Sergey Tikhonov

Failing test here https://github.com/Suor/django-cacheops/pull/408 I see multiple ways: 1. Catch `pre_delete` and traverse all m2m relations with calling "invalidate_dict" as in example workaround 2. Add m2m-related queries to "profile" conj...

It's a ten years old ticket without any movement on last 5 years. https://code.djangoproject.com/ticket/17688 So I think there will be no fix at Django side

Cache key is computed as md5 checksum of sql query with some other parameters, so if your sql vary - you will get different cache keys. `cached_as` keys and queryset...

Lots of questions. Shortly, in docs both eviction strategies are considered as experimental, so they are just not suitable in your case. Worst case for inconsistent cache state is that...

What does this line? ``` python loop.run_until_complete(asyncio_redis.Connection.create('localhost', 6379)) ``` First, it starts asyncio event loop and waits for result of `Connection.create`. Then asyncio loop is destroyed. What does `Connection.create`? 1)...

I've not seen redis client library that checks own connection, but lot's of situations when pubsub connection is closed by network hardware (and nobody know about it). connection_lost() doesn't resubscribe...

Socket opened in create_connection is local variable, so you can't without hacking find out what file descriptor to close. Other possible solutions are: 1) add timeout to asyncio in python...

> Why would we ever have to hard-code a timeout for the connection creation? > When does it happen that it would otherwise block for ever? > For example, closed...

> Maybe someone should make a fork of this project and develop it? There is a fork already https://github.com/RubinCarter/gitflow4idea-plus

I have similar problem: ``` $ kubectl -n yc1 get pod podname -o json | jq '.metadata.finalizers' [ "batch.kubernetes.io/job-tracking", "kopf.zalando.org/KopfFinalizerMarker" ] ``` Code from kopf ~ 1.36.2: ``` def allow_deletion(...