StrMap.fromRecord

[matthewleon]

create a StrMap from a homogeneous record

[hdgarrood]

My only concern is that the runtime representation of these two types might not necessarily coincide in other backends.

[matthewleon]

@hdgarrood StrMap is JS FFI only, no?

[matthewleon]

Perhaps the solution is to add a level of indirection before the unsafeCoerce?

[hdgarrood]

How would a level of indirection help?

Currently I think alternate backend maintainers have been forking the core libraries and translating the FFI code; just because there's only JS FFI stuff here, I don't think that means that we can ignore the issue of alternate backends. I think I'd prefer to address #118 before merging this. Other than that, though, this seems great. See also #125

[matthewleon]

@hdgarrood I've made a change, above, that I think should address the concern by moving the coercion to FFI code. Happy to wait for resolution to #118 .

[hdgarrood]

Ohh, I see what you mean now. Yeah, that does seem like it should address that concern. Do you think you can work out how to make the linter happy?

[matthewleon]

The linter loves me now.

[hdgarrood]

Great! Before merging I'd like to get approval from someone else, as I'm not sure I'm fully aware of the implications of adding this function. I'd also like to wait until 0.12 to merge this, because changing dependencies of core libraries is generally something that we need to take a bit of care with and has potential to break things, and so it's probably best to wait until we're breaking everything anyway.

[natefaubion]

I don't think it's a good idea for foreign functions to take constraints. It's better to declare an "unsafe" impl function for internal use, and add a constraint to the PS wrapper. Honestly, I wish PS altogether disallowed constraint syntax in foreign imports.

[matthewleon]

Thanks. Will fix.