bpo-41611: IDLE: fix freezing on completion on macOS

[taleinat]

https://bugs.python.org/issue41611

[taleinat]

CC @E-Paine, @csabella, if one of you can confirm this on macOS that would be helpful.

[E-Paine]

Fix works on my system.

[taleinat]

Fix works on my system.

Thanks @E-Paine!

[E-Paine]

Can we add a test for this, to ensure it doesn't occur in the future? Edit: is that possible?

[taleinat]

Can we add a test for this, to ensure it doesn't occur in the future? Edit: is that possible?

Possible but hard, and we don't have the infrastructure for it yet.

We'd need a way to run IDLE in a sub-process, send events, inspect what happens, and close it cleanly.

We could add a specific h-test for this, but IMO that wouldn't be worth it.

[ned-deily]

LGTM, too. It may still be possible to get this into the delayed 3.10.0b2 if it is merged to 3.10 quickly.

[taleinat]

It may still be possible to get this into the delayed 3.10.0b2 if it is merged to 3.10 quickly.

Ooh yes, let's! Merging now.

[miss-islington]

Thanks @taleinat for the PR 🌮🎉.. I'm working now to backport this PR to: 3.10, 3.9.
🐍🍒⛏🤖

[bedevere-bot]

GH-26401 is a backport of this pull request to the 3.10 branch.

[bedevere-bot]

GH-26402 is a backport of this pull request to the 3.9 branch.

[terryjreedy]

I verified the fix on installed .0b1 on Mac by #ing the line. Both Tab and .[popup] work great. Thank you Tal and others for fixing this and getting it, I hope, into .0b2!!!

The last line of autocomplete_w.py: # TODO: autocomplete/w htest here. I added this, and now put it on my classified list of issues to work on. It is definitely worthwhile because part of the test, besides working, is 'Does this widget look OK?'. This must be about the last visible widget that is not h-tested.

@E-Paine If you want to try this, use the calltip h-test as a model. (and that one should be extended if possible to test force-open).