v2.4.1

Don't filter searching for PRs by "open" state to allow updating results on closed PRs in case a PR is closed before it's results are updated.

v2.4.0

Adds support for the workflow_run event, which is significant cuz now PRs from forks can get comments with benchmark results!

See the Readme for details on setting it up

Glob support & global status update

Allow globs in path input parameter #41

Add support for globs in the path input parameter. Using this feature does cause us to have to ignore the report-id parameter though, since each result file needs it's own report-id so it can be properly updated.

Update global status section when benchmarks are running #42

When benchmarks are running, update the global status section with a relevant message.

One limitation of this feature is that if multiple jobs are running, the first job to complete will remove the status message while the others are still running. In this situation, the other benchmarks should still have the the "⏱" status icon next to them, indicating that they are still running.

In the future if we remove support for each job reporting it's own results, the above issue will go away.

Summarize Option

Add summarize option (#36)

Allow the user to define which measurements from a benchmark should be summarized. Providing the value true means include all measurements in the summary. false means include none of the measurements, and a string of comma-separated measurement names means only include the names of these measurements in the summary (e.g. summarize: measure1, measure2).

Display unsure for no percentage change (#31)

Our formatting for percents uses .toFixed(0). In cases where we would show 0% - 0% but the actual result is actually not zero (i.e. -0.5 - 0.4) we'll now show the result as unsure to avoid a situation where we would display something like "slower ❌ 0% - 0% (0.00ms - 0.00ms)"

Log running version of tachomter-reporter-action (#39)

Log the full version of tachometer-reporter-action to make it easier to debug issues

Bug fixes and single benchmark support

• Reduce and simplify comment markup (#26)
• Properly handle measurement names (#25)
• Support single benchmark results (e5c0294)
• Improve tests and code coverage (e7c894a, 63a51bd, 10227e1, 3688adf, 92fc3e1, a75c421)

Include major version in comment footer

When upgrading to from v1 to v2, the action tries to merge the contents of the v2 comment into a v1 comment. Since the comment body format changed from v1 to v2 the merged results in the summary that doesn't look right. This release fixes that by adding the major version of the action in the comment footer, meaning an action will only see and update a comment of the same major version, preventing a v2 action from trying to merge into a v1 comment, keeping results looking nice.

Support for multi-measurement benchmarks

This release adds support for Tachometer's multi-measurements per benchmark. Because it requires changing our comment format in non-backwards compatible way, this is a major release.

Any existing PRs that merge in this version with v1 comments may get comments that look a little weird. If you'd like to fix them, just delete the comment and re-run the jobs. They will recreate the comment with the new format.

List of changes:

• Support reporting multiple measurements from one report (#22)
• Let GitHub auto markup commit to get hover effect (#21)
• Remove GitHub check code and experimental use-check option

v1.1.1

• Fallback to run html URL if job html URL is not available (#20)

v1.1.0

New Features

• Add initialize option (#12) (See Readme for description)
• Sort results by report title #15

Bug fixes

• 530f2d8 Increase min random create delay to 3s
• ce15b30 Skip initial create wait if createDelayFactor is 0
• e0b3202 Fix inifinite loop issue by adding missing state transition (Fixes #19)

Another bug fix

Only run action on pull_request event (Fixes #8)