v4.0.0-alpha.9

Alpha 9 introduces some new features, and also important fixes and optimizations, especially related to FileSystems and tar archives.

⚠️ This release has some breaking changes to the API, but fairly minor all things considered. Be advised that the next release may have a new package name: archives. See #427.

• Using fs.WalkDir() is now much faster when walking large tar archives.
• A FileSystem can be created from a stream instead of just a file name (but the stream must be a Reader, ReaderAt, and Seeker).
• Added lzip compression.
• Renamed CompressedArchive to Archive since compression is optional.
• Split Archival into Archival and Extraction, since not all archive formats can archive (some, like rar and 7z, can only extract).
• Removed the filename filter argument from Extract() since it was mostly pointless and even confusing.
• Brotli-compressed files can now be identified by their stream, maybe... sometimes (it's worth a shot; brotli has no notable magic number or header so we just try our best).
• Removed ErrStopWalk in favor of fs.SkipAll which is apparently the same thing
• Renamed ErrNoMatch to NoMatch since it is a sentinel (error) value and not an actual error.

What's Changed

• Bugfix: Tar files created in their target directory shouldn't recurse infinitely by @drewstinnett in #384
• zlib format incorrectly matches on ASCII files starting with the letter x by @dpgarrick in #386
• Option to use number for user and/or group names by @breezerider in #385
• Fix create archive to a continuous writing source file failed by @halfcrazy in #388
• 7z: Fix iteration bug by @SheltonZhu in #394
• Add lzip support by @sorairolake in #401
• Expose gzip.Reader.Multistream by @rgmz in #407
• zip: Fix compression method not set without SelectiveCompression enabled (closes #418) by @hevav in #419
• chore(go.mod): Switch to upstream lzip package by @sorairolake in #422
• Upgrade rardecode to v2.0.0-beta.3 by @rgmz in #423
• Refactor FS types; improve performance by @mholt in #426

New Contributors

• @drewstinnett made their first contribution in #384
• @dpgarrick made their first contribution in #386
• @breezerider made their first contribution in #385
• @SheltonZhu made their first contribution in #394
• @sorairolake made their first contribution in #401
• @rgmz made their first contribution in #407
• @hevav made their first contribution in #419

Full Changelog: v4.0.0-alpha.8...v4.0.0-alpha.9

v4.0.0-alpha.8

This release adds support for reading 7zip archives and contains several significant fixes. Thank you to all who contributed!

What's Changed

• Retain underlying reader if it's an io.Seeker by @jhwz in #327
• Enhance FilesFromDisk (close #331) by @mholt in #332
• fs: Improve Open and Stat on ArchiveFS for streams by @mholt in #335
• Fix nested directory listing for archives without explicit directory names. by @jeremyje in #339
• Document caveats with compressed fs.FS by @WeidiDeng in #344
• return zip.Reader in FileSystem if file is a zip file by @WeidiDeng in #349
• ignore empty root dir name when compressing file by @WeidiDeng in #355
• rewrite file matching for fs Open, ReadDir and Stat by @WeidiDeng in #354
• add Archive method to SevenZip by @nikolaymatrosov in #364
• zip: fix File.Open not working if called after zip.Extract completes by @ncw in #367
• Allow Identify to take a nil io.Reader by @ncw in #370
• Rework ArchiveAsync interface to make it return any archiving errors by @ncw in #369
• Bump golang.org/x/text from 0.3.7 to 0.3.8 by @dependabot in #372

New Contributors

• @jeremyje made their first contribution in #339
• @WeidiDeng made their first contribution in #344
• @nikolaymatrosov made their first contribution in #364
• @ncw made their first contribution in #367
• @dependabot made their first contribution in #372

Full Changelog: v4.0.0-alpha.6...v4.0.0-alpha.8

v4.0.0-alpha.6

This release focuses some attention on Identify(), which has a slightly modified signature; please adjust programs accordingly. It can now accept any io.Reader and no longer requires Seek(); however, it returns three values including a new Reader that can re-read the buffered bytes consumed during identification. Callers should use that returned Reader after calling Identify().

What's Changed

• Fix for Identify() failing on empty and small files: by @congop in #319
• Change Identify API to accept only an io.Reader by @jhwz in #322

New Contributors

• @congop made their first contribution in #319
• @jhwz made their first contribution in #322

Full Changelog: v4.0.0-alpha.5...v4.0.0-alpha.6

v4.0.0-alpha.5

A couple minor enhancements (including ArchiveAsync()). Thanks for your participation and contributions!

Full Changelog: v4.0.0-alpha.4...v4.0.0-alpha.5

v4.0.0-alpha.4

A few more bug fixes and enhancements. Thanks for your participation and contributions.

Full Changelog: v4.0.0-alpha.3...v4.0.0-alpha.4

v4.0.0-alpha.3

Bug fixes and minor-moderate enhancements. Things are looking even better now! Thanks for your participation and contributions.

Full Changelog: https://github.com/mholt/archiver/compare/v4.0.0-alpha.2..v4.0.0-alpha.3

v4.0.0-alpha.2

A few enhancements and bug fixes.

v4.0.0-alpha.1

v4 is a complete rewrite with a simpler, stream-oriented API. The arc command has yet to be implemented for v4 (please use v3 for now, if you need that).

PR #302 documents the many issues fixed/closed by this new version.

Some of the major new features:

• Stream-oriented APIs (no writing to disk!)
• Simpler APIs (less code and docs!)
• File system abstraction (treat archives like directories!)
• Format identification (by filename or header!)

Many other improvements and simplifications have been made.

The v4 godoc is live. See the README for a tour and code examples. Please try it out and leave your feedback!

v3.5.1

v3.5.1

v3.5.0

Changelog

25e050d add a comment explaining backwards compatibility
b736c51 add missing and remove unused modules
ca7993e port to github/pierrec/lz4/v4