Byron/gitoxide gix-refspec-v0.7.2

gix-refspec v0.7.2

on GitHub

New Features (BREAKING)

• upgrade edition to 2021 in most crates.
  MSRV for this is 1.56, and we are now at 1.60 so should be compatible.
  This isn't more than a patch release as it should break nobody
  who is adhering to the MSRV, but let's be careful and mark it
  breaking.

  Note that git-features and git-pack are still on edition 2018
  as they make use of a workaround to support (safe) mutable access
  to non-overlapping entries in a slice which doesn't work anymore
  in edition 2021.

Changed (BREAKING)

• remove lifetime of match_group::Fix, keeping RefSpec instances instead
  That lifetime unnecessarily complicated things and wasn't worth keeping
  due to being a premature optimization.
• reject all invalid negative refspec patterns.
  Git is more lenient, but will then fail to match against such patterns
  which seems like avoidable surprising behaviour.
• upgrade bstr to 1.0.1

New Features

• add RefSpecRef::expand_prefix() method to avoid missing prefixes.
  The current implementation might cause refspecs to end up ignored as
  they don't have a prefix, and in protocol V2 it would then fail
  to add a ref-spec filter which causes them to be missed.

  With expand_prefix(), we assure that there are all possible prefixes
  that can contain partial names, similar to what git does.

• RefSpec::allow_non_fast_forward() to get information about 'force' quickly.

• RefSpecRef::prefix() to return the two-component prefix of a refspec's source. #(450)

• Simple serialization for Instruction and RefSpecRef type.
  It's also a way to normalize input strings as there is only one way
  to serialize instructions, which themselves are already normalized
  towards what's possible.

Chore

• uniformize deny attributes
• remove default link to cargo doc everywhere

Documentation

• fix typos

Bug Fixes

• Fixes build for parsing fuzz harness
• HEAD may now return itself as prefix in RefSpecRef::prefix() and expanded_prefix().
  Previously, the expanded prefix would be a list of possibilities, even
  though it's such a common case that we really want to avoid spamming the
  remote about it when asking for HEAD during clone for instance.
• RefSpecRef instruction uses the correct lifetime.

Commit Statistics

• 233 commits contributed to the release over the course of 196 calendar days.
• 14 commits were understood as conventional.
• 5 unique issues were worked on: #450, #470, #571, #691, #XXX

Thanks Clippy

Clippy helped 10 times to make code idiomatic.

Commit Details