This release updates the method for SCRYPT key derivation by correcting a bug that caused KDF to fail on systems with low memory. Thanks to@AlleyPally for reporting and testing, and@Theelx for testing. A new commandline and conf file option,--costfactor has been added. See#165 for full details. Every effort was made to preserve decrypt capability with earlier lrzip-next versions. The lrzip-fe submodule was also updated to include the--costfactor option.

Contributors

This release incorporates compile time fixes to detect new bzip3 library ABI (>=1.5.0) where thebz3_decode_block function takes an additional argument. Thanks to@chungy for the report and@kspalaiologos for assistance and@AlleyPally for testing.

## Kludge to determine if new bzip3 library is used or not based on function name## bz3_decode_block()## Will allow for conditional compilation in bzip3_decompress_buf function.## Thanks @genivia-inc## Dr. Robert van EngelenAC_COMPILE_IFELSE([AC_LANG_PROGRAM([[@%:@include <libbz3.h>]],                  [[struct bz3_state *s=0; uint8_t *b=0; bz3_decode_block(s,b,0,0,0)]])],                  [AC_DEFINE([LIBBZ3_ABI1],,[LIBBZ3 ABI check])],                  []                  )

Contributors

Bugfix release. Thanks to@Theelx.

This release contains the latest lzma SDK 24.07. It adds a new BCJ filter, RISC-V. Also includes some modest bug fixes.

Magic header has changed slightly to allow for indepent storage of delta filter values. See docs.

Git organization reverting to more traditional master branch as opposed to separate branches for each new SDK release. Older releases can still be accessed, but if really needed, just checkout an older tag in master.

Contributors

This release adds the new PROCESSORS= variable for thelrzip.conf which permits the user to change the number of default processors detected by lrzip-next. Useful if calling lrzip-next from inside another program where options may not be able to be set or if the user does not want all processors to be used. May not always work as expected!

This is a bugfix release that corrects issues#88 and#151 .

This release fixes a low probability bug in ZSTD compression where possible errors are interpreted incorrectly. ZSTD return errors are negative, but ENUM values are positive. The error became apparent when data blocks were not compressible. See#148 . This release fixes that.

Now includes Debian package for x86_64.

This release fixes bug#144 which causes a double free error in theread_config() function. It only occured when there was no lrzip.conf file.

This release includes an enhancement to lzo backend compression where functionlzo1x_999_compress() is used when compression level 9 is selected. This level improves lzo compression for level 9 compared to the current. The submodulelrzip-fe has also been updated to include the--nobemt option.

This release includes an important bugfix which would cause decompression, testing, and checking of files to fail. All compressed volumes are OK. The fix (a case of me trying to overengineer a simple problem) permits decompression and checking to proceed normally.

THIS RELEASE HAS A DECOMPRESSION BUG. USE RELEASE 0.12.2!

This release introduces the--nobemt option which stands forNOBackEndMultiThreading. Issue#133 discusses this. For lzma, it will pass the threads parameter == 1 when used. The lzma backend will interpret this to NOT use the multi threaded match finder. This was designed for a particular use case, and will have worse performance generally. This option has no meaning for other backends currently.

Starting with this release, x86_64 binary file will be uploaded. This upload will only work with glibc 2.36. The binary is signed with my key 0xEB2C5812.