Sebastian Thiel

Thanks for that wonderfully detailed and conclusive issue ! The problem described here is well known to me, and may originate in my previous misconception on the reliability of destructors....

Which version of GitPython are you using ? There has been a bug that could cause it to omit certain messages to stderr in the past. Besides, as a general...

Ah, interesting ! I think you should be able to set [custom environment variables](https://github.com/gitpython-developers/GitPython/blob/1537aabfa3bb32199e321766793c87864f36ee9a/doc/source/tutorial.rst#L337) per call or for all calls on each repository's git instance. That way you can provide...

It seems this issue turned from a question and possibly a description of an issue (i.e. missing error description from git.push) into a specific feature request. As to me GitPython...

During my work on [gitoxide](https://github.com/Byron/gitoxide) I learned that the protocol is critically important to know in this issue as well. For instance, only when using an `http` remote the git...

@tardyp thanks a lot for the suggestion and the research! If you could provide the code above in a PR I would happily merge it.

Thanks for sharing all your insights! It's much appreciated and certainly helpful to others.

This was closed via #894 and released with v3.0.0

Reopened as the fix was reverted to deal with the performance regression reported in #906 . We hope to let the fix return in a version that caches the output...

To me it looks like `pytest` isn't recent enough or it lacks a plugin. Maybe @Yobmod has a better idea about this though.