add belt.String

[bobzhang]

see #2759

[cullophid]

Should we not choose lists over arrays when possible? e.g. String.split
switching between list and array seemingly at random (I know its not :)) will be very confusing for newcomers?

[andreypopp]

Could the API use named parameters? I always forget what is what when using values like this:

val replace : t -> t ->  t ->  t

[bsansouci]

Hey could we get this merged? I'd love to have a split function finally lol

[bobzhang]

@andreypopp that makes sense.
What do other people think about this PR, it was halted since I am unclear the story on native side about string processing

[bsansouci]

The native side will need some love. I can start implementing them as soon as you merge but I wouldn't mind some help :)

[phated]

I've been writing many of these string helpers in pure Reason. Is there a desire to implement them as JS externals? If they were implemented in OCaml, wouldn't they be cross-platform?

[bsansouci]

We can implement them as externals when compiling to JS and in Reason when compiling to native, that's probably the best of both worlds :)

[phated]

@bsansouci I haven't reviewed everything but a potential downside would be if the JS APIs aren't the best APIs. For example, I'm writing a lot of my string methods to return Option.

[bsansouci]

null in JS is modeled as option in native, so you're good. To be honest, these are just string manipulation functions, there's only so far you can go with their API. We might have another version of substr for native, just because there's the weirdness with substr vs substring in JS

[phated]

@bsansouci I always felt like a goal of Belt was to remove the Js namespace from code? Doesn't nullable come from Js?

[bsansouci]

Oh well I think we'll be able to / already can compile options to unboxed nulls in JS, so there's no downsides I see to using them in both JS and native. I meant more that on the "backend" side of things, we have the power to do things slightly differently (in this very specific case of optionals), so we don't always need to re-write in Reason.

[MoOx]

I would love to see this released :)

[bobzhang]

This is a good place to use label?

[bloodyowl]

I can PR and target your branch if you want.

I'm planning on also adding Belt.String.length, Belt.String.slice/sliceToEnd, Belt.String.get, Belt.String.indexOf if that's ok with you.

should it also include Belt.String.replaceRegex, Belt.String.matchRegex ou should they just remain in Js.String?

[bobzhang]

what is the type of indexOf, I suggest we avoid char type in the API, otherwise it looks good to me

[bloodyowl]

the type would be t -> t -> int option

[cristianoc]

closing very old PRs