v0.5.2

What's Changed

• add a fuzzer that compresses the gzip header by @folkertdev in #384
• distribute zlib.h and zconf.h by @folkertdev in #385
• fix valgrind issue in test by @folkertdev in #387
• Export crc32_combine64 and adler32_combine64 for 32bit architectures by @jakos-sec in #386
• Correct naming of zlib-sys to libz-sys by @erikjee in #390
• expose the 64-bit versions of some gz functions by @folkertdev in #388
• extract SymBuf by @folkertdev in #391
• libz-rs-sys-cdylib: include LICENSE by @j-vanderstoep in #394
• clippy fixes for 1.88.0 by @folkertdev in #395
• fix valgrind install command by @folkertdev in #398
• Strengthen the lint directives related to unsafe code by @brian-pane in #396
• Export crc32_z and adler32_z by @jakos-sec in #399
• Declare BitReader::refill unsafe and document safety prerequisites by @brian-pane in #397
• expose the allocators as statics by @folkertdev in #402
• clippy by @folkertdev in #405
• Ci cleanup by @folkertdev in #406
• work around avx512vl miscompilation by @folkertdev in #408
• export get_crc_table by @folkertdev in #409

New Contributors

• @jakos-sec made their first contribution in #386

Full Changelog: v0.5.1...v0.5.2

v0.5.1

Highlights

• we now have experimental support for the gz* functions when building a C dynamic library. See the README of libz-rs-sys-cdylib. (most of the work here was done by @brianpane
• fixed a bug where uninitialized memory was read (but not used) during decompression, see #346
• fixed some logic bugs in the construction of the gzip header (unlikely to come up in practice)
• performance improvements for compression and decompression
• we support inflateCodesUsed, inflateGetDictionary and deflateGetDictionary

What's Changed

• remove CI usage of ubuntu-20.04, just use ubuntu-latest by @folkertdev in #330
• Partial implementation of gzopen and gzclose by @brian-pane in #326
• Add Linux-specific gzopen options and more tests by @brian-pane in #332
• Add some comments to gzopen_help and reduce the amount of unsafe code by @brian-pane in #333
• Fix unsafe_op_in_unsafe_fn warning in AArch64 CRC32 implementation by @brian-pane in #334
• bump codecov action version by @folkertdev in #336
• Finish implementing gz_open and add gzdopen and gzerror by @brian-pane in #335
• Add gzclearerr and gzeof functions by @brian-pane in #339
• Micro-optimization for decompression by @mat-1 in #320
• use cargo nextest with cargo llvm-cov to gather the test coverage by @folkertdev in #338
• ignore coverage of test files by @folkertdev in #342
• run the uncompress fuzzer with a custom corpus and gather fuzzer codecov by @folkertdev in #337
• stop using custom wasm simd for slide_hash by @folkertdev in #343
• Implement gzdirect by @brian-pane in #341
• Copy chunks as MaybeUninit<u8> by @folkertdev in #346
• Implement gzread and gzbuffer by @brian-pane in #345
• Implement gzwrite and gzflush by @brian-pane in #347
• Fuzz coverage error paths by @folkertdev in #344
• Make gzclose_r and gzclose_w part of the public API by @brian-pane in #348
• Implement gzoffset and gztell by @brian-pane in #349
• Implement gzputc and gzputs by @brian-pane in #351
• Implement gzgetc and gzgets by @brian-pane in #352
• Implement gzfread by @brian-pane in #353
• Implement gzfwrite and add null buffer check to gzfread by @brian-pane in #354
• Implement gzungetc by @brian-pane in #355
• Implement gzsetparams by @brian-pane in #356
• Implement gzseek for reads, as well as gzrewind by @brian-pane in #358
• Implement gzseek for writes by @brian-pane in #359
• check for and enable the bmi1 and bmi2 target feature by @folkertdev in #362
• Additional gz test coverage by @brian-pane in #363
• use more idiomatic way to create an integer out of a repeating byte by @folkertdev in #364
• use an autovectorized slide_hash_chain on neon by @folkertdev in #366
• add inflateCodesUsed by @folkertdev in #350
• Hard-code the hash function for the quick deflate algorithm by @brian-pane in #369
• faster compare256 aarch64 neon implementation by @Dr-Emann in #371
• Move gz API to libz-rs-sys-cdylib by @brian-pane in #365
• Cache locality improvement for deflate State.lookahead by @brian-pane in #372
• Hard-code the hash selection for the fast and medium deflate algorithms by @brian-pane in #373
• Optimize out some read operations from the quick deflate algorithm by @brian-pane in #374
• Optimize out some read operations from the fast deflate algorithm by @brian-pane in #375
• Reserve half of the GzState input buffer space for gzprintf. by @brian-pane in #378
• run the zlib C tests on our cdylib by @folkertdev in #377
• Fix import resolution in crc32 for the Android build by @j-vanderstoep in #379
• add gzprintf by @folkertdev in #376
• add inflateGetDictionary and deflateGetDictionary by @folkertdev in #380
• Use core::mem::transmute instead of std in wasm adler32 by @storo-dfend in #382
• cdylib: use core instead of std by @folkertdev in #383
• fix a bug in the value of the FTEXT field of the gz header by @folkertdev in #381

