Author Robert Reynolds
Recipients Aigars Mahinovs, Jim.Jewett, Robert Reynolds, chortos, dmascialino, eric.araujo, eric.snow, ezio.melotti, jjconti, loewis, ncoghlan, rbcollins, ubershmekel
Date 2020-05-28.13:51:57
SpamBayes Score -1.0
Marked as misclassified Yes
Message-id <>
I second [what was said by Aigars Mahinovs]( about long-running processes giving confusing tracebacks that make debugging very difficult. I have a Natural Language Processing pipeline extracting features from a large corpus of texts, and that process can take days to complete. If the underlying modules have since been edited, then when an Exception occurs – including a KeyboardInterrupt – then the traceback shows the wrong lines. The functions listed at the end of the line are correct, which is the only reason I was able to easily detect the source of my confusion; the line number cited was no longer inside of the listed function!

I propose one more simple thing to track that would be helpful in my situation: how many lines were in the file at call time vs now. It would be (potentially) helpful to have a warning point out that the cited module is now 17 lines longer than it was when it was imported. That way I can make more intelligent guesses about what line was actually the culprit. Obviously there could have been additions and deletions, which muddies the water, but this would at least be a starting point.
Date User Action Args
2020-05-28 13:51:57Robert Reynoldssetrecipients: + Robert Reynolds, loewis, ncoghlan, rbcollins, jjconti, ezio.melotti, eric.araujo, ubershmekel, chortos, dmascialino, eric.snow, Jim.Jewett, Aigars Mahinovs
2020-05-28 13:51:57Robert Reynoldssetmessageid: <>
2020-05-28 13:51:57Robert Reynoldslinkissue8087 messages
2020-05-28 13:51:57Robert Reynoldscreate