Normative: Add Promise.allSettled() #1583
Conversation
leobalter
added
pending stage 4
ljharb
added
the
proposal
spec.html
Outdated
| 1. Let _values_ be _F_.[[Values]]. | ||
| 1. Let _promiseCapability_ be _F_.[[Capability]]. | ||
| 1. Let _remainingElementsCount_ be _F_.[[RemainingElements]]. | ||
| 1. Let _obj_ be ! ObjectCreate(%ObjectPrototype%). |
There was a problem hiding this comment.
would it be worth (for readability) making an abstract operation here (and below) like CreateSettledResult(true/false, x)?
There was a problem hiding this comment.
That sounds like a potentially beneficial refactoring! Would you like this to happen as part of this PR, or separately as a follow-up?
There was a problem hiding this comment.
I guess it could be either, but it’d be nice to do it as part of this PR :-)
There was a problem hiding this comment.
it seems with that change we could also fold these two algorithms into one
spec.html
Outdated
| 1. Assert: ! IsConstructor(_constructor_) is *true*. | ||
| 1. Assert: _resultCapability_ is a PromiseCapability Record. | ||
| 1. Let _values_ be a new empty List. | ||
| 1. Let _remainingElementsCount_ be a new Record { [[Value]]: 1 }. |
There was a problem hiding this comment.
Perhaps it would be beneficial to add a note explaining why remainingElementsCount is a Record rather than just a Number (to pass it by reference rather than by value)?
There was a problem hiding this comment.
This just matches the precedent set by PerformPromiseAll. I'm not opposed to changing this, but maybe we can do it as a separate PR?
There was a problem hiding this comment.
I find this approach highly distasteful (altho i have no better suggestions atm), and commented such on the proposal repo, but since it does mirror the existing approach i think pursuing changing both in a separate PR is the proper approach (rather than obstructing this proposal).
ljharb
force-pushed
the
allsettled-pr
branch
from
fcc7642 to
8fbd973
Compare
spec.html
Outdated
| 1. Let _values_ be _F_.[[Values]]. | ||
| 1. Let _promiseCapability_ be _F_.[[Capability]]. | ||
| 1. Let _remainingElementsCount_ be _F_.[[RemainingElements]]. | ||
| 1. Let _obj_ be ! ObjectCreate(%ObjectPrototype%). |
There was a problem hiding this comment.
| 1. Let _obj_ be ! ObjectCreate(%ObjectPrototype%). | |
| 1. Let _obj_ be ! ObjectCreate(%Object.prototype%). |
spec.html
Outdated
| 1. Let _values_ be _F_.[[Values]]. | ||
| 1. Let _promiseCapability_ be _F_.[[Capability]]. | ||
| 1. Let _remainingElementsCount_ be _F_.[[RemainingElements]]. | ||
| 1. Let _obj_ be ! ObjectCreate(%ObjectPrototype%). |
There was a problem hiding this comment.
| 1. Let _obj_ be ! ObjectCreate(%ObjectPrototype%). | |
| 1. Let _obj_ be ! ObjectCreate(%Object.prototype%). |
ljharb
added
has stage 4
|
Is anything still blocking this? It seems we can apply @chicoxyzzy's suggestions and (given that both @ljharb and @zenparsing LGTM'd) get this merged? |
|
Note that @jasonwilliams somehow cannot apply the open suggestions:
Perhaps one of the editors can take care of this before merging? |
|
@mathiasbynens that's a github bug, i believe when the suggester and the PR author both have write access - but either way it can be applied manually on the command line. I can certainly incorporate any open suggestions prior to merging as well. |
|
Okay, so this seems good to go then! |
ljharb
force-pushed
the
allsettled-pr
branch
from
8fbd973 to
6744202
Compare
... in recently-merged `allSettled` (tc39#1583)
This is a pull request for the Stage 3 proposal https://github.com/tc39/proposal-promise-allSettled. The proposal is championed by @mathiasbynens.
test262 tests: yes tc39/test262#2131 and tc39/test262#2124
Rationale
A common use case for this combinator is wanting to take an action after multiple requests have completed, regardless of their success or failure. Other Promise combinators can short-circuit, discarding the results of input values that lose the race to reach a certain state. Promise.allSettled is unique in always waiting for all of its input values.