Jean Boussier

Right, and that something we should fix it some point. But regardless, the behavior you are adding could be made a pool config. My worry with this API is that...

I don't have full context and wonder why it was done that way, but your point make sense to me. I'd say rebase and ping me when it's done. Rafael...

To expand on Rafael's answer. Unfortunately setting the `inverse_of` can lead to compatibility issue with some patterns of code. However it does bring performance and other benefits so I think...

> We could consider gating this whole change behind a configuration flag as to offer a deprecation period for older code. Thinking more about this, I think it's indeed necessary...

@rosa this change will be behind a configuration flag for Rails 7.2: https://github.com/rails/rails/pull/50883 It was known from the start that while the change makes thing better, existing code may be...

Alright, I guess I don't quite understand the issue. If you think there's a bug, it would be best to open an issue with a repro script.

> this change seems related to `has_many ..., through:` Yeah, the title is a bit misleading, it doesn't apply exclusively to those. > it is not clear to me what...

I'm in vacation this week. I'll try to have a look this week end.

I considered this but wasn't expecting anyone to use aliased attributes in this case. Is this something you are doing?

> But if they are explicitly not supported, I think it's fine. Yeah, I don't really see a use case for it to be honest. That said I initially took...