🤖 New build scheduled with the buildbot fleet by @iritkatriel for commit e1cc312 🤖

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

Shouldn’t pdb.pm() be changed to use sys.last_exc?

Yes, idlelib too. But I'm not sure whether we do that now or in 3.14 when it won't interfere with backports.

Shouldn’t pdb.pm() be changed to use sys.last_exc?

Yes, idlelib too. But I'm not sure whether we do that now or in 3.14 when it won't interfere with backports.

I would do it now, but look both (in case user code sets one or the other): if sys.last_exc exists, use that, else if sys.last_(type,value,traceback) exist, use those, else fail. Then in the future we can delete the code checking for the latter.

🤖 New build scheduled with the buildbot fleet by @iritkatriel for commit 2ce2c26 🤖

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

Please revert the changes to idlelib. (Why the rush before I could respond.) I dislike the awkward workaround changes, which IDLE does not need, and they cannot be backported. I would much prefer to make more extensive changes that get rid of nearly all references to the deprecated attributes and which can be backported.

For instance, in test_stackviewer setup, replace svs.last_type ... = svs.exec_info() with svs.last_exc = svs.exec_info()[1]. In the teardown, replace del ... with del svs.last_exc, leaving no usage of the deprecated names. In other modules, add sys.last_exc = sys.last_value and eliminate all other uses of the old names. In June 2024, eliminate the use of last_value.

Let me request again (from the previous PR that was closed) that idlelib changes be put in a separate PR that I can review and backport.

Sorry, will revert and fix.

In other modules, add sys.last_exc = sys.last_value and eliminate all other uses of the old names.

If a user tries to access sys.last_type or sys.last_traceback then it won't work. Is this not an issue?

Yes, sys.last_type, sys.last_value, sys.last_traceback = excinfo should remain as long as the REPL does the same. I should add a comment in the file. I believe other settings are only for internal use for stackviewer and its test, but will check.