@michaelficarra are you seriously opening issues during PureScript Conf?

@Fresheyeball Maybe. It's inspiring!

👍

Is requiring forall a feature? I thought we only did that becasue we hadn't got around to allowing Haskell-style implicit typevars.

I find class forall a. Sized a where [...] a bit strange - first, you wouldn't be allowed to use a real type in that context (that is, class Sized Foo where [...] isn't allowed), and also, I think of class Sized a where as introducing a type variable a into the inner scope already, so there shouldn't be a need for an extra type-variable-introducing construct (that is, forall).

👍 for the instance part, though, I do find the current behaviour a bit jarring.

I agree with @hdgarrood in theory, but in practice, this change is massive for relatively little gain IMO.

Also agreed with @hdgarrood. I still think this is an inconsistency worth fixing eventually. It's unfortunate that this got passed over for 0.7.0.

One could at least allow explicit quantification syntax for a few versions.

Ah, good idea. If both are allowed, the current syntax can be deprecated using our new warning system.

Backlog?

Yeah, I think so.

I personally would still really like to see this. I find reading implicitly quantified instances difficult, and have even made silly typo mistakes due to this.

I'd like to work on this feature, but am new to purescript. @natefaubion does #3824 affect this feature or the other way around?

@jrairigh It's the other way around. Implementing this feature would likely affect the syntax of #3824.

I'm guessing the changes would go somewhere in the CST parsing code. Is this correct? If not, could someone please point me in the right direction?

This would indeed go somewhere in the CST parsing code; probably by adding new cases to instHead in lib/purescript-ast/src/Language/PureScript/CST/Parser.y. I think we might also want to start emitting warnings for implicitly quantified variables in instances, and then turn that into an error later.

I've added a new deprecated warning constructor to ParserWarningType called WarnDeprecatedInstHeadForallSyntax. What should the deprecation warning text to be?

e.g. Implicit 'forall' in instance head is deprecated and will be removed in a future release. Add 'forall' keyword to instance head.

Also, am I correct in assuming I should create a branch off of master and create the PR off of branch? Are there any other steps I've missed?