News for hackers. Stuff that matters.
Re:Cancelable Promises: Why was this proposal withdrawn? by BrendanEich on Monday December 19, 2016 @03:20PM (#13214674)
You miss the point that TC39 uses a champions model: 1 or 2 people who do design, with the committee reviewing.

That broke down here, to the extent that the champion, Domenic, withdrew under some kind of internal (to Google if not his own state of mind) duress. TC39 never had a chance to consider the argument or whatever it was that caused this.

More to say after January's meeting, I'm sure. I'm not going to throw stones at Google but it would not surprise me if there were backroom objections. If so, they should be brought to the committee cleanly. There is still time.

Re:Cancelable Promises: Why was this proposal withdrawn? by wodencafe on Monday December 19, 2016 @05:48PM (#13215699)
This is a great summary of the situation Brendan, thanks for the clarification about TC-39 committee not actually getting to review the proposal.

The main issue regarding the proposal and its sudden withdrawal seems to be an all-around lack of transparency.

Re:Cancelable Promises: Why was this proposal withdrawn? by mcguire on Monday December 19, 2016 @06:29PM (#13215956)
Do the champions provide an implementation of the proposal for the committee to review?
Re:Cancelable Promises: Why was this proposal withdrawn? by BrendanEich on Tuesday December 20, 2016 @01:45PM (#13222654)
See https://tc39.github.io/process-document/. To get real engine implementations may take cooperation among champion and one or two browser vendors, as the "real engines" (meaning ones that can be tested against the Web at scale) are all owned by browser vendors.

Examples:

* SIMD started from Dart's proposal, which John McCutchan then championed at first for JS, and later handed off to others. Intel, Mozilla, and Google all cooperated.

* ES6 Proxies, where Mark Miller and Tom Van Cutsem were the champions and did a terrific job covering the design space and finding the sweet spots. Andreas Gal implemented in SpiderMonkey and then the spec changed, based on the implementation (but in a mostly co-expressive way, for wins beyond expressiveness; the new proxy implementation layered on the internals of the old). Proxies were quite challenging as they exposed observable internal order of operation details and required reformulating spec-mandated invariants that matter for not only interoperation but also security.

Re:Cancelable Promises: Why was this proposal withdrawn? by bzbarsky on Monday December 19, 2016 @07:18PM (#13216249)
They could, but I'd expect that they generally do not.