1314959 - Turn on pref for requestIdleCallback

Status: RESOLVED FIXED

(Core :: DOM: Core & HTML, defect)

Comment 1

[Andreas Farre [:farre]]

Attachment: 0001-Bug-1314959-Enable-requestIdleCallback-by-default.-r.patch

Bug 1315260, Bug 1313989, Bug 1313864 have all been fixed, which was what was stopping us from shipping requestIdleCallback.

Bug 1318720 is still remaining, but I'm not sure how important that is to fix that before shipping requestIdleCallback.

Any thoughts Ben?

Comment 2

[Ben Kelly [:bkelly, not reviewing]]

(In reply to Andreas Farre [:farre] from comment #1)
> Bug 1318720 is still remaining, but I'm not sure how important that is to
> fix that before shipping requestIdleCallback.

This is a compat issue, but we could still probably ship as long as we fix it soonish.  IMO anyway.

Comment 3

[Olli Pettay [:smaug][[e-mail]]]

Bug 1318720 feels like reasonable important since it is a spec violation and other browsers don't behave that way (IIRC).

Comment 4

[Olli Pettay [:smaug][[e-mail]]]

Is Bug 1318720 somehow hard to implement? In my mind it feels rather simple. One could for example mark requests with some generation and not ever execute requests which has higher generation than the current one.

Comment 5

[Olli Pettay [:smaug][[e-mail]]]

Comment on attachment 8840463 [details] [diff] [review]
0001-Bug-1314959-Enable-requestIdleCallback-by-default.-r.patch

I think I'd like to see the previous comment answered and Bug 1318720 possibly fixed before reviewing this.

Comment 6

[Andreas Farre [:farre]]

(In reply to Olli Pettay [:smaug] (pto-ish for couple of days) from comment #4)
> Is Bug 1318720 somehow hard to implement? In my mind it feels rather simple.
> One could for example mark requests with some generation and not ever
> execute requests which has higher generation than the current one.

I'm going to set a "chaining limit", based on the id generator in nsGlobalWindow, in the executor and only allow requests with lower id to be executed. And whenever execution/idle dispatch stops due to that I'll set a timeout handler to the deadline of the last idle period to start dispatching again. So no, not that hard.

Comment 7

[Andrew Overholt [:overholt]]

Seems like we'll block on bug 1318720 here so setting dependency.

Comment 8

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/3ec06476d730
Enable requestIdleCallback by default. r=smaug

Comment 9

[Wes Kocher (:KWierso) (Not reading bugmail; email directly if needed)]

https://hg.mozilla.org/mozilla-central/rev/3ec06476d730

Comment 10

[Andrew Overholt [:overholt]]

Are we considering uplifting this to 54?

Comment 11

[Andreas Farre [:farre]]

I would really want to say yes to this, but that means that uplifting bug 1318720 as well.

Hmm, I thought that there was more. Ok, that might actually be doable.

Comment 12

[Andrew Overholt [:overholt]]

I'll leave it up to you but if we're confident enough to turn this on in 54 I'm happy to hear it.

Comment 13

[Andreas Farre [:farre]]

Decided to not uplift, forgot to clear needinfo.

Comment 14

[Eric Shepherd [:sheppy]]

The documentation for this API had already been written but it has been updated some and now reflects that the API was enabled by default in Firefox 55.

Pages updated:

https://developer.mozilla.org/en-US/docs/Web/API/Background_Tasks_API

https://developer.mozilla.org/en-US/docs/Web/API/Window/requestIdleCallback

https://developer.mozilla.org/en-US/docs/Web/API/Window/cancelIdleCallback

https://developer.mozilla.org/en-US/docs/Web/API/IdleDeadline

https://developer.mozilla.org/en-US/docs/Web/API/IdleDeadline/didTimeout

https://developer.mozilla.org/en-US/docs/Web/API/IdleDeadline/timeRemaining

https://developer.mozilla.org/en-US/Firefox/Releases/55