jouet test

• Usage: jouet test [FLAGS] [EXTRA]…

Run the test suite, or only the tests a change affects

Without --affected, runs the repository's test task as the full suite: a mise test task, a package.json test script, Cargo.toml (cargo test --workspace), a justfile or Makefile test target or go.mod, else the whole suite of each test runner found. With --affected, runs only the tests that the changes since the merge base with --since (default: the upstream, else the default branch), uncommitted and untracked files included, affect. Adapters delegate to each runner's own selection: vitest's and jest's module graphs, pytest-testmon when it is installed, else a static import graph. The whole suite runs instead when a lockfile, tool versions, CI, the test configuration, a conftest or setup file, or a file no adapter can trace changed; nothing runs for docs only. Each test is listed with the chain of imports that picked it, and the command. Affected runs count as partial; they do not verify the full suite. Words after -- go to the runner.

With --cmd, the words after -- are the whole command instead, run directly rather than through a shell: one test, an e2e suite or a build, such as jouet test --cmd -- cargo build.

Every run's output, stdout and stderr together, goes to a log under jouet's log directory, kept for a week. For an agent (and in plain display), that is all it goes to: a passing run prints one line with the counts and the log's path, and a failing one each failure's name and the lines that explain it (the assertion, the panic, the compiler error), up to --budget bytes, then the names of the failures that did not fit. Read more from the log rather than running it again. For a person at a terminal, the output streams as it runs. Failures are recognised in the output of cargo test and nextest, rustc, tsc, go test and go build, pytest, unittest, jest, vitest and TAP (bats), and in the formats declared in a formats/ directory (your config's, a trusted repository's .jouet/, a plugin's) for other runners; for any other output, a failing run shows the end of it.

With --baseline, a failing run is run again in a temporary worktree of the merge base with the upstream or default branch (or --since REF). If every test that failed also fails there, the failures were there before the change, from the environment or an earlier commit, and the run passes, naming them; any new failure still fails it. It needs the output to name its failures, and costs one more run of the failed suites. What the merge base did is kept, so a later failure of the same command compares with it without running it again. With --affected, it replays the selected commands at the merge base without selecting tests again; a missing or uncomparable test does not count as an existing failure. Without --baseline, a failing full suite is compared on its own, and so is a failing --cmd run that took under a minute; a slower one uses what is kept for it, or names the --baseline command that compares it. When every failure is a Go test that go test named with its package, those tests alone run at the merge base first (go test -run '^(TestA|TestB)$' PKG...), however slow the run was; only when that does not show each one failing the same way does the comparison go on as above. [test] baseline = "off" (JOUET_TEST_BASELINE=off) turns that off.

A failing run whose every failing test is defined in a package the change does not reach (by the code graph: --affected selects nothing there, and no changed file is in it) says so on an unrelated: line, is not run at the merge base, and does not hold up finishing; it still exits 1.

Exits 1 when tests fail, 2 for a usage error (the runner's too), 4 when no runner is found or installed.

Arguments

• [EXTRA]… — Arguments for the test runner, after --; with --cmd, the command

Flags

• --affected — Run only the tests the changes affect
• --since <REF> — Measure changes from the merge base with REF (default: the upstream, else the default branch)
• --adapter <NAME> — Use only this adapter: vitest, jest, pytest, unittest, cargo or graph (Go, Ruby and Java, from the code graph)
• --dry-run — Print what would run and why, running nothing
• --cmd — Run the words after -- as the command, instead of the test task
• --baseline — When the run fails, run it again at the merge base (with --since, of REF) and pass it if every failure also fails there
• --fresh — Run the command even when result reuse is enabled and a matching run exists
• --budget <BYTES> — Bytes of failure text to show (default 4000; 0 shows every failure whole)
• -h --help — Print help