New Contributors

• @mat-1 made their first contribution in #320
• @Dr-Emann made their first contribution in #371
• @j-vanderstoep made their first contribution in #379
• @storo-dfend made their first contribution in #382

Full Changelog: v0.5.0...v0.5.1

v0.5.0

This is technically a breaking change, but unless you rely on rust exporting C symbols, you should not run into any issues when upgrading.

highlights

• the export-symbols feature now controls whether C symbols are exported. This feature is off by default (that is why this is a breaking change; the behavior of earlier versions was to export those symbols). For rust users that is almost certainly the correct default, because it means zlib-rs can be used together with stock zlib or zlib-ng in the same project. This should fix some issues with symbol clashes between libz-sys (used e.g. by openssl) and zlib-rs.
• the cdylib crate now uses panic=abort. For C users, that is the only safe way to panic, because unwinding to C is unsound.
• Some simplifications to simd code: in a number of cases, LLVM optimizes safe rust code to the equivalent assembly that we achieved with explicit simd before.

What's Changed

• make the fallback implementation of slidehash vectorize better by @folkertdev in #296
• Clippy 1.85.0 fixes by @folkertdev in #299
• the standard slide_hash implementation vectorizes well on avx2 by @folkertdev in #298
• remove explicit SIMD reads and writes by @folkertdev in #303
• Store mode on the stack in dispatch by @bjorn3 in #304
• add Cargo.lock by @folkertdev in #308
• zlig-ng -> zlib-ng by @folkertdev in #310
• use a more straightforward bit counting operaration by @folkertdev in #311
• use safe chunking in wasm32/aarch64 compare256 by @folkertdev in #312
• Inflate cleanups by @folkertdev in #313
• Show sponsor button by @erikjee in #316
• add semver-prefix feature (like we have for libbz2-rs-sys) by @folkertdev in #318
• explicitly document memory management by @folkertdev in #317
• Loop match preparations by @folkertdev in #319
• add export-symbols feature flag by @folkertdev in #322
• add panic=abort to the cdylib by @folkertdev in #323
• run clippy for more scenarios by @folkertdev in #325
• add zlibCompileFlags by @folkertdev in #324
• refactor zeroed allocation api by @folkertdev in #327
• fix alignment in zeroed allocations by @folkertdev in #328
• investigate CI failure by @folkertdev in #329

Full Changelog: v0.4.2...v0.5.0

v0.4.2

Lots of performance improvements. We're now substantially faster than other implementations for inflate (decompression), and faster in the most important cases for compression.

What's Changed

• Speed up emit_dist with precomputed length codings for static trees by @brian-pane in #264
• Add a comment to explain how Crc32HashCalc is sometimes Crc32C by @brian-pane in #266
• Fix the usage message for crc32_bench by @brian-pane in #267
• Add some comments explaining the chained hash data structure. by @brian-pane in #269
• Cleanups for struct State by @brian-pane in #274
• Begin converting match offsets in struct State to u16 by @brian-pane in #275
• Remove the cloning of scan_start & scan_end in longest_match_help by @brian-pane in #268
• Simplify deflate window by @folkertdev in #272
• cleanup of fill_window by @folkertdev in #271
• Add checks for State memory layout by @brian-pane in #278
• Update zlib-ng version 2.2.3 by @folkertdev in #282
• Move the TreeDesc fields and finish adding cache line markers to State by @brian-pane in #279
• Make various fields in the State structure smaller by @brian-pane in #281
• Hint to the compiler to inline the fill_window function by @brian-pane in #270
• Remove the padding from the State structure by @brian-pane in #283
• Optimize away compare256 for short matches by @brian-pane in #284
• Faster comparisons for priority queue by @brian-pane in #285
• Fix CI by @bjorn3 in #288
• Add CI benchmarking by @bjorn3 in #287
• fuzzer: make sure the buffer is large enough to store the input by @folkertdev in #286
• compare our and stock zlib compression levels in README by @folkertdev in #291
• specialize inflate_fast_help for avx2 by @folkertdev in #289
• gather benchmark data for chromium (only inflate is relevant) by @folkertdev in #290
• manual struct target feature by @folkertdev in #292
• Make the matches field smaller to improve cache locality by @brian-pane in #293
• apply the crc target feature to the acle crc32 implementation by @folkertdev in #294

Full Changelog: v0.4.1...v0.4.2

Version 0.4.1

A minor release, but with some nice goodies.

