Issue metadata
Sign in to add a comment
|
26.2% regression in blink_perf.dom at 380376:380387 |
||||||||||||||||||||
Issue descriptionSee the link to graphs below.
,
Mar 17 2016
Started additional bisect here: https://chromeperf.appspot.com/buildbucket_job_status/9017915268927787888
,
Mar 27 2016
Timed out waiting for a build in the v8 roll. Launching again.
,
Apr 4 2016
Bisect re-kicked with wider revision range. Details on the below link: ------------------------- https://chromeperf.appspot.com/group_report?bug_id=594053 Thank you!
,
Apr 4 2016
=== Auto-CCing suspected CL author tkent@chromium.org === Hi tkent@chromium.org, the bisect results pointed to your CL below as possibly causing a regression. Please have a look at this info and see whether your CL be related. ===== BISECT JOB RESULTS ===== Status: completed ===== SUSPECTED CL(s) ===== Subject : TEXTAREA: Cutting last line without EOL should not remove the remaining EOL in the previous line. Author : tkent Commit description: Our TEXTAREA element assumes there is an extra trailing <BR> or \n if the last character of the value is \n. So we remove the last \n when we generate value from innerEditor content. However, the assumption was broken if we cut the last line content without EOL. Editing code doesn't add an extra <BR> in such case. We removed the last \n in the value mistakenly. Solution: We should make sure there no <BR> elements in innerEditor except the extra trailing one so that we can ignore it safely on generating TEXTAREA value. Also, we should make sure editing code doesn't add \n for placeholder newline. - HTMLTextAreaElement simply ignore the last <BR> on generating value. - HTMLTextAreaElement makes sure that innerEditor has the last <BR> after every editing operations. - InsertLineBreakCommand always inserts \n for plain text editing field, however it produces <BR> as an extra line break for TEXTAREA. - ReplaceSelectionCommand appends a <BR> for an InterchangeNewline at the end. - InsertTextCommand should not produce TabSpans for plain text editing. BUG= 522144 , 593184 Review URL: https://codereview.chromium.org/1785603002 Cr-Commit-Position: refs/heads/master@{#380385} Commit : 37f57d892222dc0ed816ae004074d7c80600ad46 Date : Thu Mar 10 10:54:04 2016 ===== TESTED REVISIONS ===== Revision Mean Value Std. Dev. Num Values Good? chromium@380320 0.701506 0.021221 5 good chromium@380375 0.691197 0.02133 5 good chromium@380382 0.684803 0.024374 5 good chromium@380384 0.654152 0.018258 5 good chromium@380385 0.537369 0.019831 5 bad <- chromium@380386 0.54572 0.01324 5 bad chromium@380389 0.553621 0.014898 5 bad chromium@380403 0.560446 0.013574 5 bad chromium@380430 0.544261 0.015154 5 bad Bisect job ran on: android_nexus9_perf_bisect Bug ID: 594053 Test Command: src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --also-run-disabled-tests blink_perf.dom Test Metric: textarea-edit/textarea-edit Relative Change: 22.42% Score: 99.9 Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/android_nexus9_perf_bisect/builds/1704 Job details: https://chromeperf.appspot.com/buildbucket_job_status/9016332853490204080 Not what you expected? We'll investigate and get back to you! https://chromeperf.appspot.com/bad_bisect?try_job_id=594053 | O O | Visit http://www.chromium.org/developers/speed-infra/perf-bug-faq | X | for more information addressing perf regression bugs. For feedback, | / \ | file a bug with component Tests>AutoBisect. Thank you!
,
Apr 4 2016
Basically this is unavoidable, and fixing issue 109587 will recover this. I'm the author of blink_perf.dom/textarea-edit, and it's ok to have this performance regression for a while.
,
Apr 4 2016
===== BISECT JOB RESULTS ===== Status: completed ===== SUSPECTED CL(s) ===== Subject : TEXTAREA: Cutting last line without EOL should not remove the remaining EOL in the previous line. Author : tkent Commit description: Our TEXTAREA element assumes there is an extra trailing <BR> or \n if the last character of the value is \n. So we remove the last \n when we generate value from innerEditor content. However, the assumption was broken if we cut the last line content without EOL. Editing code doesn't add an extra <BR> in such case. We removed the last \n in the value mistakenly. Solution: We should make sure there no <BR> elements in innerEditor except the extra trailing one so that we can ignore it safely on generating TEXTAREA value. Also, we should make sure editing code doesn't add \n for placeholder newline. - HTMLTextAreaElement simply ignore the last <BR> on generating value. - HTMLTextAreaElement makes sure that innerEditor has the last <BR> after every editing operations. - InsertLineBreakCommand always inserts \n for plain text editing field, however it produces <BR> as an extra line break for TEXTAREA. - ReplaceSelectionCommand appends a <BR> for an InterchangeNewline at the end. - InsertTextCommand should not produce TabSpans for plain text editing. BUG= 522144 , 593184 Review URL: https://codereview.chromium.org/1785603002 Cr-Commit-Position: refs/heads/master@{#380385} Commit : 37f57d892222dc0ed816ae004074d7c80600ad46 Date : Thu Mar 10 10:54:04 2016 ===== TESTED REVISIONS ===== Revision Mean Value Std. Dev. Num Values Good? chromium@380320 0.673256 0.019992 5 good chromium@380375 0.699156 0.022742 5 good chromium@380382 0.655968 0.024032 5 good chromium@380384 0.672923 0.034417 5 good chromium@380385 0.536574 0.014831 5 bad <- chromium@380386 0.540741 0.018042 5 bad chromium@380389 0.554387 0.012838 5 bad chromium@380403 0.561227 0.022499 5 bad chromium@380430 0.543567 0.015298 5 bad Bisect job ran on: android_nexus9_perf_bisect Bug ID: 594053 Test Command: src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --also-run-disabled-tests blink_perf.dom Test Metric: textarea-edit/textarea-edit Relative Change: 19.26% Score: 99.9 Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/android_nexus9_perf_bisect/builds/1705 Job details: https://chromeperf.appspot.com/buildbucket_job_status/9016329272585091600 Not what you expected? We'll investigate and get back to you! https://chromeperf.appspot.com/bad_bisect?try_job_id=594053 | O O | Visit http://www.chromium.org/developers/speed-infra/perf-bug-faq | X | for more information addressing perf regression bugs. For feedback, | / \ | file a bug with component Tests>AutoBisect. Thank you! |
|||||||||||||||||||||
►
Sign in to add a comment |
|||||||||||||||||||||
Comment 1 by kouhei@chromium.org
, Mar 11 2016