michael-small

> For those following this issue, what would you think of providing this feature by providing errors via the unified control state event observable ? (related to https://github.com/angular/angular/pull/54579 & https://github.com/angular/angular/issues/48947)...

Does `@let` fall into the scope of this change? For example, I tried the following in 18.1 and it did something interesting. Context: fun things are being found on the...

> @michael-small Exactly, this would not be shadowed. Nice. Good to know, thank you.

Do you have an idea for the API surface, of a set for starters? That is what I would think would be the first method to base things off of....

> That seems good to me @michael-small ! I'll take a look into this approach then. > For naming, I like the simplicity of `set` to be honest, but I'm...

Update on this issue: I have a somewhat working version of the group setter going. I think it is rather naive and missing some things, but it is coming along....

> Great catch, I think we should try to keep as close to possible to the Angular behavior. Feel free to update the validator to also accept empty strings. I...

Ah I see, it would make sense to not have tests if it is a proof of concept. ### My point, directly about adding tests > At least some e2e...

> The open question is - should we have `readonlyState: true` by default? Yes. I think @LMFinney's point above encapsulates that well. That said, I think refactors which would be...

> Backward compatibility is not a big problem in my opinion. If we decide to go with readonlyState: true by default, the migration script that adds readonlyState: false to all...