Peter Baumgartner

This is a pretty common use case for Django (and I assume Python in general). Use a standard Node.js pipeline for your static frontend assets and then build your Python...

Here's an example we see in a pretty vanilla Django application. Based on the context, I suspect it is a `boto3` call to S3 doing a `PutObject`. 

> If you suspect which call this could be, what are the reasons to not instrument it manually? It defeats the utility of Sentry. When I follow [the directions here](https://docs.sentry.io/platforms/python/guides/django/#configure),...

What I've seen from another APM is tracing with a very small sample set of requests getting profiled. That lets me identify a problem and then deep-dive into the profile...

This library is now part of the mainstream Redis library as of 4.2.0-rc1 https://github.com/redis/redis-py/pull/1899

I dug into this a bit, could you explain what the reasoning is for this? https://github.com/django/channels_redis/blob/f5eef165017a702a1e53fe19a673036c0e337c95/channels_redis/core.py#L73-L79 It's creating a single connection pool then treating and referring to it as a...

Yeah, it's not available here https://github.com/redis/redis-py/blob/c5d19b8571d2b15a29637f56a51b0da560072945/redis/asyncio/connection.py#L1379-L1538 I'm guessing we can iterate over the connections ourselves and verify they have all been closed.

This is the only place we call `asyncio.get_event_loop()` in the new code: https://github.com/django/channels_redis/blob/52e240e7d59b12df498dcdb0b1fc165919c0f3d0/channels_redis/core.py#L315 Does it make sense to replace it with the technique used in https://github.com/django/channels_redis/commit/f5e4799e11f472cc267598e9f78099a160f81550? From a cursory glance,...

The tests require a number of different Redis server instances running. The project uses this script to install and test as part of their CI https://github.com/redis/redis-py/blob/e5e265de87dfe7ef8cc5cad8c247d067c74fa44d/.github/workflows/install_and_test.sh

@dfreilich Yes, testing then publishing would be a better workflow, but that would prevent me from using `--cache-image` @ekcasey Yes, allowing `--cache-image` to work without `--publish` would solve my issue,...