probably no application tries to do that, so this
1501 shouldn't be a big problem in practice.
1503 - Document that lzma_block_buffer_encode(),
1504 lzma_easy_buffer_encode(), lzma_stream_encoder(), and
1505 lzma_stream_buffer_encode() may return LZMA_UNSUPPORTED_CHECK.
1507 - The return values of the _memusage() functions are now
1508 documented better.
1510 * Support for multithreaded compression was added using the simplest
1511 method, which splits the input data into blocks and compresses
1512 them independently. Other methods will be added in the future.
1513 The current method has room for improvement, e.g. it is possible
1514 to reduce the memory usage.
1516 * Added the options --single-stream and --block-size=SIZE to xz.
1518 * xzdiff and xzgrep now support .lzo files if lzop is installed.
1519 The .tzo suffix is also recognized as a shorthand for .tar.lzo.
1521 * Support for short 8.3 filenames under DOS was added to xz. It is
1522 experimental and may change before it gets into a stable release.
1525 5.0.8 (2014-12-21)
1527 * Fixed an old bug in xzgrep that affected OpenBSD and probably
1528 a few other operating systems too.
1530 * Updated French and German translations.
1532 * Added support for detecting the amount of RAM on AmigaOS/AROS.
1534 * Minor build system updates.
1537 5.0.7 (2014-09-20)
1539 * Fix regressions introduced in 5.0.6:
1541 - Fix building with non-GNU make.
1543 - Fix invalid Libs.private value in liblzma.pc which broke
1544 static linking against liblzma if the linker flags were
1545 taken from pkg-config.
1548 5.0.6 (2014-09-14)
1550 * xzgrep now exits with status 0 if at least one file matched.
1552 * A few minor portability and build system fixes
1555 5.0.5 (2013-06-30)
1557 * lzmadec and liblzma's lzma_alone_decoder(): Support decompressing
1558 .lzma files that have less common settings in the headers
1559 (dictionary size other than 2^n or 2^n + 2^(n-1), or uncompressed
1560 size greater than 256 GiB). The limitations existed to avoid false
1561 positives when detecting .lzma files. The lc + lp <= 4 limitation
1562 still remains since liblzma's LZMA decoder has that limitation.
1564 NOTE: xz's .lzma support or liblzma's lzma_auto_decoder() are NOT
1565 affected by this change. They still consider uncommon .lzma headers
1566 as not being in the .lzma format. Changing this would give way too
1567 many false positives.
1569 * xz:
1571 - Interaction of preset and custom filter chain options was
1572 made less illogical. This affects only certain less typical
1573 uses cases so few people are expected to notice this change.
1575 Now when a custom filter chain option (e.g. --lzma2) is
1576 specified, all preset options (-0 ... -9, -e) earlier are on
1577 the command line are completely forgotten. Similarly, when
1578 a preset option is specified, all custom filter chain options
1579 earlier on the command line are completely forgotten.
1581 Example 1: "xz -9 --lzma2=preset=5 -e" is equivalent to "xz -e"
1582 which is equivalent to "xz -6e". Earlier -e didn't put xz back
1583 into preset mode and thus the example command was equivalent
1584 to "xz --lzma2=preset=5".
1586 Example 2: "xz -9e --lzma2=preset=5 -7" is equivalent to
1587 "xz -7". Earlier a custom filter chain option didn't make
1588 xz forget the -e option so the example was equivalent to
1589 "xz -7e".
1591 - Fixes and improvements to error handling.
1593 - Various fixes to the man page.
1595 * xzless: Fixed to work with "less" versions 448 and later.
1597 * xzgrep: Made -h an alias for --no-filename.
1599 * Include the previously missing debug/translation.bash which can
1600 be useful for translators.
1602 * Include a build script for Mac OS X. This has been in the Git
1603 repository since 2010 but due to a mistake in Makefile.am the
1604 script hasn't been included in a release tarball before.
1607 5.0.4 (2012-06-22)
1609 * liblzma:
1611 - Fix lzma_index_init(). It could crash if memory allocation
1612 failed.
1614 - Fix the possibility of an incorrect LZMA_BUF_ERROR when a BCJ
1615 filter is used and the application only provides exactly as
1616 much output space as is the uncompressed size of the file.
1618 - Fix a bug in doc/examples_old/xz_pipe_decompress.c. It didn't
1619 check if the last call to lzma_code() really returned
1620 LZMA_STREAM_END, which made the program think that truncated
1621 files are valid.
1623 - New example programs in doc/examples (old programs are now in
1624 doc/examples_old). These have more comments and more detailed
1625 error handling.
1627 * Fix "xz -lvv foo.xz". It could crash on some corrupted files.
1629 * Fix output of "xz --robot -lv" and "xz --robot -lvv" which
1630 incorrectly printed the filename also in the "foo (x/x)" format.
1632 * Fix exit status of "xzdiff foo.xz bar.xz".
1634 * Fix exit status of "xzgrep foo binary_file".
