Fix cyclic-import edge exclusion with --jobs

[jacobtylerwalls]

Type of Changes

	Type
✓	🐛 Bug fix

Description

Follow-up to bf8051b. I need to also combine the data collected during the multiprocessing about excluded edges. I wondered about this during development but unfortunately didn't fully test it.

(Sorry for all the PR volume! It won't be like this forever. No rush on the reviews.)

[codecov]

Codecov Report

Merging #8820 (7910e2f) into main (b00d3c9) will increase coverage by 0.00%.
The diff coverage is 83.33%.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #8820   +/-   ##
=======================================
  Coverage   95.89%   95.89%           
=======================================
  Files         173      173           
  Lines       18508    18511    +3     
=======================================
+ Hits        17749    17752    +3     
  Misses        759      759           

Impacted Files	Coverage Δ
pylint/checkers/imports.py	94.21% <83.33%> (+0.03%)	⬆️

... and 1 file with indirect coverage changes

[github-actions]

🤖 According to the primer, this change has no effect on the checked open source code. 🤖🎉

This comment was generated for commit 7910e2f

[Pierre-Sassoulas]

LGTM !

Sorry for all the PR volume! It won't be like this forever. No rush on the reviews.

Ho no, I was hoping you'd fix all the major problems in pylint in 2 weeks, considering the frequency and importance of problems resolved, you were on track to do it 😄

[cdce8p]

I'm currently catching up on the latest changes. This one seems to break my custom CI runs for Home Assistant. Every time I include it, shortly before I expect pylint to finish, I get a Error: The operation was canceled. message.

Reverting just f815e9f does fix it though.

Example CI runs:
Good (768c026): https://github.com/cdce8p/ha-core/actions/runs/5496467930/jobs/10016508120#step:17:32
Bad (f815e9f): https://github.com/cdce8p/ha-core/actions/runs/5496471817/jobs/10016514653#step:17:33

[jacobtylerwalls]

I'll take a look. Does disabling cyclic-import also resolve the issue?

[jacobtylerwalls]

Every time I include it, shortly before I expect pylint to finish, I get a Error: The operation was canceled. message.

What I'm worried about is that the memory cost of the cyclic import checker algorithm is the problem, rather than this PR, and that it's been swept under the rug this whole time because the whole check never even ran with --jobs before. If that's the case, how do you think we should proceed?

[cdce8p]

Does disabling cyclic-import also resolve the issue?

It was already disabled, so unfortunately no.

Every time I include it, shortly before I expect pylint to finish, I get a Error: The operation was canceled. message.

What I'm worried about is that the memory cost of the cyclic import checker algorithm is the problem, rather than this PR

I've never before gotten the message The operation was canceled without any indication to what actually caused it, so it could be a memory issue. However, I also just ran a test with jobs=1 which worked fine.

That leads me to believe it's actually the step that combines the results which might be the problem.

[jacobtylerwalls]

It was already disabled, so unfortunately no.

That sounds like an easy problem to address. Opened #8838 for it.