Percent less

[let-def]

Allow extension sugar without having to put %:

let async f = ...
/* as well as */
switch async / try async / for async / if async / ...

The sugar applies for any lowercase identifier that is not a keyword.

The let f() = ... / let f() { ... } syntax had to be removed (functions are now always bound like any other expressions, let f = () => ...).

[hcarty]

OCaml supports ;%extend - is there a way to support a Reason equivalent with this syntax? It would be useful for operations used purely for side effects like printing/logging.

[jordwalke]

It's a very nice surprise that Reason's syntax enabled us to omit the % without any ambiguities. This looks really approachable.

A couple of thoughts: We would probably want to start with make omitting the % opt-in, just in case we for some reason need to use a different solution in the next release. In that way people could try it out that it's subject to change.
We also might want to only add a few whitelisted ppx names to be printed with the % omitted because it might look confusing for other ppx names.

For some reason it's failing on Mac OS only with " Error: 8 states have shift/reduce conflicts."

[let-def]

The conflicts are due to the patch forcing parentheses around switch/try scrutinee being reverted.

[ELLIOTTCABLE]

Wow. This would really be the killer feature of the Reason syntax, IMO.

Well, assuming the BuckleScript team relaxes its position on AST modification. 🤣

[jaredly]

ummm I have strong reservations about elevating ppx extension points to this level of "prominence" in the syntax. I believe it should be fairly "loud" when a syntax transform is happening.
@chenglou and I have talked about an alternate proposal that would allow monads to be handled gracefully without relying on ppx's (basically through first-class CPS support).
it would involve introducing a let!foo x = y; ...etc form that would be translated to foo(y, x => ...etc)
This allows us to handle async, optionals, results, etc, while also allowing you to cmd-click into foo to see what's happening. It feels much less magic than ppxs, which I like a lot.

[hcarty]

@jaredly There are a few relevant active PRs upstream which may be of interest and worth chiming in on to make sure the Reason community's needs are taken into account.

Monadic let operators - ocaml/ocaml#1947

match operators - ocaml/ocaml#1955

[jaredly]

@hcarty wow, that's a very cool proposal!
it would be cool to have something like that for reason, especially if we found a way of having readable text (let!foo) as opposed to relying on symbols exclusively.