bpo-44461: Fix bug with pdb's handling of import error due to a packa… #26937
Conversation
…ge which does not have a __main__ module
iritkatriel
added
needs backport to 3.9
needs backport to 3.10
jaraco
left a comment
jaraco
left a comment
There was a problem hiding this comment.
Really appreciate the PR here and your analysis in the bug. Nice work.
One thing I notice is that .runmodule and .runscript are symmetric operations, and one of the reasons that .reset() isn't invoked is because in .runmodule, the viability of the target (-m package.mod) isn't checked until in the runmodule step, whereas in the file-based approach, the presence of the file is checked early and SyntaxErrors are trapped separately.
Perhaps runmodule should do something similar and handle non-existence of an executable module (here) by raising a distinct exception to be handled separately in the main loop. Then, the conditions that lead to .botframe not being set would never get triggered (similar to how things are handled when a script is missing or has a SyntaxError).
WDYT?
| if pdb._user_requested_quit: | ||
| break |
There was a problem hiding this comment.
After experimenting, I understand the motivation for this change here, but I also worry that it might affect other workflows we haven't considered. Could there be other use-cases where a different Exception occurred and pdb._user_requested_quit is True, but we would want that to fall through (not break)?
There was a problem hiding this comment.
Good question. self._user_requested_quit will be True only when we get quit()/exit() or EOF. ISTM that it's safe to exit in both cases.
I'm pretty sure I've had to kill pdb processes in the past, when they stopped responding to quit(), so I would guess it's more likely that there are other code paths that would benefit from this change.
Co-authored-by: Jason R. Coombs <[email protected]>
|
Closing as this is no longer under consideration. |
3 participants
…ge which does not have a main module
https://bugs.python.org/issue44461