1636 * Fix portability to EBCDIC systems.
1638 * Fix a configure issue on AIX with the XL C compiler. See INSTALL
1639 for details.
1641 * Update French, German, Italian, and Polish translations.
1644 5.0.3 (2011-05-21)
1646 * liblzma fixes:
1648 - A memory leak was fixed.
1650 - lzma_stream_buffer_encode() no longer creates an empty .xz
1651 Block if encoding an empty buffer. Such an empty Block with
1652 LZMA2 data would trigger a bug in 5.0.1 and older (see the
1653 first bullet point in 5.0.2 notes). When releasing 5.0.2,
1654 I thought that no encoder creates this kind of files but
1655 I was wrong.
1657 - Validate function arguments better in a few functions. Most
1658 importantly, specifying an unsupported integrity check to
1659 lzma_stream_buffer_encode() no longer creates a corrupt .xz
1660 file. Probably no application tries to do that, so this
1661 shouldn't be a big problem in practice.
1663 - Document that lzma_block_buffer_encode(),
1664 lzma_easy_buffer_encode(), lzma_stream_encoder(), and
1665 lzma_stream_buffer_encode() may return LZMA_UNSUPPORTED_CHECK.
1667 - The return values of the _memusage() functions are now
1668 documented better.
1670 * Fix command name detection in xzgrep. xzegrep and xzfgrep now
1671 correctly use egrep and fgrep instead of grep.
1673 * French translation was added.
1676 5.0.2 (2011-04-01)
1678 * LZMA2 decompressor now correctly accepts LZMA2 streams with no
1679 uncompressed data. Previously it considered them corrupt. The
1680 bug can affect applications that use raw LZMA2 streams. It is
1681 very unlikely to affect .xz files because no compressor creates
1682 .xz files with empty LZMA2 streams. (Empty .xz files are a
1683 different thing than empty LZMA2 streams.)
1685 * "xz --suffix=.foo filename.foo" now refuses to compress the
1686 file due to it already having the suffix .foo. It was already
1687 documented on the man page, but the code lacked the test.
1689 * "xzgrep -l foo bar.xz" works now.
1691 * Polish translation was added.
1694 5.0.1 (2011-01-29)
1696 * xz --force now (de)compresses files that have setuid, setgid,
1697 or sticky bit set and files that have multiple hard links.
1698 The man page had it documented this way already, but the code
1699 had a bug.
1701 * gzip and bzip2 support in xzdiff was fixed.
1703 * Portability fixes
1705 * Minor fix to Czech translation
1708 5.0.0 (2010-10-23)
1710 Only the most important changes compared to 4.999.9beta are listed
1711 here. One change is especially important:
1713 * The memory usage limit is now disabled by default. Some scripts
1714 written before this change may have used --memory=max on xz command
1715 line or in XZ_OPT. THESE USES OF --memory=max SHOULD BE REMOVED
1716 NOW, because they interfere with user's ability to set the memory
1717 usage limit himself. If user-specified limit causes problems to
1718 your script, blame the user.
1720 Other significant changes:
1722 * Added support for XZ_DEFAULTS environment variable. This variable
1723 allows users to set default options for xz, e.g. default memory
1724 usage limit or default compression level. Scripts that use xz
1725 must never set or unset XZ_DEFAULTS. Scripts should use XZ_OPT
1726 instead if they need a way to pass options to xz via an
1727 environment variable.
1729 * The compression settings associated with the preset levels
1730 -0 ... -9 have been changed. --extreme was changed a little too.
1731 It is now less likely to make compression worse, but with some
1732 files the new --extreme may compress slightly worse than the old
1733 --extreme.
1735 * If a preset level (-0 ... -9) is specified after a custom filter
1736 chain options have been used (e.g. --lzma2), the custom filter
1737 chain will be forgotten. Earlier the preset options were
1738 completely ignored after custom filter chain options had been
1739 seen.
1741 * xz will create sparse files when decompressing if the uncompressed
1742 data contains long sequences of binary zeros. This is done even
1743 when writing to standard output that is connected to a regular
1744 file and certain additional conditions are met to make it safe.
1746 * Support for "xz --list" was added. Combine with --verbose or
1747 --verbose --verbose (-vv) for detailed output.
1749 * I had hoped that liblzma API would have been stable after
1750 4.999.9beta, but there have been a couple of changes in the
1751 advanced features, which don't affect most applications:
1753 - Index handling code was revised. If you were using the old
1754 API, you will get a compiler error (so it's easy to notice).
1756 - A subtle but important change was made to the Block handling
1757 API. lzma_block.version has to be initialized even for
1758 lzma_block_header_decode(). Code that doesn't do it will work
1759 for now, but might break in the future, which makes this API
1760 change easy to miss.
1762 * The major soname has been bumped to 5.0.0. liblzma API and ABI
1763 are now stable, so the need to recompile programs linking against
1764 liblzma shouldn't arise soon.