The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/faa55192bc07e10058fef50dfcc3531afeef6a13

commit faa55192bc07e10058fef50dfcc3531afeef6a13
Author: bmcquade <bmcquade@chromium.org>
Date: Fri Jul 29 17:23:33 2016

Add equality and empty tests for first_contentful_paint

ksakamoto discovered that our PageLoadTiming code is missing
equality/empty tests for the first_contentful_paint field.

This means that when FCP is different from FP, and no subsequent
metrics are logged for a given page, FCP may erroneously not get
logged for that page.

BUG= 632765 

Review-Url: https://codereview.chromium.org/2195743002
Cr-Commit-Position: refs/heads/master@{#408674}

[modify] https://crrev.com/faa55192bc07e10058fef50dfcc3531afeef6a13/chrome/common/page_load_metrics/page_load_timing.cc

Your change meets the bar and is auto-approved for M53 (branch: 2785)

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/05dac4c5dbc4e0e2e5610edba038f7dd5ebdaa7e

commit 05dac4c5dbc4e0e2e5610edba038f7dd5ebdaa7e
Author: bmcquade <bmcquade@chromium.org>
Date: Sun Jul 31 14:22:26 2016

Add equality and empty tests for first_contentful_paint

ksakamoto discovered that our PageLoadTiming code is missing
equality/empty tests for the first_contentful_paint field.

This means that when FCP is different from FP, and no subsequent
metrics are logged for a given page, FCP may erroneously not get
logged for that page.

BUG= 632765 
TBR=csharrison
NOTRY=true
NOPRESUBMIT=true

Review-Url: https://codereview.chromium.org/2195743002
Cr-Commit-Position: refs/heads/master@{#408674}
(cherry picked from commit faa55192bc07e10058fef50dfcc3531afeef6a13)

Review-Url: https://codereview.chromium.org/2192383002
Cr-Commit-Position: refs/branch-heads/2785@{#421}
Cr-Branched-From: 68623971be0cfc492a2cb0427d7f478e7b214c24-refs/heads/master@{#403382}

[modify] https://crrev.com/05dac4c5dbc4e0e2e5610edba038f7dd5ebdaa7e/components/page_load_metrics/common/page_load_timing.cc

Thank you for fixing this, I should have created a separate patch just for this fix.

Fortunately, I think this bug affects only very limited cases,

- If first contentful paint is text or image, first_{text,image}_paint must be changed at the same time, so equality / empty tests doesn't (wrongly) pass.

- If first contentful paint is SVG or canvas, FCP may not be logged immediately, but once text or image paint happens first_{text,image}_paint is updated and it makes first_contentful_paint also logged.

Thanks! Yes I agree that it seems like it should be rare. I merged the fix into M53 so we'll start getting data from beta soon which will allow us to confirm that it's rare.