We now support building C-compatible dynamic libraries with cargo c. The readme of the libz-rs-sys-cdylib crate has all the details.

Additionally, performance has improved drastically. We fixed an incorrect loop bound and started using the enable-dfa-jump-thread llvm flag in our benchmarks, which give inflate performance that is consistently better than zlib-ng. Meanwhile @brian-pane has been making some smaller improvements to deflate performance, where we already beat zlib-ng in some, but not yet in all cases.

What's Changed

• run the flate2 test suite on CI by @folkertdev in #250
• Add a release checklist by @rnijveld in #251
• Revert "load the inflate reader and writer to the stack" by @folkertdev in #252
• Add the metadata needed by cargo-c by @lu-zero in #176
• rust 1.83.0 clippy fixes by @folkertdev in #254
• Avoid an out-of-bounds panic in deflate_stored by @bjorn3 in #255
• cdylib cleanup by @folkertdev in #256
• inflate: make use of enable-dfa-jump-thread by @folkertdev in #257
• Parallelize the checking of the first two bytes of a potential match. by @brian-pane in #259
• Fix inflate_fast_help loop bound by @folkertdev in #260
• Small additional speedup for deflate_quick by @brian-pane in #262
• Inline the fast path of BitWriter::send_bits by @brian-pane in #258
• release version 0.4.1 by @folkertdev in #265

New Contributors

• @lu-zero made their first contribution in #176
• @brian-pane made their first contribution in #259

Full Changelog: v0.4.0...v0.4.1

v0.4.0

What's Changed

• Clippy update by @squell in #178
• It is time to have a security policy by @squell in #177
• wasm: add a compare256 wasm simd implementation by @folkertdev in #179
• inline copy_match_help by @folkertdev in #180
• inflate: use custom Writer by @folkertdev in #181
• simplify struct Table by @folkertdev in #182
• Inflate cache opts by @folkertdev in #183
• add chunked and adler32 measurements by @folkertdev in #184
• unroll copy_chunk_unchecked by @folkertdev in #185
• cleanup Writer api usage by @folkertdev in #186
• Extend from window by @folkertdev in #187
• add specific tests for copy_match by @folkertdev in #188
• remove adler32 COPY variant (memcpy is faster) by @folkertdev in #189
• remove casting noise around drop_bits by @folkertdev in #191
• Writer tuning by @folkertdev in #192
• remove libz-sys dependency from zlib-rs crate by @folkertdev in #194
• use wider loads/stores in Writer on aarch64 by @folkertdev in #196
• wasm: use wider loads/stores in copy_match by @folkertdev in #197
• wasm: SIMD adler32 by @folkertdev in #198
• wasm: SIMD slide_hash by @folkertdev in #199
• wasm: allow unaligned reads in longest_match by @folkertdev in #202
• fix UB in inflate::State::extra by @folkertdev in #203
• Reenable deflate::test::insufficient_compress_space under Miri by @inahga in #204
• move the repository (and fix links) by @folkertdev in #208
• Add Prossimo as initiator by @erikjee in #209
• fix for reading/writing uninitialized memory by @folkertdev in #211
• try this codecov thing by @folkertdev in #215
• fix bug in inflate_fast when refilling by @folkertdev in #216
• run more deflate:: tests under miri by @folkertdev in #214
• set head.done = -1 if a gzip header was configured but the input is not gzip data by @folkertdev in #217
• Run test-libz-rs-sys tests on wasm in CI by @bjorn3 in #220
• fix panic in the stored algorithm by @folkertdev in #221
• fix panic in crc32 hash calc by @folkertdev in #222
• optimize the medium algorithm by @folkertdev in #223
• Replace usage of LD_PRELOAD with LD_LIBRARY_PATH by @bjorn3 in #227
• Fix a couple of warnings when building for wasm by @bjorn3 in #226
• Stop passing -fno-pic when building for wasm by @bjorn3 in #225
• fix 2 memory leaks in deflate tests by @folkertdev in #228
• gzip: make header functions unsafe and document preconditions by @inahga in #235
• just always initialize the window buffer with zeros by @folkertdev in #234
• crc32: mark some unsafe functions properly, add some safety documentation by @inahga in #236
• Miri stacked borrows improvements for inflate by @folkertdev in #237
• longest_match: handle underflow in assertion by @inahga in #240
• run deflate tests under stacked borrows by @folkertdev in #239
• adler32: document some unsafe blocks by @inahga in #238
• fuzzing: inflate fuzz suite for catching UB by @inahga in #231
• inflate/bitreader: fix unsoundness in advance() by @inahga in #242
• deflate: safety documentation, relax safety requirements of wasm intrinsics by @inahga in #241
• follow MaybeUninit::slice_assume_init_ref implementation by @folkertdev in #243
• inflate: some safety documentation by @inahga in #244
• libz-rs-sys: more safety requirements on inflate() and deflate() by @inahga in #245
• add section on using; change spelling to US by @squell in #247
• Loop plus match by @folkertdev in #248

