PYTHON-3959 - NULL Initialize PyObjects

[aclark4life]

NULL initialize PyObjects before calling PyErr_*

[evergreen-ci-prod]

There is an existing patch(es) for this commit SHA:

• Evergreen patch with PR

Please note that the status that is posted is not in the context of this PR but rather the (latest) existing patch and that may affect some tests that may depend on the particular PR. If your tests do not rely on any PR-specific values (like base or head branch name) then your tests will report the same status. If you would like a patch to run in the context of this PR and abort the other(s), comment 'evergreen retry'.

[Jibola]

LGTM

[Jibola]

Actually, I see there are some non NULL-initialized PyObjects. I think we should target all the PyObject* instantiations as I'm assuming this has only targeted the ones that leverage PyErr, the original scope of the PR. Looking for suggestions from others.

[ShaneHarvey]

The jira ticket, PYTHON-3959, mentions initializing vars for PyErr_Fetch. Should this PR be updating those?

[aclark4life]

The jira ticket, PYTHON-3959, mentions initializing vars for PyErr_Fetch. Should this PR be updating those?

PYTHON-3959 mentions initializing PyObject variables so that PyErr_* calls are less noisy in Coverity reports. This PR only initializes PyObject variables that precede calls to PyErr_*. As @Jibola mentions above we could also initialize all PyObject variables too for "good measure", but if there are other proposed changes to be made in addition to that, I'm not aware of them yet.

[ShaneHarvey]

So are we going to update the PyErr_Fetch calls?

[aclark4life]

So are we going to update the PyErr_Fetch calls?

Yes! Sorry, In my first pass I misread which PyObject variables were called by which PyErr_ functions. Fixed.

[ShaneHarvey]

I know this is only changing NULLS but just to be safe we should wait until 4.9 is released to merge.

[ShaneHarvey]

I'll defer to Jib for the rest.

[Jibola]

Missed two in _cbsonmodule.c

• 1
• 2

Thanks for the work thus far, though! I imagine navigating this c file is a bit pesky.

[aclark4life]

Thanks for the work thus far, though! I imagine navigating this c file is a bit pesky.

Got'em, thanks! C is a pleasure to work with 👼

[Jibola]

LGTM!!!

[blink1073]

Let's hold off on merging this until we address https://jira.mongodb.org/browse/PYTHON-4636