GH-101673: Fix pdb bug that local variable changes lost after longlist

[gaogaotiantian]

The bug is described in #101673.

Using f_code.co_name to check whether it's a frame module avoids reading f_local whose side effect reverts local variable changes in pdb.

• Issue: pdb loses local variable change after command longlist #101673

[bedevere-bot]

Most changes to Python require a NEWS entry.

Please add it using the blurb_it web app or the blurb command-line tool.

[ghost]

All commit authors signed the Contributor License Agreement.

[bedevere-bot]

Most changes to Python require a NEWS entry.

Please add it using the blurb_it web app or the blurb command-line tool.

[artemmukhin]

@iritkatriel I am wondering if you could be an appropriate reviewer for this. Could you please take a look?

[iritkatriel]

Thanks for adding the test. Before merging this I would need to spend some time to understand why this happened and to convince myself that this is the right solution. If someone else reviews it before I get around to it that's obviously fine.

[artemmukhin]

@iritkatriel Thank you for your quick response.

I have investigated the issue further and discovered that the problematic obj.f_globals is obj.f_locals check was introduced in 5ed2b5a to work around the limitation of inspect.getsourcelines, as the latter did not check that the given frame is a module-level frame.

However, inspect.getsourcelines was independently corrected in 91cb298 using the same object.f_code.co_name == "<module>" technique.

So I believe that we should consider replacing the custom pdb.getsourcelines function with direct inspect.getsourcelines invocations. After making this change, the updated test_pdb still passes. @gaogaotiantian WDYT?

[gaogaotiantian]

Yeah I think that’s the way to go. The original customized pdb version was to avoid module level frame issue and it would be nice to not to keep multiple identical code pieces around. The functionality seems to be the same. I’ll make the changes to use inspect version directly and remove the pdb version.

[gaogaotiantian]

As @ortem said, the reason we had a separate getsourcelines in pdb was because the inspect version did not deal with the module level frame. It does now so we can just switch back to inspect version and kept one less duplicated code piece. I did include the regression test for the issue.

[iritkatriel]

Makes sense. Thank you!

[miss-islington]

Thanks @gaogaotiantian for the PR, and @iritkatriel for merging it 🌮🎉.. I'm working now to backport this PR to: 3.10, 3.11.
🐍🍒⛏🤖

[miss-islington]

Sorry, @gaogaotiantian and @iritkatriel, I could not cleanly backport this to 3.10 due to a conflict.
Please backport using cherry_picker on command line.
cherry_picker 5d677c556f03a34d1c2d86e4cc96025870c20c12 3.10

[bedevere-bot]

GH-102632 is a backport of this pull request to the 3.11 branch.

[bedevere-bot]

GH-102633 is a backport of this pull request to the 3.10 branch.