Increase new primer timeout to 60m

[jacobtylerwalls]

See last run on main. The 3.7 job sometimes takes just barely over 45 minutes. The old primer timeout was set to 60m.

Clearly we need to add some concurrency. Rather than parallelizing the primer message comparison infrastructure, my hope is that we can just (later this year or next) solve the underlying issues with --jobs (see label topic: multiprocessing) and use that.

[Pierre-Sassoulas]

We can't count on the github runner machine to have more than one core consistently, so we probably need both parallelization

[jacobtylerwalls]

Oooh, that's a much larger ask then. I looked at that briefly a month ago and then realized there's no good solution for streaming JSON results to a file.

[coveralls]

Pull Request Test Coverage Report for Build 2645542754

• 0 of 0 changed or added relevant lines in 0 files are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage remained the same at 95.385%

---

Totals
Change from base Build 2644912543:	0.0%
Covered Lines:	16700
Relevant Lines:	17508

---

💛 - Coveralls

[DanielNoord]

We can't count on the github runner machine to have more than one core consistently, so we probably need both parallelization

I do think the "partial" cores the runner machines offer could be used if we created a more barebones way of parallelisation. Instead of creating a completely new PyLinter just pass a FileItem and some basic options and run check_file.

I really don't have time to look into this in the coming weeks due to deadlines and holidays, but I would certainly be interested in a (Discord/Zoom) call in August or September to discuss some possibilities. It seems this requires such large refactoring efforts that it would be good to both discuss the approach and perhaps even divide some tasks.

[jacobtylerwalls]

My time is also limited these next few months but in general 👍🏻

[DanielNoord]

@jacobtylerwalls https://github.com/PyCQA/pylint/runs/7271165374?check_suite_focus=true

Am I overseeing something or is pip install -e . in the Run pylint primer step overwriting the install of astroid on the latest commit on main? 😅

Edit: Ah, this is funny. Because we haven't upped the constraint in pylint yet the commit on main doesn't satisfy the version requirement so we're downloading the latest version available on PyPI. This is also why all primer comments are messed up.

Anybody got a good solution?

[Pierre-Sassoulas]

Should we relax the requirement even more ? I don't think a lot could go wrong (as long as we set it correctly before release). We can use tbump for that.