implement spread for jsx3

[IwanKaramazow]

#1428

<Foo> ...bar </Foo>;

<Foo> ...((a) => 1) </Foo>;

Is parsed as:

(Foo.createElement ~children:bar ()) [@JSX]);;
(Foo.createElement ~children:(fun a  -> 1) ())[@JSX];;

[chenglou]

Does this work with not just ...foo but also ...<div />? Aka <Foo> ...<div /></Foo> -> Foo.createElement children::(<div />) () [@JSX]

[IwanKaramazow]

Supported:

<Foo> ...bar </Foo>;
<Foo> ...((a) => 1) </Foo>;
<Foo> ...<div /> </Foo>;
<Foo> ...[|a|] </Foo>;
<Foo> ...(1, 2) </Foo>;

Desugars to the following in the old syntax:

Foo.createElement children::bar () [@JSX];
Foo.createElement children::(fun a => 1) () [@JSX];
Foo.createElement children::<div /> () [@JSX];
Foo.createElement children::[|a|] () [@JSX];
Foo.createElement children::(1, 2) () [@JSX];

It **does not ** support:

<Foo> ...[x] </Foo>;

Should I add support for ...[] ?

[chenglou]

Ehhh good one, I think we should? What do you think?

[jaredly]

Yeah it would be weird to be inconsistent

[IwanKaramazow]

Added the <Navbar> ...[myList] </Navbar> variant.
Note that this will reformat to a single child because their equivalent.

<Navbar> ...[myList] </Navbar>;
(Navbar.createElement ~children:[myList] ())[@JSX]
<Navbar> myList </Navbar>;
(Navbar.createElement ~children:[myList] ())[@JSX]

List syntax inside jsx children without spread operator <Navbar> [list1] [list2] </Navbar>, is broken atm & will be addressed in a different PR.

[chenglou]

Hah, that's troublesome now. Because from the react jsx ppx perspective, <Foo> ...[bar] </Foo> is definitely not the same as <Foo> bar </Foo>. Latter's an array...

[chenglou]

Non-idempotent formatting isn't great. If this is the only solution then I think we should just format <Foo> ...[bar] </Foo> to <Foo> bar </Foo>. And then hopefully people don't actually use the former with the reactjs jsx ppx

[jaredly]

Hmmm yeah are you sure <Bar>...[x]</Bar> would refmt to <Bar>x</Bar>?

[jordwalke]

This wouldn't be "non-idempotent" as @IwanKaramazow described. I think it would be exactly what you would expect. <X>...any_expression</X> means "the children should exactly be the expression any_expression. If you happen to supply <x>...[literal, list]</x>, then the ...[ ] was not necessary in the first place and would then be equivalent to <x>literal list</x>.