All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=747805

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=2b74bda0e2fe1d0e36e9bc9c01b1d8ed2c357712324801722dc0c3fddeef8392


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

chromium-rel-mac-retina
chromium-rel-mac11
chromium-rel-mac11-air
chromium-rel-mac11-pro
chromium-rel-mac12
chromium-rel-win7-dual
chromium-rel-win7-gpu-ati
chromium-rel-win7-gpu-intel
chromium-rel-win7-gpu-nvidia
chromium-rel-win8-dual
linux-release
win-high-dpi

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8973206293196625824

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8973205866620402528

=== Auto-CCing suspected CL author vogelheim@chromium.org ===

Hi vogelheim@chromium.org, the bisect results pointed to your CL, please take a look at the
results.


=== BISECT JOB RESULTS ===
Perf regression found with culprit

Suspected Commit
  Author : Daniel Vogelheim
  Commit : 5578c07d877cb8bf3ee41439ea1c8ff9c0dd7394
  Date   : Fri Jul 21 14:03:19 2017
  Subject: ScriptRunner can initiate script streaming.

Bisect Details
  Configuration: win_perf_bisect
  Benchmark    : v8.runtimestats.browsing_desktop
  Metric       : Parse-Background:duration_avg/browse_social/browse_social_twitter
  Change       : 59.63% | 30.1941666667 -> 48.1993333333

Revision             Result                  N
chromium@488620      30.1942 +- 5.42345      6      good
chromium@488639      29.8812 +- 4.91845      6      good
chromium@488644      29.9335 +- 7.03059      6      good
chromium@488647      31.226 +- 7.50026       6      good
chromium@488648      50.6378 +- 22.7619      6      bad       <--
chromium@488649      48.1087 +- 9.79475      6      bad
chromium@488658      48.1993 +- 5.51587      6      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=browse.social.twitter v8.runtimestats.browsing_desktop

More information on addressing performance regressions:
  http://g.co/ChromePerformanceRegressions

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8973206293196625824


For feedback, file a bug with component Speed>Bisection

Observations:

1, The CL identified here is part of my last V8 action item, finishing the script streaming. This would have an effect on parse time, so the alert matches.

2, My feature is behind a feature flag, and so should not have a substantial effect anything. This doesn't match.

3, The alert identifies an increase in *background* parsing, so this matches, no. 2 notwithstanding.

4, More background parsing (unless accompanied with an increase in overall load time time) is generally desirable.

5, Streaming has a race condition, in that the streamer can only parse one stream at a time, and hence the first stream to request the streamer will get it. (Improving this is the action item in question.) So it's conceivable that a subtle change in the logic - despite the 'real' feature being disabled - would change which stream gets picked, which might affect performance.

6, The performance change seems to only affect one site, twitter.com, while all other sites appear unaffected.

--------------

- I'm mostly concerned #2. I'll double check that.
- #4 suggests that this is largely harmless, since paging loading itself doesn't increase. so I'll probably close this anyhow.
- #5 + #6 suggests that the root cause is probably that the streamer, in a few cases, ends up picking a different script. That should be fine.

 Issue 747861  has been merged into this issue.

=== BISECT JOB RESULTS ===
NO Perf regression found

Bisect Details
  Configuration: mac_10_12_mini_8gb_perf_bisect
  Benchmark    : v8.runtimestats.browsing_desktop
  Metric       : Parse-Background:duration_avg/browse_social/browse_social_twitter

Revision             Result                  N
chromium@488577      62.1465 +- 28.5613      21      good
chromium@488631      59.4993 +- 16.0758      21      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=browse.social.twitter v8.runtimestats.browsing_desktop

More information on addressing performance regressions:
  http://g.co/ChromePerformanceRegressions

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8973205866620402528


For feedback, file a bug with component Speed>Bisection

 Issue 747936  has been merged into this issue.

The regression reported in #6 /  Issue 747861  seems far more concerning, but at least on first try I can't reproduce it locally.

That is, my change doesn't seem to dramatically impact time-to-first-meaningful-paint for that example. I just did one run, tough; also on x64.release; but on Linux. (Report is on Win7. My change has no platform specific parts.)

 Issue 749354  has been merged into this issue.