Results don't publish if you re-run job

[flcdrg]

I've noticed that while the action doesn't fail (and indeed in the output I see a 201 Created status), the test report never appears when you try and manually re-run the jobs (using the re-run jobs button when viewing a workflow results.

I wonder if there's some extra tracking info that needs to be supplied to differentiate between different job runs?

[flcdrg]

I dunno... Maybe it's GitHub Actions being weird. Sometimes it shows up the checks and sometimes it doesn't. Really odd.. Still trying to find a predictable pattern as to when it doesn't do the right thing.

[ebekker]

One thing to keep in mind, I'm using the Run attachments in a way that was not really intended or designed, as evidenced by the 65K limit errors or the fact that if you have multiple simultaneous Runs executing tests and generating reports, all the reports from all these Runs get attached to a single Run.

So if you're seeing sporadic, incorrect behavior it could be related to this impedance of how Runs and their attachments are supposed to be used and how I use it. In any event, I do recommend switching to the Gist-based report publication approach as it's more consistent and predictable and has several other benefits.

[flcdrg]

There's definitely some strange things that seem to happen sometimes. I've also seen the run check ending up being attached to a different workflow in the UI. I'm going to see if I can take this up with the GitHub folks - it should be more reliable.

It's interesting to compare to how the Azure DevOps API works with pull requests (where that API allows you to attach to a specific 'iteration'), but I guess check runs are slightly different in that they are just based on a commit (not a PR)