Jean Boussier

> one option could be to add a silencer like the following: Bundler's path is already silenced in the parent class.

> it seems like a better configuration point would be to allow customizing `Rails.backtrace_cleaner` i Yes, I didn't realize `Rails.backtrace_cleaner =` isn't defined. It definitely should. But yes, I don't...

> Is this analysis correct Yes.

You don't need to ask for permission.

Hum, not sure if you are aware, but I did implement `.available_processors_count` specifically to use inside Rails' `puma.rb` config. But turns out a bunch of providers don't expose the cgroups...

> Hopefully this PR will be reviewed soon PRs aren't generally monitored by devs, hence why this has been sitting here for so long, it wasn't reported where devs look...

I'm at a conference all weeks so probably won't get back to this until Monday. But @Edouard-chin mentioned this to me today, and quickly: - IMO this is a bug...

> tricky thing for us (it affects our in-house queue adapter in a similar way) is that deserializing job arguments can be expensive Could you expand a bit of that....

Reverted in 53928ebf2f7528ff0f6685bf26ea69bce1728689. I'm however not closing this issue, because in my opinion, not being able to evolve the serialization format is a bug. We should be able to evolve...

To add one more datapoint, Solid Queue seem to solve this problem almost like I think it should: https://github.com/rails/solid_queue/blob/6dc5c6681fecb0b3c712bf24d89ddbbc43a4b14b/lib/active_job/concurrency_controls.rb#L29 It extends Active Job to allow defining an explicit `concurrency_key`. I...