Author vstinner
Recipients Timothy Lusk, Timothy Lusk2, benjamin.peterson, tim.peters, vstinner
Date 2020-03-26.01:18:45
SpamBayes Score -1.0
Marked as misclassified Yes
Message-id <1585185525.28.0.497112578827.issue37434@roundup.psfhosted.org>
In-reply-to
Content
I'm unable to reproduce the crash on Fedora 31.

A crash during a garbage collection usually means that there is a corrupted object somewhere, and suddenly, we discover the inconsistent. The crash can be far from where the inconsistency was created.

You may try python3.7 -X dev which might provide more information. The best would be to test a Python 3.9 compiled in debug mode.

The setup.py file of Python binding of protobuf has an optional --cpp_implementation option to build C++ extension modules. It seems like they are not built by default. I'm not sure if this bug requires these C++ extension modules, or if they must miss.

It seems like "pip install protobuf" installs protobuf without these C++ extension modules.

See also https://pythondev.readthedocs.io/debug_tools.html to debug a crash.

It's likely a bug in protobuf, rather than a bug in Python itself. I close the issue.
History
Date User Action Args
2020-03-26 01:18:45vstinnersetrecipients: + vstinner, tim.peters, benjamin.peterson, Timothy Lusk, Timothy Lusk2
2020-03-26 01:18:45vstinnersetmessageid: <1585185525.28.0.497112578827.issue37434@roundup.psfhosted.org>
2020-03-26 01:18:45vstinnerlinkissue37434 messages
2020-03-26 01:18:45vstinnercreate