Operators

[icecream17]

• Previously blocked by Add more operators tree-sitter/tree-sitter-javascript#152 (and other updating prs)
• Previously blocked by ⬆️ update tree-sitter packages to 0.19.0 #692
  • Blocked until Uncaught RangeError after updating tree-sitter atom#22129 is fixed
  • Well, it works in legacy tree-mate, but not in tree-sitter.
  • Help is very appreciated

Description of the Change

Edit: update

Adds optional chaining a?.b a?.[5] a?.()
Adds **=
Adds logical assignment &&= ||= ??=
Add shift class to >>= >>>= <<=
Add **

Ugh unfinished edit this later

Alternate Designs

I don't see how to make the ?. and . distinct without making everything that uses it seperate:

It's much bigger than this but the idea is the same:

Original:
"var.prop" -> "variable" "meta.delimiter.period" "property"

Two way I thought of to support ?.
"var(?. or .)prop" -> "var" "meta.delimiter" "property"

or

"var.prop" -> "var" "meta.delimiter.period" "property"
"var?.prop" -> "var" "meta.delimiter.optional" "property"

I still added ?. as "meta.delimiter.optional" in the tree-sitter, but in the non-tree-sitter, they're the same.
Edit: They're separate selectors, there's a later update below.

I've added tests for the ?. operator, and included ?? in the logic operator test. (Works!!!!!)

Benefits

Fixes #680
Fixes #640
Fixes #715 (duplicate)

Possible Drawbacks

A lot of comments
There might be a way to make them separate without adding a bunch of redundant code, but I haven't figured it out yet. I actually did figure it out, again there's an update below

Applicable Issues

#640 #680

[icecream17]

Only seems to fix ?? right now, so the current code doesn't do what I thought (of course...)

It thinks the ? is ternary, and the period to be a.. regular period.

[icecream17]

Oh the latest issue made me realize I didn't even put in
hmm?.[2]
and
hmm.jump?.()

I think I should reset the files to atom:master to retry fixing this

[utkarshgupta137]

Shouldn't we also include "??=" in operators?

[icecream17]

I added the missing operators but I broke something and the tests are failing

[darangi]

Hi @icecream17 could you take a look into the failing tests?

[icecream17]

Current status
exampleFunc?.() not supported
exampleArr[7]?.() not supported
a?.b should be support if it works
a?.b() should also be supported if it works

Shouldn't we also include "??=" in operators?

Oh yeah. That's added
As for the errors, I try to figure them out tomorrow - I don't know what's wrong, except that regular unrelated code isnt parsing so there's some catastrophic error somewhere

For later testing:
#666 (comment)

[icecream17]

359 fails, now down to 66, though 6f32f5a has only 13 errors, so maybe I should just use that instead....

maybe I should provide less updates?
edit: Now at 18 wrong
edit 2: It is sooo close to working

[icecream17]

There's some precedence thing with the operator parsing. I'm not sure how it works.

I found the actual parser: https://github.com/tree-sitter/tree-sitter-javascript
And the grammar file: https://github.com/tree-sitter/tree-sitter-javascript/blob/master/src/grammar.json

---

^{Note to self: https://github.com/atom/language-javascript/blob/master/grammars/tree-sitter-javascript.cson just maps parsed nodes to names.}

[icecream17]

The tree-sitter parser atom uses can't actually parse some operators yet. Will update when it can.

Edit: The tests at the repo above show that it does support optional chaining.... but it doesn't parse correctly still
Theoretically it would work if the parser worked too.

Edit 3?: ...see tree-sitter/tree-sitter-javascript#152

Edit 6: See #692

[icecream17]

Update:
See atom/atom#22129
Because of that error it's blocked until atom is updated: atom/atom#22130

[icecream17]

Ok, even though it still doesn't work in tree-sitter (see blocking), it could still be merged as long as there's no bad changes.

---

Incompatible changes

Currently I've "added scopes" in 743fcd5 but https://github.com/atom/language-javascript/pull/690/files takes care of that.

However the scopes for rest vs spread on lines 233/234 are still needed