[Webkit-unassigned] [Bug 193730] Web Inspector: CPU Usage Timeline

bugzilla-daemon at webkit.org bugzilla-daemon at webkit.org
Wed Jan 23 13:59:07 PST 2019


https://bugs.webkit.org/show_bug.cgi?id=193730

--- Comment #10 from Devin Rousso <drousso at apple.com> ---
Comment on attachment 359929
  --> https://bugs.webkit.org/attachment.cgi?id=359929
[PATCH] Proposed Fix

View in context: https://bugs.webkit.org/attachment.cgi?id=359929&action=review

>> Source/WebInspectorUI/UserInterface/Views/CPUTimelineOverviewGraph.js:119
>> +        }
> 
> Currently this sizes bars at 500ms wide, where the center of the bar is the time of the sample. Devin has pointed out that we may want the end of the bar to be the time of the sample, so that if developers select the range of the bar they will see all of the activity leading up to the sample and not some activity that may have happened after the sample. I think that is valid, but at the low sample rate, I'm not sure it matters as much. I think we will want to end up with finer grained categorization (something at a 1ms interval), and if we get in the detail view then the placement of the samples is not as important.

I would agree that if we have more often sampling, this matters less, but a lot can happen in 500ms, so at larger samples it may make all the difference.  I think the tradeoff point would be somewhere around 16ms (e.g. a single frame) as we can already do some "guess logic" per-frame using the rendering frames view, but anything above that and the record bar "correctness" or "accuracy" matters much more.

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-unassigned/attachments/20190123/613c9792/attachment.html>


More information about the webkit-unassigned mailing list