Release 0.33.0

Sunday, February 15, 2026

🔧 New features

Test output on failure

When a test fails due to a runtime error, bashunit now displays the captured stdout/stderr output directly alongside the failure, making it much easier to diagnose what went wrong.

Output

✗ Failed: test_should_parse_config
  Expected '0' but got '1'
  stdout/stderr:
    parse_config: line 12: unexpected token '}'

Disable

bashunit --no-output-on-failure tests/
# or
BASHUNIT_SHOW_OUTPUT_ON_FAILURE=false bashunit tests/

This is enabled by default. Disable it with --no-output-on-failure or the BASHUNIT_SHOW_OUTPUT_ON_FAILURE=false environment variable.

Suppress progress display

A new --no-progress flag suppresses real-time progress output, useful for CI environments or when piping output:

Usage

bashunit --no-progress tests/
# or
BASHUNIT_NO_PROGRESS=true bashunit tests/

Opt out of test-level parallelism

Add the # bashunit: no-parallel-tests directive at the top of a test file to run its tests sequentially, even when using --parallel:

Usage

#!/usr/bin/env bash
# bashunit: no-parallel-tests

function test_must_run_sequentially() {
  # Tests in this file won't be parallelized
  ...
}

This is useful for test files with shared state or ordering requirements.

⚡ Improvements

Bash 3.0 support

The minimum Bash version requirement has been lowered from 3.2 to 3.0, broadening compatibility with older systems.

Faster parallel execution

Parallel test execution is now 30-40% faster on large test suites thanks to internal optimizations in the parallel runner.

🐛 Bug fixes

• Data providers now work without the function keyword on test functions
• Install tests now pass in sandboxed/no-network environments
• Parallel test execution now works correctly in strict mode (bash -e -o pipefail)

---

See the full changelog on GitHub