New Contributors

• @squell made their first contribution in #178
• @inahga made their first contribution in #204
• @erikjee made their first contribution in #209

Full Changelog: v0.3.0...v0.4.0

v0.3.1

What's changed

• fix out-of-bounds write in Pending by @folkertdev
• fix UB in inflate::State::extra by @folkertdev
• backport ptr aliasing bugfix by @folkertdev

Full Changelog: v0.3.0...v0.3.1

Version 0.3.0

What's Changed

• add custom-prefix feature flag by @folkertdev in #125
• silence some warnings in dynamic-libz-sys by @folkertdev in #128
• use zlib-rs as a dynamic c library by @folkertdev in #127
• Update README.md by @rnijveld in #129
• Symbol visibility fixes by @folkertdev in #130
• add ROADMAP.md by @folkertdev in #131
• Simplify test suite 1 by @folkertdev in #132
• move libz-rs-sys tests into its own crate by @folkertdev in #133
• gate the crc32_hash_calc test on x86, x86_64 and aarch64 by @folkertdev in #137
• fix test suite on s390x-unknown-linux-gnu by @folkertdev in #138
• test i686 and s390x on CI by @folkertdev in #139
• clean up hash_calc selection logic by @folkertdev in #140
• Ci cross compile by @folkertdev in #141
• centralize fuzzing helpers by @folkertdev in #142
• guarantee that the alloc/free function pointer always point to the same thing by @folkertdev in #144
• handle NULL in adler32 and crc32 functions by @folkertdev in #145
• add safety comments & tests to (un)compress by @folkertdev in #146
• tests/comments for foundational inflate functions by @folkertdev in #147
• document the rest of the inflate functions by @folkertdev in #148
• Deflate docs by @folkertdev in #149
• test edge cases in the public C api by @folkertdev in #150
• handle partially uninitialized streams properly by @folkertdev in #151
• cover some uncovered paths by @folkertdev in #152
• switch the instruction used for acle crc32 calculation by @folkertdev in #154
• add more tests for inflateGetHeader by @folkertdev in #153
• mirror zlib-ng behavior for i686 by @folkertdev in #155
• mod read_buf: remove unused methods by @folkertdev in #158
• refinements to unsafe code by @folkertdev in #157
• correct target features (based on miri) by @folkertdev in #159
• add target features for the SIMD compare256 operations by @folkertdev in #160
• correct target features in hash_calc by @folkertdev in #161
• quick algorithm quick cleanup by @folkertdev in #162
• fix bug in the quick algorithm (Z_BEST_SPEED) by @folkertdev in #165
• centralize cpu feature detection by @folkertdev in #167
• fix oversight in one of the quick deflate macros by @folkertdev in #170
• add a separate cdylib crate by @folkertdev in #168
• Use the "C-unwind" abi for functions we cannot guarantee won't panic by @folkertdev in #171
• fix issue 172 by @folkertdev in #173
• remove unneeded slice operation by @folkertdev in #175

Full Changelog: v0.2.1...v0.3.0

Version 0.2.1

What's Changed

• Fix inflate() flush modes and implement state.data_type by @cjgriscom in #113
• stop using crc32 hash on aarch64 by @folkertdev in #112
• fix inflate::uncompress bug by @folkertdev in #116
• include LICENSE file in published crates by @decathorpe in #118
• inflate::reset_keep: clear more fields by @folkertdev in #121
• Refactor dynamic loading by @folkertdev in #122
• make libz-rs-sys and zlib-rs use their local README.md for crates.io by @folkertdev in #120

New Contributors

• @cjgriscom made their first contribution in #113
• @decathorpe made their first contribution in #118

Full Changelog: v0.2.0...v0.2.1

Version 0.2.0

• Fix bug with window padding
• Add no_std support