Reason: Support for private extensible variants?

[rickyvetter]

Right now it doesn't look like private constructors are respected in reason.

Original OCaml:

type foo += private C of int

Actual:

type foo +=
  | C int;

Expected:

type foo +=
  | private C int; /* Error: Unbound value pri */

[chenglou]

Hot damn what are you trying to do

[rickyvetter]

@chenglou don't worry about it 🕵🏻

[jordwalke]

Great first task and high pri because it's regarding completeness!

[IwanKaramazow]

Should be familiar terrain for me, will look into it

[SanderSpies]

@IwanKaramazow 💪

[IwanKaramazow]

Ok, this was already implemented.

type water = ..;
type molecules = string;
type water += pri Ocean molecules;

Reason uses pri instead of private, this probably originates from @SanderSpies's new object syntax. Do we just document this? Or should I make it private in this case?

[rickyvetter]

Ahh interesting - when I refmt that code I get:

type water = ..
type molecules = string
type water +=
  | Ocean of molecules

So maybe it's just an issue with refmt not handling it? It doesn't work in any direction ml <=> re or self formatting.

[IwanKaramazow]

So it parses, but prints incorrectly. I'll fix it.

[SanderSpies]

In general it would be great if we could try to convert more OCaml libraries to Reason and see if it works.

[jordwalke]

Might want to do this on top of Fred's diff.

[IwanKaramazow]

Opened a PR that should be compatible with Fred's diff, cherry-picking the commit should suffice?