New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.

Issue 654430 link

Starred by 2 users

Issue metadata

Status: Fixed
Owner:
Last visit > 30 days ago
Closed: Mar 2017
Cc:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 3
Type: Bug



Sign in to add a comment

Identify long term plan for remaining PageLoad.Timing2 histograms

Project Member Reported by bmcquade@chromium.org, Oct 10 2016

Issue description

Most Timing2 histograms have been migrated to an appropriate category such as PaintTiming or DocumentTiming.

However the following Timing2 histograms still remain:

PageLoad.Timing2.NavigationToCommit
PageLoad.Timing2.NavigationToCommit.Background PageLoad.Timing2.NavigationToFailedProvisionalLoad PageLoad.Timing2.NavigationToFirstBackground
PageLoad.Timing2.NavigationToFirstBackground.AfterCommit.BeforePaint PageLoad.Timing2.NavigationToFirstBackground.AfterPaint.Before1sDelayedInteraction
PageLoad.Timing2.NavigationToFirstBackground.AfterPaint.BeforeInteraction PageLoad.Timing2.NavigationToFirstBackground.BeforeCommit
PageLoad.Timing2.NavigationToFirstBackground.DuringParse PageLoad.Timing2.NavigationToFirstForeground

We should figure out the right long term plan for each of these. Some may be deprecated, and others should be moved to existing or new PageLoad histogram categories.
 
NavigationToCommit and NavigationToFailedProvisionalLoad are somewhat internal. We're intending to migrate consumers away from tracking time to commit, which is somewhat less well defined in a cross-browser way, and towards NavigationToParseStart, which is similar to commit time but has a more well defined start time. Thus, I propose that we migrate these to the PageLoad.Internal namespace for the time being.
NavigationToFirstBackground are very closely related to abort metrics. They aren't quite aborts, but I believe we could incorporate the concept of being backgrounded before paint into AbortTiming or perhaps rename AbortTiming to something slighly more general that incorporates both aborts and the user backgrounding the page. For example
PageLoad.AbortTiming.Background.BeforeCommit

Since other aborts are only logged in cases where the abort occurred in the foreground, 'background' aborts will not overlap with any existing abort categories.

Charles, what do you think about treating 'background' as its own abort category? If we don't do this, where should these NavigationToFirstBackground histograms live?
AbortTiming.Background sounds fine to me.
Great, will send out a change shortly, thanks!

This leaves PageLoad.Timing2.NavigationToFirstForeground - not quite sure what to do with it. I may just leave it in Timing2 for now.
Project Member

Comment 5 by bugdroid1@chromium.org, Oct 12 2016

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/3f54b6c611497a48e138d03555be0388c11d60dc

commit 3f54b6c611497a48e138d03555be0388c11d60dc
Author: bmcquade <bmcquade@chromium.org>
Date: Wed Oct 12 01:28:09 2016

Add new background abort type.

These new metrics will replace the existing NavigationToFirstBackground
metrics, which will be removed after it is determined that the new and
old metrics are logging similar values for a few days in canary. Note
that these histograms do log under slightly different conditions, so are
not expected to log identical values.

BUG= 654430 

Review-Url: https://codereview.chromium.org/2403113004
Cr-Commit-Position: refs/heads/master@{#424635}

[modify] https://crrev.com/3f54b6c611497a48e138d03555be0388c11d60dc/chrome/browser/page_load_metrics/metrics_web_contents_observer.cc
[modify] https://crrev.com/3f54b6c611497a48e138d03555be0388c11d60dc/chrome/browser/page_load_metrics/observers/aborts_page_load_metrics_observer.cc
[modify] https://crrev.com/3f54b6c611497a48e138d03555be0388c11d60dc/chrome/browser/page_load_metrics/observers/aborts_page_load_metrics_observer.h
[modify] https://crrev.com/3f54b6c611497a48e138d03555be0388c11d60dc/chrome/browser/page_load_metrics/observers/aborts_page_load_metrics_observer_unittest.cc
[modify] https://crrev.com/3f54b6c611497a48e138d03555be0388c11d60dc/chrome/browser/page_load_metrics/observers/from_gws_page_load_metrics_observer.cc
[modify] https://crrev.com/3f54b6c611497a48e138d03555be0388c11d60dc/chrome/browser/page_load_metrics/page_load_metrics_observer.h
[modify] https://crrev.com/3f54b6c611497a48e138d03555be0388c11d60dc/tools/metrics/histograms/histograms.xml

Project Member

Comment 7 by bugdroid1@chromium.org, Oct 18 2016

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/ab53067e0c3938944cf7aa9b120811e8a031a4f8

commit ab53067e0c3938944cf7aa9b120811e8a031a4f8
Author: bmcquade <bmcquade@chromium.org>
Date: Tue Oct 18 02:43:23 2016

Remove PageLoad.Timing2.NavigationToCommit histograms.

This histogram uses a commit time recorded in the browser process,
which is incorrect. Additionally, we are encouraging consumers to
move to 'parse start' since it's a more well defined timestamp in
the render process that is conceptually similar to commit.

Until we have a customer that needs to track actual commit time,
we are removing it. If we need to support this in the future,
we should instrument the commit time properly in the render process
as part of that work.

BUG= 654430 

Review-Url: https://codereview.chromium.org/2418763005
Cr-Commit-Position: refs/heads/master@{#425868}

[modify] https://crrev.com/ab53067e0c3938944cf7aa9b120811e8a031a4f8/chrome/browser/page_load_metrics/metrics_web_contents_observer.cc
[modify] https://crrev.com/ab53067e0c3938944cf7aa9b120811e8a031a4f8/chrome/browser/page_load_metrics/metrics_web_contents_observer.h
[modify] https://crrev.com/ab53067e0c3938944cf7aa9b120811e8a031a4f8/chrome/browser/page_load_metrics/observers/core_page_load_metrics_observer.cc
[modify] https://crrev.com/ab53067e0c3938944cf7aa9b120811e8a031a4f8/chrome/browser/page_load_metrics/observers/core_page_load_metrics_observer.h
[modify] https://crrev.com/ab53067e0c3938944cf7aa9b120811e8a031a4f8/chrome/browser/page_load_metrics/observers/core_page_load_metrics_observer_unittest.cc
[modify] https://crrev.com/ab53067e0c3938944cf7aa9b120811e8a031a4f8/chrome/browser/page_load_metrics/observers/https_engagement_metrics/https_engagement_page_load_metrics_observer.cc
[modify] https://crrev.com/ab53067e0c3938944cf7aa9b120811e8a031a4f8/chrome/browser/page_load_metrics/page_load_metrics_browsertest.cc
[modify] https://crrev.com/ab53067e0c3938944cf7aa9b120811e8a031a4f8/chrome/browser/page_load_metrics/page_load_metrics_observer.cc
[modify] https://crrev.com/ab53067e0c3938944cf7aa9b120811e8a031a4f8/chrome/browser/page_load_metrics/page_load_metrics_observer.h
[modify] https://crrev.com/ab53067e0c3938944cf7aa9b120811e8a031a4f8/tools/metrics/histograms/histograms.xml

the remaining PageLoad.Timing2 histograms were moved to PageLoad.PageTiming.
Status: Fixed (was: Started)

Sign in to add a comment