All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=827145

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=ea5529c316a8ff647f5f7356faddce8a4e1a32325c9618b88b996c8171d615c8


Bot(s) for this bug's original alert(s):

chromium-rel-win7-dual

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/11bd877b440000

📍 Couldn't reproduce a difference.
https://pinpoint-dot-chromeperf.appspot.com/job/11bd877b440000

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/1295b8d8c40000

Expanding the range since the value bumps up over multiple data points.

📍 Found significant differences after each of 2 commits.
https://pinpoint-dot-chromeperf.appspot.com/job/1295b8d8c40000

Add components_perftests to chromium.perf.json by csharrison@chromium.org
https://chromium.googlesource.com/chromium/src/+/a33ba78b7502b9bfad38cfd48fa5f8ff698bac79

Fix startup tracing in renderers by chiniforooshan@chromium.org
https://chromium.googlesource.com/chromium/src/+/f192e8a25e1937336d9cc16f9273b1d13b99530a

Understanding performance regressions:
  http://g.co/ChromePerformanceRegressions

I believe this is not a regression. It is a logging issue that my CL fixed.

- My CL fixed startup tracing, which means now we see more early trace events.

- The way timeToFirstPaint is calculated is that for every firstPaint event of the main frame we find the corresponding navigationStart event before it and compute the duration from navigationStart to firstPaint.

- Before my CL, the first navigationStart event was not logged and so one large timeToFirstPaint was not reported. After my CL, we see all navigationStart events and so report more timeToFirstPaint values.

+dproy@ for verification.

 Issue 827146  has been merged into this issue.

 Issue 827149  has been merged into this issue.

 Issue 827150  has been merged into this issue.

Verified c#7. 

Although it looks like we now have two firstPaint samples, and the the URL for the new first paint is "https://myaccount.google.com/?pli=1" (so this is the login page), while the URL for the other first paint is "mail.google.com/mail". Someone who has more context into this benchmark should verify this was the original intention. 

+Ned and Kouhei about the multiple urls in #11.

 Issue 826827  has been merged into this issue.