CI: split benchmarks from other checks

[afeld]

That job was the longest-running, so this should parallelize the various code checks with the benchmarks to get done in less wall clock time.

This pull request builds on #39745.

cc @fangchenli

• closes #xxxx
• tests added / passed
• Ensure all linting tests pass, see here for how to run them
• whatsnew entry

[jreback]

not sure this is actually running we had 19 checks before, should show up in the list.

[jreback]

can this be in more of the builds? (you change a few i c); normally do this as a separate PR (e.g. when you refactor want to do one thing), but ok to leave.

[afeld]

can this be in more of the builds?

Maybe! The environment gets set up various ways for various builds. I wasn't sure why, so just changed the ones that looked the same.

normally do this as a separate PR

The refactoring was in service of splitting the two jobs, but happy to take that to a separate pull request if the splitting of the job doesn't work.

[afeld]

not sure this is actually running we had 19 checks before

Ah, whoops, I thought that was a GitHub Actions glitch, but as usual, Problem Exists Between Chair and Keyboard 😛 Hadn't realized that when using an action in the same repository:

• The path needs to start with ./
• The job needs to run actions/checkout before that action can be used (which makes sense, now that I think about it)
• Need to specify the shell for every step (defaults aren't inherited)

20 checks running now, as intended!

[jreback]

@afeld can you merge master

[jreback]

can you merge master

[jreback]

i merged this one: #39745

[afeld]

Hmm, went from the longest run being CI / Checks at 48 minutes to CI / Benchmarks taking 46. Better, but not a lot better 😕

Flaky tests failing, but CI ones are 🟢.

[github-actions]

This pull request is stale because it has been open for thirty days with no activity. Please update or respond to this comment if you're still interested in working on this.

[jbrockmendel]

@afeld is this ready?

[datapythonista]

Not sure if this is worth. We've got an overheat of 15 minutes, and we are saving 5 from 50 for an extra build. No strong opinion, but I think our CI should be faster without this with the amount of builds we usually have running (unless we've got many more than the 20 GH actions jobs, that I don't think we did).

[datapythonista]

Not sure if this is worth. We've got an overheat of 15 minutes, and we are saving 5 from 50 for an extra build. No strong opinion, but I think our CI should be faster without this with the amount of builds we usually have running (unless we've got many more than the 20 GH actions jobs, that I don't think we did).

@afeld thoughts on this?

[simonjayhawkins]

@afeld closing as stale. feel free to reopen if you want to continue with this. (@datapythonista has a question)