[3.7] gh-80254: Disallow recursive usage of cursors in sqlite3 converters

[erlend-aasland]

(cherry picked from commit c908dc5)

Co-authored-by: Sergey Fedoseev fedoseev.sergey@gmail.com
Co-authored-by: Jelle Zijlstra jelle.zijlstra@gmail.com

[bedevere-bot]

🤖 New build scheduled with the buildbot fleet by @erlend-aasland for commit d52fd31 🤖

If you want to schedule another build, you need to add the ":hammer: test-with-buildbots" label again.

[erlend-aasland]

CI cannot find merge base, and decides that this is not worthy running the tests. Buildbots seem to ignore this fact though, so I guess that it's fine if they are green.

UPDATE: a little bit too fast there; AMD64 FreeBSD Non-Debug is green, but it did not run. It failed before even compiling.

[erlend-aasland]

This buildbot run does not look good. Do you experience this often on the 3.7 branch, @ambv?

[JelleZijlstra]

We're not allowed to merge until regular CI passes (the check is required). Code looks good to me but we'll have to figure out what's up with CI.

[erlend-aasland]

I think it is related to 4ac923f

[erlend-aasland]

See also GH-21806

[erlend-aasland]

I think backporting both GH-24004 and GH-21806 will fix it. Maybe not both are required, but I think both should be backported.

[JelleZijlstra]

I think backporting both GH-24004 and GH-21806 will fix it. Maybe not both are required, but I think both should be backported.

Seems reasonable. I triggered #92341 and #92342, but I am not authorized to merge to the 3.7 branch.

[erlend-aasland]

Great! I'll pull in 3.7 when the backports land.

[vstinner]

The 3.7 branch no longer accept bugfixes, only security fixes: https://devguide.python.org/#status-of-python-branches

Do you consider that this change fix a security issue?

[ned-deily]

Do you consider that this change fix a security issue?

Yes, see discussion here.