Pun record declaration fields when possible.

[IwanKaramazow]

type a = string
type b = int

type r = {a, b}

Requested in #47

[cristianoc]

@IwanKaramazow was there something to be careful about for unary records?

[IwanKaramazow]

That was in record expressions. Not yet sure how we could tackle that one

[chenglou]

Let's hold off on this one. Not sure about printing record punning yet

[cknitt]

@chenglou What exactly are you unsure about / what are your concerns?

[chenglou]

cknitt it's possible that we reconsider punning in favor of some other syntax possibilities

[bobzhang]

If we allow punning in writing, we should do punning in printing,
otherwise we should not allow punning in writing either.

This would create confusion for example, people write code:

type t = { hi, lo}

and formatted as

type t = { hi : hi, lo : lo} 

People will read it that you are against such style, if you are against such style, why not prohibit in the beginning (in the parsing)?.

I am not against/for punning, but we should be consistent here

[cknitt]

@chenglou I agree with @bobzhang that we should be consistent here.

Also, I think it would be a mistake to remove punning support because:

1. It would make code more verbose/less readable.
2. It was supported in the old Reason parser.
3. Why support it for named function arguments, but not for record fields?
4. I thought the idea for Reason syntax was to be familiar to JS developers, and JS supports punning for objects fields, too.

In short, not supporting punning would feel like a regression to me, both compared to the old Reason parser and to JS.

So I am quite curious about what the "other syntax possibilities" are that outweigh these concerns.

[bobzhang]

@cknitt I digged into a little bit. The proposal is to remove punning in type declarations not in expressions where punning is mostly used. Does it sound good to you?

[cknitt]

@bobzhang Thanks for digging! That's good news. I had understood Cheng's comments to mean that he would like to remove punning altogether.

@cknitt I digged into a little bit. The proposal is to remove punning in type declarations not in expressions where punning is mostly used. Does it sound good to you?

[pckilgore]

I don't mind losing punning on types fields (data fields a different story) but agree with Bob that I would dislike the inconsistency of considering it valid syntax but rewriting it on fmt.

[fhammerschmidt]

One year later I still would love to have this one back. Code just gets more verbose than in Reason.

[ryyppy]

@IwanKaramazow I guess the conclusion of this discussion is that we want to keep the record field declarations always explicit. For punning record values, we already merged #480.

So I'd propose to close this for now.