architecttap-arc 📋
A small TAP reporter with spec-like output, streaming, and failure diffing.
tap-arc on npmjs.org »
Contents:
Install
•
Development
•
FAQ
Objectives
- minimal, informative spec-like output for all assertion types
- helpful diffing for failures
- shallow graph of maintained dependencies
Installation and Usage
Compatible with Node.js v16+ -- v14 also works but is not recommended.
Save tap-arc as a development dependency:
npm i -D tap-arcSimply pipe tap output to tap-arc.
Example npm test script:
// package.json
"scripts": {
"test": "tape test/**/*.js | tap-arc"
}[!IMPORTANT]
💁tap-arcwill format output from any tap reporter.tapeis our favorite and was used for testing.
tap-arc --help
Usage:
tap-arc <options>
Parses TAP data from stdin, and outputs a "spec-like" formatted result.
Options:
-v | --verbose
Output full stack trace, TAP version, and plan
-p | --pessimistic | --bail
Immediately exit upon encountering a failure
example: tap-arc -p
--no-diff
Do not show diff for failed assertions
example: tap-arc --no-diff
--no-color
Output without ANSI escape sequences for colors
example: tap-arc --no-color
--fail-bad-count
Fail when the number of assertions parsed does not match the plan
example: tap-arc --fail-bad-countDevelopment
When building tap-arc, it's helpful to try various TAP outputs. See package.json "scripts" for useful "tap-arc.*" commands to test passing and failing TAP.
npm run tap-arc.simple # used to create the screen shot aboveDev Tips
- To see previous exit code, run:
echo $?Testing
Primarily, tap-arc is tested to output the correct exit code based on your test suite's TAP output.
Testing could be improved by unit testing the printer and diff maker.
FAQ
n tests, but found < n"</summary>
What happened?
✅ The TAP parser found zero failing tests
✅ The final tally from the raw TAP shows n of n passed
🤨 But the TAP plan called for more tests than were found, counted, and parsed.
💁♀️ Currently, when this case is detected, tap-arc will exit with a successful status code.
This can be overridden with the --fail-bad-count flag.
Why, though?
This has been observed specifically on Windows, where the TAP output is buffered to another stream and not piped to tap-arc.
Libraries like mock-fs tinker with stdout and subsequent TAP output is lost. Try closing those helpers before making an assertion that generates TAP.
Yes. At least one passing test is required to pass the suite.
This helps ensures there wasn't a silent, catastrophic failure in the test suite.
tap-arc get to decide these things?</summary>
tap-arc is responsible for the test suite's exit code. If your entire CI stack is piped to a reporter, it's an important job. So tap-arc is a bit skeptical by default to help ensure your suite is passing.
If you'd like to see different behavior from tap-arc, please open an issue or PR. We'd love to hear your use case.
Credit & Inspiration
- tap-spec ol' reliable, but a bit stale and vulnerable
- tap-difflet inspired output and diffing, also vulnerable