Byron/gitoxide gix-testtools-v0.11.0

gix-testtools v0.11.0

on GitHub

Chore

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

New Features

• set_current_dir() to change the CWD and reset it to previous version on drop.

• allow execution of scripts without 'bash'.
  This works by trying to execute the file directly, and on failure, use 'bash'
  as interpreter.

  That way we are finally able to support a wider variety of fixture generators
  and make the crate more useful to a wieder audience.

• spawn_git_daemon() to spawn a git daemon hosting a working directoy…
  …with support for multiple at a time thanks to port selection (allowing
  tests to run in parallel) as well as auto-kill on drop.

• increase the waiting time on MacOS for file base locks
  It appears that these workers run into timeouts more and more,
  they got slower or maybe there are just more tests.

• add Env::unset() for convenience

• Provide GIT_VERSION information along with a way to skip a test based on it.

• expose on_ci in the top-level.

• Allow to re-execute scripts into temp directories.
  This is important in cases where the files created by the script
  contain absolute mentions of locations. That way, when copying
  files over, the test might accidentally return to the original
  read-only location, and write into it making future test runs fail.

• publicly accessible Result type

Bug Fixes

• gix-testtools use the latest dependencies
  These should compile properly.

• note that crates have been renamed from git-* to gix-*.
  This also means that the git-* prefixed crates of the gitoxide project
  are effectively unmaintained.
  Use the crates with the gix-* prefix instead.

  If you were using git-repository, then gix is its substitute.

• don't overwrite unexpanded git-lfs pointer files.
  It's possible for those with incomplete git-lfs installations
  (and many more situations) to end up in a spot where pointer files
  aren't expanded. If we overwrite the with archives, files look
  changed which can be confusing and lead to even bigger messes
  to happen.

  Now we don't overwrite those files anyomre.

• Disable tag.gpgSign in test scripts
  This is done for the same reason that commit.gpgsign is disabled for test
  scripts. It prevents test failures if the user has tag.gpgsign enabled in
  their global git config when invoking tests.

• Allow multiple scripts to run at the same time, if they are not the same.
  Previously, per integration test and thus per crate, one would
  effectively only be able to run a single script at a time because of the
  global identity lock. This was required previously before the additional
  file based lock, per script name, was introduced.

  This is now fixed by dropping the lock after the script identity was
  obtained.

• _with_args(…) functions now allow non-static strings

Other

• try to disable GPG signing with environment variables…
  …but it's not picked up at all even though it's definitely present.

Test

• ensure tests use 'merge.ff false' and recreate fixtures on each run

Changed (BREAKING)

• rename scripted_fixture_* to not contain 'repo' in the name.
  Further make clear in the documentation that bash is used to execute
  the fixture scripts, previously it wasn't even implied and got
  lost in history.
• upgrade bstr to 1.0.1

New Features (BREAKING)

• remove hex_to_id() and add various *_standalone() versions of existing methods.
  With these it's possible to handle fully standalone (as in with their own Cargo.toml) integration
  tests which are needed to resolve cyclic dependencies.

• 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.

Commit Statistics

• 246 commits contributed to the release over the course of 938 calendar days.
• 23 commits were understood as conventional.
• 20 unique issues were worked on: #198, #266, #293, #298, #301, #331, #364, #366, #382, #384, #391, #393, #427, #450, #470, #488, #509, #607, #650, #XXX

Thanks Clippy

Clippy helped 11 times to make code idiomatic.

Commit Details