Nathan Faubion

I'm not opposed to this, but is there a significant advantage to desugaring to top-level declarations with name munging, vs just supporting them in the type checker under it's normal...

Something else to consider is scoped type variables within kind signatures of local type synonyms. ```purescript foo :: forall k (a :: k). Proxy a -> ... foo _ =...

Yes, it is something I would expect to work with PolyKinds. Kinds and types are in a unified namespace, so scoped type variables apply. This is tricky to do with...

The first. I don't think we should `Type`-default.

It seems like this would require a solution to https://github.com/purescript/purescript/issues/2478 first, at which point it would have a straightforward desugaring.

> As a point of information, this feature was added by Phil as discussed in https://github.com/purescript/purescript/issues/839. (A link in that issue points to [this usage](https://github.com/natefaubion/purescript-psa-utils/blob/5e9c60602ca3065e8e95e79e49c09de3fc4a0cf6/src/Psa/Printer/Default.purs#L150-L159) by @natefaubion of seven years...

> I definitely can do the work, that is "the small effort" in my opinion. And it will also benefit me in learning the codebase better. The work is more...

I wrote up this on discourse a while back, but I don't have the time to see it through. It may be useful as a starting point: https://discourse.purescript.org/t/adding-syntax-for-annotations-on-declarations/988

Regarding this issue specifically, new annotation syntax and semantics is a significant amount of work involving quite a bit of bikeshedding, whereas including more comments in corefn seems fairly uncontroversial....

> If we instead started offering guarantees about which comments in which locations would be associated with which AST nodes. Should it be done either way? As you said, it's...