gh-101266: Fix __sizeof__ for subclasses of int

[mdickinson]

This PR fixes the behaviour of the __sizeof__ method (and hence the results returned by sys.getsizeof) for subclasses of int.

• Issue: int and set subclass __sizeof__ under-reports the instance dictionary pointer #101266

[mdickinson]

@ionite34 This should fix the int portion of the bug you reported.

@corona10 Do you have enough spare cycles to review?

[mdickinson]

Darnit; this fails for bool; looks like we'll have to special case in boolobject.c. Working on it.

[mdickinson]

this fails for bool

That's because we had somewhat bogus numbers for tp_basicsize and tp_itemsize for the bool class. Now fixed.

I'm going to remove the "backport to 3.10" label: the bug does technically still exist in 3.10, but it's messier to fix - it looks like we would need to add a dedicated __sizeof__ implementation for bool in that case, and it doesn't seem worth the effort.

[corona10]

Do you have enough spare cycles to review?

Thanks I will take a look

[corona10]

@mdickinson We have to resolve the conflict before reviewing this PR.
Mark Shannon refactored the longobject ;)
c1b1f51

[bedevere-bot]

🤖 New build scheduled with the buildbot fleet by @corona10 for commit 9cd5d01 🤖

If you want to schedule another build, you need to add the :hammer: test-with-refleak-buildbots label again.

[mdickinson]

@corona10

@mdickinson We have to resolve the conflict before reviewing this PR.

Thanks! Updated.

[bedevere-bot]

🤖 New build scheduled with the buildbot fleet by @corona10 for commit ae6229b 🤖

If you want to schedule another build, you need to add the :hammer: test-with-refleak-buildbots label again.

[corona10]

lgtm,

Just one more thing,
Would you like to change the PyBool_Type initialization with C99 style through this chance?

PyTypeObject PyBool_Type = {
    PyVarObject_HEAD_INIT(&PyType_Type, 0)
    "bool",
     .tp_basicsize = offsetof(struct _longobject, long_value.ob_digit),
     .tp_itemsize = sizeof(digit),
     .tp_dealloc = bool_dealloc,
      ...
     .tp_vectorcall = bool_vectorcall

[mdickinson]

Would you like to change the PyBool_Type initialization with C99 style through this chance?

Good idea! It may make sense to do that in a separate PR, though: I was hoping to backport this bugfix PR to 3.11, and the C99 style change sounds like a cosmetic update that shouldn't be backported.

[corona10]

It may make sense to do that in a separate PR,

Great! Feel free to merge anytime you want :)

[miss-islington]

Thanks @mdickinson for the PR 🌮🎉.. I'm working now to backport this PR to: 3.11.
🐍🍒⛏🤖

[miss-islington]

Sorry, @mdickinson, I could not cleanly backport this to 3.11 due to a conflict.
Please backport using cherry_picker on command line.
cherry_picker 39017e04b55d4c110787551dc9a0cb753f27d700 3.11

[mdickinson]

Hmm. Looks like I'll need to do a manual cherry-pick backport to 3.11, excluding the last few commits. I'll do that later today.

[bedevere-bot]

GH-101579 is a backport of this pull request to the 3.11 branch.

[mdickinson]

See #101579 for a manual backport to 3.11 (cherry-picking the commits up to but excluding the main merge).

[sobolevn]

This PR was the first one to fail refleak bot (test_importlib for some reason): https://buildbot.python.org/all/#/builders/802/builds/582

See #101766

[mdickinson]

@sobolevn

Hmm. We could try reverting it in a PR, and then running the refleak buildbots on that PR to see what happens. It seems rather implausible that it was this particular PR that introduced the refleak, though.

[sobolevn]

Yes, seems suspicious. I will investigate!