1322411 - Use requestIdleCallback for getting job data updates

Status: RESOLVED FIXED

(Tree Management :: Treeherder, defect)

Description

[James Graham [:jgraham]]

This is a background task, so we can let the browser schedule it at some convenient time when nothing more important is happening.

Comment 1

[GitHub Autolander Bot]

Attachment: [treeherder] mozilla:requestidle_jobs > mozilla:master

Comment 2

[Treeherder GitHub Bugbot]

Commit pushed to master at https://github.com/mozilla/treeherder

https://github.com/mozilla/treeherder/commit/1a85cf974aa1b53b2d28517ef89146c71e8a446b
Bug 1322411 - Use requestIdleCallback to get jobs data updates. (#2022)

* Bug 1322411 - Refactor some resultset_store.js code to use es6 features

In particualr use some fat arrow functions and reformat a little
code for improved brevity and clarity.

* Bug 1322411 - Use requestIdleCallback for scheduling job data refreshes.

This is a background task, so we should allow the browser to schedule it at some
time it isn't busy doing something more significant.

Comment 3

[Treeherder GitHub Bugbot]

Commit pushed to master at https://github.com/mozilla/treeherder

https://github.com/mozilla/treeherder/commit/60f6c3022ef00852f054d1fde715064b00081170
Revert "Bug 1322411 - Use requestIdleCallback to get jobs data updates. (#2022)"

This reverts commit 1a85cf974aa1b53b2d28517ef89146c71e8a446b.

Comment 4

[William Lachance (:wlach)]

Sorry, but I had to revert this as it was causing the recent jobs lookup to hammer the API. It appears that the behaviour of this function is to *call back in at most X* as opposed to *callback at X or some convenient time*.

:needinfo'ing jgraham to look at this when he gets back. In Treeherder the change is simple but we need to also sort out the issue of bad api / mdn documentation.

Comment 5

[James Graham [:jgraham]]

Ah, it seems that gecko doesn't implement that parameter yet :(

Want to fix it in th to use a timeout + rIC in the interim?

Comment 6

[James Graham [:jgraham]]

Sorry I totally misread the spec here. I didn't notice that there's a race between invoking the callbak normally and invoking it via the timeout.

The timeout parameter is indeed a maximum time (kind of, since the UA may actually take an indefinite amount of extra time, so I don't really know what the use case is). We just need to add a timeout before the requestIdleCallback.

Comment 7

[GitHub Autolander Bot]

Attachment: [treeherder] mozilla:requestidle_jobs > mozilla:master

Comment 8

[James Graham [:jgraham]]

Comment on attachment 8820807 [details] [review]
[treeherder] mozilla:requestidle_jobs > mozilla:master

Only change from the previous patch is putting the call to `requestIdleCallback` in a `$timeout`.

Comment 9

[William Lachance (:wlach)]

Comment on attachment 8820807 [details] [review]
[treeherder] mozilla:requestidle_jobs > mozilla:master

This looks fine, I'm going to assume that you've tested this locally (if not, please do before landing)

Comment 10

[Treeherder GitHub Bugbot]

Commits pushed to master at https://github.com/mozilla/treeherder

https://github.com/mozilla/treeherder/commit/79a0c8b92e4263f286b7905d6e7d58ed3974deb4
Bug 1322411 - Refactor some resultset_store.js code to use es6 features

In particualr use some fat arrow functions and reformat a little
code for improved brevity and clarity.

https://github.com/mozilla/treeherder/commit/c2c36a01a7f2508ba7fbebef59d2f871370fd864
Bug 1322411 - Use requestIdleCallback for scheduling job data refreshes.

This is a background task, so we should allow the browser to schedule it at some
time it isn't busy doing something more significant.