Arnaud Spiwack

I'll reject for the MVP. We can refine later.

Hi @erikd thanks for the feedback. Let me, first, clarify that the error messages haven't been designed: they're just the bare minimum that I needed to work on the meat...

I will try to finish a robust monomorphic implementation before I make a polymorphic one. When the multiplicity-polymorphic implementation is out, you can define your own multiplicity-parametric `Arrow` type class...

I'm not sure what your question about promotion is?

I see. Well, I don't see a reason why this could be done automatically as GHC does not provide an implicit user-defined coercion mechanism. I believe that you'll end up...

It may be possible to use polymorphism if set right. But it's not entirely obvious: there is no subtyping between linear arrows and variable-multiplicity arrows. I think it is possible...

> When you say “arrows”, are you really talking about arrows (values of instances of the Arrow class), or are you talking about functions? I meant function types, sorry. >...

Just to check consistency, are there other extensions which enable some unicode syntax? If so what error message do they give when trying to use the unicode syntax without `-XUnicodeSyntax`?

Note: it's better to use tweag/ghc to report such issues. There is already a conversation on error messages, by the way: tweag/ghc#5 . The short story is: error messages are...

Almost certainly not, unfortunately. There will be a need of eta-expanding the function here and there. Of course we can do ```haskell (++') :: [a] -o [a] -o [a] (++)...