Restoring the ability to test on local forks

[davidcok]

Being able to test on local forks (on a push event, without a PR having been created) saves CI resources and improves turnaround time on test failures. Note that it is not on by default -- only if the TEST_ON_FORK variable is set in the forked repo.

By submitting this pull request, I confirm that my contribution is made under the terms of the MIT license.

[robin-aws]

First off let me apologize for removing the push trigger without spotting the additional logic designed to restrict that only to forks. I agree we should support running CI on forks easily somehow.

My typical workflow has been to create a dummy draft PR on my fork, based on my fork's copy of master (which I occasionally update when necessary). Once I have the PR cut every subsequent push triggers the CI, and I find it handy to browse the current set of changes in the PR anyway.

Would this workflow be acceptable instead of this solution? Part of the reason I prefer it is that it's much simpler to implement - we don't need all these extra if conditions everywhere.

[davidcok]

So with a draft PR:

• we can run the CI locally?
• and it does not run on master/main-* ?

[davidcok]

Draft PRs do have the disadvantage of spamming other people's email

[robin-aws]

Draft PR on my fork though, so it won't spam anybody but me. :)

[davidcok]

Closing as abandoned