Jean Boussier

Closing in favor of https://github.com/rails/rails/pull/51349

> `"dev:001>" and "prod:001>"`? Is the `001` even worth it? I don't quite see the value of it. My `irbrc` just strips it.

I think the application name would be valuable yes, especially for people who often juggle between multiple applications. My preference goes to `my-app(dev)>` but no strong opinion here.

I don't think we should backport it, because it's fairly big of a change in behavior. Do you have a more targeted fix?

Thank would be great yes.

But also since the fix doesn't target `main` don't sweat it too much if testing is hard, since that code won't see as much changes as if it was on...

CI is failing for legitimate reasons: ``` Error: PrimaryKeysTest#test_update_counters_should_quote_pkey_and_quote_counter_columns: ActiveRecord::NoDatabaseError: Database not found /rails/activerecord/lib/active_record/connection_adapters/sqlite3_adapter.rb:34:in `rescue in new_client' /rails/activerecord/lib/active_record/connection_adapters/sqlite3_adapter.rb:31:in `new_client' /rails/activerecord/lib/active_record/connection_adapters/sqlite3_adapter.rb:732:in `connect' /rails/activerecord/lib/active_record/connection_adapters/sqlite3_adapter.rb:741:in `reconnect' /rails/activerecord/lib/active_record/connection_adapters/abstract_adapter.rb:692:in `block in reconnect!' /rails/activesupport/lib/active_support/concurrency/null_lock.rb:9:in `synchronize' ```

I'll have to dig deeper (cc @casperisfine), but I think this is semi-intended. If you perform a request or run a job as part of the test, they should start...

Sorry but this report is way too obscure to be actionable. Please try to include some code samples or something like that.

Wouldn't it make more sense to disable it on the entire pool? IIRC with SolidCache you are supposed (or heavily recommended?) to give it its own pool, right?