Gab, any plans for an idle task class in the executor? Something like this would fit in nicely. For example we might want to wait until the user has been idle for a few seconds before running things like this.

Yes, TaskTraits.WithPriority(TaskPriority::BACKGROUND) in base/task_scheduler. These will run on a thread running at background (I/O, CPU, and memory) priority at the OS-level (i.e. will essentially be desched until the kernel decides it has time to run it.

Related:  issue 223430 .

The TaskScheduler will be ready to accept work in Q3 and this could be a good first candidate :-).

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/2def4f5e4c629472298fa679b9803d382960a807

commit 2def4f5e4c629472298fa679b9803d382960a807
Author: vmiura <vmiura@chromium.org>
Date: Mon Jul 25 20:19:41 2016

telemetry: Disable top sites in thread_times benchmarks.

Top sites update tasks add noise to these benchmarks.  Disabling
until tasks are converted to use the TaskScheduler APIs.

BUG=588745
BUG=621128

Review-Url: https://codereview.chromium.org/2168313004
Cr-Commit-Position: refs/heads/master@{#407560}

[modify] https://crrev.com/2def4f5e4c629472298fa679b9803d382960a807/tools/perf/benchmarks/silk_flags.py
[modify] https://crrev.com/2def4f5e4c629472298fa679b9803d382960a807/tools/perf/benchmarks/thread_times.py

This issue has been Available for over a year. If it's no longer important or seems unlikely to be fixed, please consider closing it out. If it is important, please re-triage the issue.

Sorry for the inconvenience if the bug really should have been left as Available. If you change it back, also remove the "Hotlist-Recharge-Cold" label.

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot

I think we can consider this fixed as of https://chromium-review.googlesource.com/c/chromium/src/+/678115. Over to the author to verify and close if appropriate :)

This code is converted to TaskScheduler APIs now. The problem is that one task runner is used for multiple tasks with different priorities. It seems that there is a potential for improvement here with SequenceScheduler.

I did make several improvements on TopSitesBackend::UpdateTopSites. 

One is to remove polling in the topsites update (see CL as treib@ mentioned in comment 7). Thanks to this fix, UpdateTopSites is not "running periodically" anymore which is mentioned in the bug description.

Another improvement is to optimize UpdateTopSites so each run takes less time. see crbug/763103.

The issue that gab@ and altimin@ mentioned is that TopSitesBackend::UpdateTopSites and some other tasks are currently running on db_task_runner_ which has a USER_VISIBLE priority. However, we want UpdateTopSites to run in background. Another related bug is crbug/223430.

 Issue 223430  has been merged into this issue.