Matthias Bach

I also ran into this when working on the SUSE package. Weirdly, while the `setup.py` would generate an atlas, that was actually broken. However, this only showed when starting a...

For systems that run Systemd 235 or higher this could also be solved by adding `RuntimeDirectory=keybase/kbfs` to the Service section of `kbfs.service`. This obviously would not fix things for custom...

I am seeing the same issue on openSUSE.

This is an interesting suggestion. I have actually had this use-case myself before. But it vanished because we now `pip-compile` all requirements used in our tests. The tricky part about...

Thank you for your feedback! In fact, devpi-builder could use that information. Internally it uses the original devpi client for communication with devpi and thus if we would not explicitly...

Curiously, the build failures triggered by this PR are exactly the ones that #407 fixes. Not sure how best to proceed here. Ideally we can agree to merge #407 and...

Yes, this still works. I rebased onto latest master make this explicit. Regarding your question as to why one would repackage this within a distribution. You are right. In a...

Yes, I still need this patch. It also seems I now triggered this on CI in #408 so it's no longer exclusive to openSUSE Tumblweed.

So in case this was the reason I was mentioned here: I just added the required dev packages to my openSUSE box and use that as a dev environment. I...

In Postgres 9 the empty string is a valid password. The error in this case stems from the client library. It seems Psycopg2 does not even attempt password authentication in...