Stephan Jaekel

If Celery is down, the same happens as desribed earlier (stale cache kept and extended until a new cache version was generated).

I'm not sure if passing serializing the whole context can be done easily or fast enough.

Sounds complex to me. Feel free to build an proof of concept but I'm really not sure if this should land in cacheback. Maybe an extra library built up on...

Interesting issue. Besides not knowing if this would work if the partial issue is fixed, the partial issue by itself is worth being fixed. I'm more than happy to review...

As the QuerySetJob should be used for caching purposes, I think its ok if get_or_create is not supported. Can you provide a use case which required get_or_create on QuerySetJob?

How yould you handle the "created" flag in this case? To be consistent, it should only be true for the first "get" call. Later cached "get" calls would need to...

> Yeah the Python 3.8 tests fail for some unknown reason related to the version pinning within Black itself (the tests pass when run independently). And since Black is unpinned,...

Thank you for your report. As you stated correctly, output capturing is complicated and sadly, isort uses print to stderr directly with no option to configure it. As we call...

After investigating further, the actual issue is, that the module tries to delete the "speed" field from a interface that doesn't have that option (in this case, a xe interface)....