Release 0.41.0

Saturday, July 11, 2026

New features

Three flags give you finer control over a run, and each one composes with --parallel:

--retry <n> (or BASHUNIT_RETRY) re-runs a failed test up to n extra times and reports it as passed if any attempt passes. It fails only after every attempt fails, and a test that recovers on retry is annotated so the flakiness stays visible instead of being hidden. Works with --stop-on-failure too (#737).
--random-order shuffles the order of test files and of the tests within each file to surface hidden inter-test coupling. Pair it with --seed <n> (or BASHUNIT_SEED) to replay a specific shuffle; when you omit the seed, bashunit generates one and prints it in the run header (#738).
--shard <index>/<total> runs a deterministic, non-overlapping subset of the test files, so you can split one suite across several parallel CI runners (#739).

bash

bashunit --retry 2 tests/
bashunit --random-order --seed 12345 tests/
bashunit --shard 1/4 tests/   # runner 1 of 4

Alongside the streaming --output tap, you can now write reports to a file:

--report-json <file> writes results as JSON — summary counts plus per-test records — with no jq dependency (#741).
--report-tap <file> writes a TAP v13 report to a file, complementing the existing streaming output (#740).

bash

bashunit --report-json report.json --report-tap report.tap tests/

assert_file_permissions <mode> <file> — asserts a file's octal permission mode, portable across GNU and BSD stat (#742).
assert_array_length <n> <array> — asserts an array has exactly n elements (#743).
assert_within_delta <expected> <actual> <delta> — asserts a number is within a tolerance, floats included (#744).

The watch subcommand crashed with bashunit::watch::run: command not found in the released binary because src/watch.sh was missing from the build. It is now bundled (#735).

Faster test execution: subprocess forks were removed from the hot paths, with no change in behaviour.

See the full changelog for 0.41.0 on GitHub.