It also got me thinking about the orphan instances for Array again - we're pretty inconsistent about what's provided in Prelude. We have instances for Eq, Ord, Show, yet not for any of the Control stuff. Should we fix that?

Looks good.

I think the do notation is quite useful actually. I find the laws for Monad much easier to read that way, and I don't think it's hard to make the connection to valid do notation. I would like to support braces and semicolons anyway.

For Array, I think I'd rather do things in the other direction, long term, even moving Array out of Prim and into purescript-arrays.

Yeah, I did wonder about that. I guess that's a “Meltdown” change, as I assume it ties in with the removal or change in pattern matching and [] syntax that was/is being discussed.

It would be nice to have some way of dealing with kind of flexible syntax for that kind of thing actually - it would also help with the int and char literal problem. OverloadedStrings in GHC isn't wonderful, but it's damn useful.

some way of dealing with kind of flexible syntax

Templates?

myCharLiteral = [char|x]
myIntLiteral = [int|100]

Verbose, but very flexible.

Yeah, I meant perhaps something less flexible so it would also be less verbose than just calling a char or int function. ;) Like the IsString stuff.

Well one advantage of templates is that they can verify validity at compile time.

[char|actually a string] would give a compile time error.

Ah, that's a good point.

Tests failing as I misplaced (%) by accident it would seem. Should we remove it anyway, and promote purescript-integers for that usage?