MatthewMckee4/karva

karva

Performance History

Latest Results

main

2 days ago

Bump semver from 1.0.27 to 1.0.28 Bumps [semver](https://github.com/dtolnay/semver) from 1.0.27 to 1.0.28. - [Release notes](https://github.com/dtolnay/semver/releases) - [Commits](https://github.com/dtolnay/semver/compare/1.0.27...1.0.28) --- updated-dependencies: - dependency-name: semver dependency-version: 1.0.28 dependency-type: direct:production update-type: version-update:semver-patch ... Signed-off-by: dependabot[bot] <support@github.com>

dependabot/cargo/semver-1.0.28

3 days ago

Match worker color decision to orchestrator's TTY Worker stdout is now a pipe, so the worker's TTY-based auto-detection always disables ANSI — colors got dropped on the way through the drain even when the orchestrator's stdout was a terminal. Forward an explicit --color=always or --color=never to every worker based on the orchestrator's own should_colorize() check (or the user's --color override if one was given). The drain pipes bytes through verbatim, so the orchestrator-side TTY decision is the one that should drive colorization.

worktree-fluttering-growing-rose

6 days ago

Pipe worker stdout into the drain to fix print interleaving Workers were spawned with Stdio::inherit(), so any byte they wrote directly to FD 1 — Python print() in a user test, panic messages, etc. — landed on the orchestrator's terminal without going through the file-based output drain. When that hit FD 1 in the middle of an orchestrator writeln!, the worker's PASS line came out spliced through foreign bytes (issue #502 in another guise). Spawn workers with Stdio::piped() for stdout, take the read end at spawn time, and hand it to a per-worker reader thread inside OutputDrain. The reader pushes whole lines through an mpsc channel that drain_loop polls alongside the per-worker output files; both sources emit through the same bar.suspend() path so nothing tears against the bar either. Stderr stays inherited because the orchestrator also writes tracing to stderr in real time and the drain's poll interval would shuffle the relative order of those lines. Two follow-on fixes were needed for ordering: The drain now reads the pipe channel before polling the result files in each iteration. A print() runs before the test returns and the reporter writes its PASS line, so within one iteration the pipe line should land in the emit batch ahead of the file line. attach_with_output now reconfigures sys.stdout / sys.stderr to line-buffer when show_output is true. Python defaults to block buffering when its stdout isn't a TTY; with the pipe in place that delayed every print() until the interpreter shut down, which is after the reporter had already emitted PASS.

worktree-fluttering-growing-rose

6 days ago

Tick progress per test function, not per parametrize variant

worktree-fluttering-growing-rose

6 days ago

Serialize worker output via per-worker file drain

worktree-fluttering-growing-rose

6 days ago

Add required-version field to configuration (#771) ## Summary Closes #579. Adds a top-level `required-version` field to `karva.toml` and the `[tool.karva]` table in `pyproject.toml`. The value is a SemVer requirement; if the installed karva does not satisfy it, karva exits with a clear error before any tests run, which avoids the confusing failures that show up when CI or a teammate is on an older version that does not understand newer config keys. ```toml required-version = ">=0.5.0" ``` The check runs at every config-loading entry point (`karva.toml`, `--config-file` / `KARVA_CONFIG_FILE`, and a discovered `[tool.karva]` table). Parsing uses `semver::VersionReq`, so an invalid specifier surfaces as a regular TOML parse error pointing at the offending value. Because `required-version` is not part of any profile, it is special-cased in the doc generator so the configuration reference still gets the section even though the metadata-driven walker only sees per-profile options. ## Test Plan ci

main

6 days ago

Add --show-progress for live progress display (closes #570)

worktree-fluttering-growing-rose

7 days ago

Latest Branches

-3%

Bump semver from 1.0.27 to 1.0.28#774

3 days ago

8248413

dependabot/cargo/semver-1.0.28

-12%

Add --show-progress and serialize worker output to fix interleaving#773

6 days ago

6fe0775

worktree-fluttering-growing-rose

+10%

Add per-test configuration overrides#770

7 days ago

40c3e12

worktree-nested-wandering-acorn

Home Terms Privacy Docs