Ngan Pham
Ngan Pham
I agree as well. 🤔 @frankh do you have specific examples for why it would break existing pipelines if we change this default in a minor release?
👋 Gusto person here! > I don't see why we'd want to tank system test performance to avoid it. Agree. System tests sharing the same connection as Puma is the...
Yup, lots of failures: ``` ActiveRecord::StatementInvalid: -- Â | Mysql2::Error: This connection is in use by: # ``` I just swapped out the lock in abstract adapter though (for a fake/noop...
> Let me try patching the actual lock later. Same problem.
I think our problem was initially brought up here: https://github.com/rails/rails/pull/44219#issuecomment-1125524583 In which you responded, @byroot: https://github.com/rails/rails/pull/44219#issuecomment-1126125529
At this point, I'm wondering if this is a fix in Ruby or a fix in Rails. This basic ruby code: ``` require "monitor" lock = Monitor.new lock.synchronize do Enumerator.new...
> Is that for system test or regular tests? As far as I can tell it’s system tests
> do that synchronization at a higher level. interesting. Are you thinking maybe at the request level? Maybe in a test-only middleware?
@deivid-rodriguez does this make sense? Would be happy to open PR.
Hm, I was thinking that if I don’t specify install, it would try to use the existing installed gems and err if it can’t satisfy all dependency requirements with what’s...