object destructuring

[zdenko]

@GeoffreyBooth I've made some improvements regarding the object destructuring assignment and rest elements.
Basically, implement proposal for rest properties for ECMAScript in CS: {a, b, c...} = x

Since this proposal isn't yet at stage-4, CS doesn't compile to ES.
In my PR I catch rest element and assign remaining values to it:
{a, b, c...} = x compiles to:
a = x.a, b = x.b, c = extractObjectWithoutKeys(x, ['a', 'b'])

Multiple rest elements are disallowed. ES also requires the rest element to be the last, so compiler currently throws an error.
IMHO, CS should allow rest element anywhere, just like for arrays.
Later, when proposal reaches stage-4 we could implement similar conversion as you did for the function parameters, and ensure rest element is the last in the compiled JS.

[zdenko]

I removed the requirement for the position of the rest element.
CS will always compile it as the last element.

Currently, {a, b, c...} = x and {a, c..., b} = x are compiled to
a = x.a, b = x.b, c = extractObjectWithoutKeys(x, ['a', 'b']).

When ES proposal hits the stage-4 it will compile to {a, b, ...c} = x

[GeoffreyBooth]

I haven’t reviewed the guts of what you’ve done, just some superficial comments on the stuff around it. Bravo though for implementing this feature without needing to output the stage-3 syntax. Perhaps you should add some comments like “When XXX is finalized, remove these lines” or somesuch so that we know what to revise once the syntax gets final approval?

[GeoffreyBooth]

Should this be in here?

[GeoffreyBooth]

CS2 assumes ES2015+. Do we need this function? Is there no ES function like Object.keys that can do some or all of this work for us?

I’m not opposed to the function if we need it; just trying to avoid helpers unless we need them.

[GeoffreyBooth]

Please remove commented-out code.

[GeoffreyBooth]

Ditto. Please also follow the style of the file, two empty lines between tests.

[GeoffreyBooth]

Please follow the style of the file, without empty lines here.

[zdenko]

I've made some progress yesterday and currently, I'm able to compile:

{a, b:z, c} = {a:1, b:2, c:3, d:4, e:5}

obj1 = {f:1, g:2, h:3}
{f, g} = obj1

into

var obj1;

({a, b:z, c} = {a:1, b:2, c:3, d:4, e:5});

obj1 = {f:1, g:2, h:3};
({f, g} = obj1);

and with rest element

{a, b:z, c, r...} = {a:1, b:2, c:3, d:4, e:5}

obj1 = {f:1, g:2, h:3}
{f, q...} = obj1

into

var obj1, q, r, ref;

ref = {a: 1, b: 2, c: 3, d: 4, e: 5}, ({a, b: z, c} = ref), 
r = Object.keys(ref).filter((k) => ['a','b','c'].indexOf(k) == -1).reduce((a, c) => (a[c] = ref[c], a), {});

obj1 = {f: 1, g: 2, h: 3};
({f} = obj1), 
q = Object.keys(obj1).filter((k) => ['f'].indexOf(k) == -1).reduce((a, c) => (a[c] = obj1[c], a), {});

I'm stuck with declarition of the variables from the object at the top of the scope.
Any hints or suggestions how to approach this?

So this:

obj1 = {f:1, g:2, h:3}
{f, g} = obj1

should compile to

var obj1, f, g;

obj1 = {f:1, g:2, h:3};
({f, g} = obj1);

[GeoffreyBooth]

Take a look at jashkenas#4478. I was working last night too and I got that part working. Merge my branch into yours.

The declaration happens automatically when you do new Assign.

[zdenko]

So, with the latest commit, I covered simple and nested destructuring assignment:

obj = {a:1, b:2, c:3, d:4, e:5}
{a, b:z, d, r...} = obj

a1={}; b1={}; c1={}; d1={}
obj = {
    a: a1
    b: {
      'c': {
        d: {
          b1
          e: c1
          f: d1
        }
      }
    }
    b2: {b1, c1}
}
{a:w, 'b':{c:{d:{b1:bb, r1...}}}, r2...} = obj

I added the compileObjectDestruct method to the Assign class, in which I basically traverse thru the object, remove splats and create a list of assignments which are returned back to the compileNode.
For that reason, I also added answers variable in the compileNode where possible splat assignments are stored.
When ES proposal hits stage-4, it should be pretty simple to remove compileObjectDestruct method and answers variable.

I don't check for the position of the splat. Currently, it is removed and the compiled code is correct, and afterward, it can be always compiled as the last element: {a, r..., b} => {a, b, ...r}

I guess the next step could be splat in function parameters, which could compile similar as array splats:

fn = ({a, b, r...}) -> {}

fn = function(arg) {
   var a, b, r;
   ({a, b} = arg), r = Object.keys(arg)......
}

[GeoffreyBooth]

Well . . . what about function parameters, though? As we've been discussing in jashkenas#4478, parameters don't get passed through Assign.

[zdenko]

I've made attempt to enable splats in function parameters destructuring.

fn = ({a, b, r...}) -> {}

fn = function(arg) {
   var arg, a, b, r;
   ({a, b} = arg), r = Object.keys(arg)......
}

When parsing parameters I'm checking for possible splats. If they exists, I pass the object through Assign, push results to the function body and create required variables in function scope.

[zdenko]

I saw you closed this PR. Any feedback on last commits?
Should I continue with the work?
I'm almost done with the spread properties
https://github.com/tc39/proposal-object-rest-spread/blob/master/Spread.md

[GeoffreyBooth]

Sorry I didn’t intentionally close it, it got closed because my destructuring branch got merged (and then deleted). We need to change the target of this PR to the jashkenas:coffeescript 2 branch.

[GeoffreyBooth]

Looks like I can’t modify the merge target. Can you edit the PR to point to the 2 branch?

And yes please continue working, I’m eager to get this merged in! Thanks!

[zdenko]

Did you perhaps check the function parameters destructuring (the last commit)?
Is the approach acceptable?
And sorry, I wasn't paying full attention to the discussion in jashkenas#4478
Bussy days. Plus, I just dived into this without a deeper overview of CS internals.