You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Should results always be reported as a check in addition to the comment (if possible)
Should results only be reported as a check if comment failed?
Should results be reported as a check based on an action input?
Some combination of the above?
Right now I'm leaning towards 3.
IIRC, I think using the Checks API inside of an action will result in an additional check being reported in the checks list. In other words, if you run this action in a workflow called "benchmarks" and the check is called "benchmarks check", then in the list of checks at the bottom of the PR will include a "benchmarks" line item (representing that the workflow successfully completed) and a "benchmarks check" line item for the check the workflow manually created.
That duplicate line item probably doesn't impact lit-html as much since it runs all of it benchmarks in one workflow, but for Preact (which runs each benchmark in its own workflow) it would mean we'd have 8 line items related to benchmarks compared to 4. Hence I'm leaning to 3.
Add an option use-check that takes 3 values: always, never, on-forks with the default being on-forks.
always: Always report as GitHub check
never: Never report as GitHub check
on-forks: If PR is opened is a fork, then report as check
To me, this seems to be the best cause it provides a smart default for people, but gives them the option to opt-out or always opt-in based on preference.
Open questions:
Right now I'm leaning towards 3.
IIRC, I think using the Checks API inside of an action will result in an additional check being reported in the checks list. In other words, if you run this action in a workflow called "benchmarks" and the check is called "benchmarks check", then in the list of checks at the bottom of the PR will include a "benchmarks" line item (representing that the workflow successfully completed) and a "benchmarks check" line item for the check the workflow manually created.
That duplicate line item probably doesn't impact lit-html as much since it runs all of it benchmarks in one workflow, but for Preact (which runs each benchmark in its own workflow) it would mean we'd have 8 line items related to benchmarks compared to 4. Hence I'm leaning to 3.
Relevant commit: 96645e6
The text was updated successfully, but these errors were encountered: