[es6] Make sure we do not invoke setter when calling Promise.all

[es6] Make sure we do not invoke setter when calling Promise.all

We were calling the setter for Object.prototype.promise if it existed
when calling PromiseAll.

BUG=v9:4232
LOG=N
R=rossberg@chromium.org, adamk@chromium.org

Committed: https://crrev.com/473badf5fabf0856f89aa6be3d25b639fa217a7e
Cr-Commit-Position: refs/heads/master@{#29357}

[arv (Not doing code reviews), 2015-06-29 15:34:52]

PTAL

[rossberg, 2015-06-29 16:05:53]

https://codereview.chromium.org/1219623005/diff/20001/src/promise.js
File src/promise.js (right):

https://codereview.chromium.org/1219623005/diff/20001/src/promise.js#newcode192
src/promise.js:192: var result = {__proto__: null};
This is visible in the public API (even if .defer is not official). Can you do

var result = {promise: UNDEFINED, resolve: UNDEFINED, reject: UNDEFINED};

instead?

https://codereview.chromium.org/1219623005/diff/20001/test/mjsunit/es6/promis...
File test/mjsunit/es6/promise-internal-setter.js (right):

https://codereview.chromium.org/1219623005/diff/20001/test/mjsunit/es6/promis...
test/mjsunit/es6/promise-internal-setter.js:8: set(x) {
Nit: I supposed this can be abbreviated to {set: assertUnreachable}

https://codereview.chromium.org/1219623005/diff/20001/test/mjsunit/es6/promis...
test/mjsunit/es6/promise-internal-setter.js:14: P.all([Promise.resolve('ok')]);
Can you add a case for .race and .defer as well?
Is the subclassing actually relevant?

[arv (Not doing code reviews), 2015-06-29 16:44:16]

Use O.p as prototype to proprietary API object

[arv (Not doing code reviews), 2015-06-29 16:45:00]

PTAL

https://codereview.chromium.org/1219623005/diff/20001/src/promise.js
File src/promise.js (right):

https://codereview.chromium.org/1219623005/diff/20001/src/promise.js#newcode192
src/promise.js:192: var result = {__proto__: null};
On 2015/06/29 16:05:53, rossberg wrote:
> This is visible in the public API (even if .defer is not official). Can you do
> 
> var result = {promise: UNDEFINED, resolve: UNDEFINED, reject: UNDEFINED};
> 
> instead?

I'm fine doing that.

Where is the spec for this public API =P

https://codereview.chromium.org/1219623005/diff/20001/test/mjsunit/es6/promis...
File test/mjsunit/es6/promise-internal-setter.js (right):

https://codereview.chromium.org/1219623005/diff/20001/test/mjsunit/es6/promis...
test/mjsunit/es6/promise-internal-setter.js:8: set(x) {
On 2015/06/29 16:05:53, rossberg wrote:
> Nit: I supposed this can be abbreviated to {set: assertUnreachable}

Done.

https://codereview.chromium.org/1219623005/diff/20001/test/mjsunit/es6/promis...
test/mjsunit/es6/promise-internal-setter.js:14: P.all([Promise.resolve('ok')]);
On 2015/06/29 16:05:53, rossberg wrote:
> Can you add a case for .race and .defer as well?
> Is the subclassing actually relevant?

It needs to be a function that has a static resolve function

[rossberg, 2015-06-29 16:53:25]

lgtm

[arv (Not doing code reviews), 2015-06-29 17:36:40]

The CQ bit was checked by arv@chromium.org

[commit-bot: I haz the power, 2015-06-29 17:36:52]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1219623005/40001

[commit-bot: I haz the power, 2015-06-29 18:14:35]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2015-06-29 18:14:49]

Patchset 3 (id:??) landed as
https://crrev.com/473badf5fabf0856f89aa6be3d25b639fa217a7e
Cr-Commit-Position: refs/heads/master@{#29357}