gh-129640: Add tests for reference cycle in gzip.GzipFile

[Mr-Sunglasses]

Issue: #129640

Related to: #129726

• Issue: gzip.GzipFile creates reference cycle that requires a deep garbage collection cycle to cleanup. #129640

[ZeroIntensity]

Some triage nits.

[Mr-Sunglasses]

Thanks @ZeroIntensity for the suggestions, I've made the change 😄

[hauntsaninja]

Why is the weakref not sufficient?

[bedevere-app]

A Python core developer has requested some changes be made to your pull request before we can consider merging it. If you could please address their requests along with any other requests in other reviews from core developers that would be appreciated.

Once you have made the requested changes, please leave a comment on this pull request containing the phrase I have made the requested changes; please review again. I will then notify any core developers who have left a review that you're ready for them to take another look at this pull request.

[cmaloney]

I think just timing / this issue was added before, PR added after, didn't see this issue when I was working on the other one.

Could you check if the issue is resolved with current main and/or 3.12 branch? I think #130055 should have fixed it as well as what it was directly trying to address (bad destruction order) by adding a weakref.

I do like the testing for number of GCd object test / think that would be good to add regardless

[Mr-Sunglasses]

Why is the weakref not sufficient?

I think just timing / this issue was added before, PR added after, didn't see this issue when I was working on the other one.

Could you check if the issue is resolved with current main and/or 3.12 branch? I think #130055 should have fixed it as well as what it was directly trying to address (bad destruction order) by adding a weakref.

I do like the testing for number of GCd object test / think that would be good to add regardless

Thanks @cmaloney the issue is fixed in the main branch and 3.12 branch and sorry while looking in this issue, I was testing in my Python 3.13 and there it is reproducible, that is why i opened this PR and I did not see #130055 ( my bad 😅 ), So I think we need to backport #130055 it to 3.13.2 ? Also I'll remove the change, but I think as @cmaloney suggested we can add the tests.

cc. @hauntsaninja

TiA.

[cmaloney]

#130055 is backported to 3.12 and 3.13 (#130670 and #130669), so should be in the next minor release of both. According to the 3.13 Release Schedule (https://peps.python.org/pep-0719/), 3.13.3 bugfix release which should contain it is expected 2025-04-08.

[Mr-Sunglasses]

@cmaloney @hauntsaninja as of our discussions, I changed this PR for the tests for reference cycle in gzip.GzipFile.

• Also I'm not sure if it needs any NEWS entry, please let me know it need one.

[Mr-Sunglasses]

Requesting @hauntsaninja for a review.

[hauntsaninja]

This test passes on the commit prior to #130055, so I'm not sure it's testing what we want it to test. len(gc.get_objects()) is quite coarse, where does the magic 10 number come from? Can we instead get a weakref to the _WriteBufferStream and confirm it gets collected?