GitoxideLabs/gitoxide gix-filter-v0.35.0

gix-filter v0.35.0

on GitHub

Changed (BREAKING)

• migrate errors to gix-error

  rubberstamp

• raise MSRV to Rust 1.88

  The newly published dua-core 3.3 release used by linked-worktree removal
  requires Rust 1.88, so raise every workspace crate and the advertised badge
  together.

  Keep the MSRV checks buildable by selecting the latest sysinfo and rusqlite
  release lines that support Rust 1.88.

Bug Fixes

• reap long-running filter processes

  A gix-filter shutdown test deliberately selected the non-waiting mode, so cargo nextest reported its child as leaked. Unobserved State drops likewise closed the
  pipes without reaping the process.

  Keep shutdown(WaitForProcesses) as the explicit, status-reporting path and use
  it in all workspace callers. Drain the process map during explicit shutdown so
  State can also wait for any remaining children on Drop while ignoring errors.

  This matches Git’s sub-process.c cleanup: subprocess_exit_handler closes both
  pipes and calls finish_command() to wait for completion.

Bug Fixes (BREAKING)

• propagate long-running filter shutdown failures

  BREAKING CHANGE: State::shutdown() now takes &mut self and returns
  shutdown::Outcome; Mode::Ignore is renamed to Mode::Zombify.

  A long-running process filter can finish its protocol work and still exit
  unsuccessfully during teardown. Returning raw ExitStatus values made that
  failure easy to overlook, while consuming State made explicit cleanup awkward
  for owners that reuse a filter pipeline.

  Keep State reusable after shutdown(), return an Outcome that preserves
  every observed status, and add Outcome::into_result() to turn the
  first unsuccessful exit into a structured error. Rename Mode::Ignore to
  Mode::Zombify so opting out of wait() makes the resulting Unix process
  lifecycle explicit.

  Git considers a completed conversion successful even if a long-running
  process filter exits non-zero after EOF, so Outcome::into_result()
  isn't actually called.

Commit Statistics

• 11 commits contributed to the release over the course of 33 calendar days.
• 34 days passed between releases.
• 4 commits were understood as conventional.
• 1 unique issue was worked on: #2947

Commit Details