We want to add some description for each run, so that people can know what each run is testing.
Created attachment 142394 [details] Patch
Comment on attachment 142394 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=142394&action=review > Tools/ChangeLog:16 > + DESCRIPTION: Benchmark for DOM attributes that return a Node object. Nice! > Tools/Scripts/webkitpy/performance_tests/perftest.py:98 > + description_regex = re.compile(r'^description (?P<description>.*)$') We should probably ignore cases here :) > PerformanceTests/resources/runner.js:97 > + this.log("description " + description); Why don't we capitalize description and put a comma as in: "Description: ~~~" > PerformanceTests/resources/runner.js:126 > + this.logStatistics(this._results, this._description); It seems odd that a function named logStatistics also prints description. Why don't we put the code here instead?
Created attachment 142397 [details] patch for landing
(In reply to comment #2) > > Tools/Scripts/webkitpy/performance_tests/perftest.py:98 > > + description_regex = re.compile(r'^description (?P<description>.*)$') > > We should probably ignore cases here :) Done. > > PerformanceTests/resources/runner.js:97 > > + this.log("description " + description); > > Why don't we capitalize description and put a comma as in: "Description: ~~~" Done. > > PerformanceTests/resources/runner.js:126 > > + this.logStatistics(this._results, this._description); > > It seems odd that a function named logStatistics also prints description. Why don't we put the code here instead? Done.
Comment on attachment 142397 [details] patch for landing Clearing flags on attachment: 142397 Committed r117397: <http://trac.webkit.org/changeset/117397>
(In reply to comment #5) > (From update of attachment 142397 [details]) > Clearing flags on attachment: 142397 > > Committed r117397: <http://trac.webkit.org/changeset/117397> It broke perf tests on the bot: Running PageLoad/svg/files/42450-under the see.svg (53 of 71) Traceback (most recent call last): File "./Tools/Scripts/run-perf-tests", line 39, in <module> sys.exit(PerfTestsRunner().run()) File "/home/webkitbuildbot/slaves/release64bitWebKit2-perf/buildslave/qt-linux-64-release-wk2-perf-tests/build/Tools/Scripts/webkitpy/performance_tests/perftestsrunner.py", line 138, in run unexpected = self._run_tests_set(sorted(list(tests), key=lambda test: test.test_name()), self._port) File "/home/webkitbuildbot/slaves/release64bitWebKit2-perf/buildslave/qt-linux-64-release-wk2-perf-tests/build/Tools/Scripts/webkitpy/performance_tests/perftestsrunner.py", line 226, in _run_tests_set if self._run_single_test(test, driver): File "/home/webkitbuildbot/slaves/release64bitWebKit2-perf/buildslave/qt-linux-64-release-wk2-perf-tests/build/Tools/Scripts/webkitpy/performance_tests/perftestsrunner.py", line 240, in _run_single_test new_results = test.run(driver, self._options.time_out_ms) File "/home/webkitbuildbot/slaves/release64bitWebKit2-perf/buildslave/qt-linux-64-release-wk2-perf-tests/build/Tools/Scripts/webkitpy/performance_tests/perftest.py", line 194, in run self.output_statistics(self.test_name(), results) File "/home/webkitbuildbot/slaves/release64bitWebKit2-perf/buildslave/qt-linux-64-release-wk2-perf-tests/build/Tools/Scripts/webkitpy/performance_tests/perftest.py", line 132, in output_statistics if results['description']: KeyError: 'description'
(In reply to comment #6) > It broke perf tests on the bot: looking
Comment on attachment 142397 [details] patch for landing View in context: https://bugs.webkit.org/attachment.cgi?id=142397&action=review > Tools/Scripts/webkitpy/performance_tests/perftest.py:132 > + if results['description']: Oops, you can't assume that results['description'] is always yet :(
Reopening to attach new patch.
Created attachment 142431 [details] Patch
Comment on attachment 142431 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=142431&action=review > Tools/Scripts/webkitpy/performance_tests/perftest.py:193 > + 'description': '', How about ChromiumStylePerfTest ?
(In reply to comment #11) > How about ChromiumStylePerfTest ? I want to fix it and confirm the bahavior, but how can I run ChromiumStylePerfTest?
(In reply to comment #12) > (In reply to comment #11) > > How about ChromiumStylePerfTest ? > > I want to fix it and confirm the bahavior, but how can I run ChromiumStylePerfTest? Tests in PerformanceTests/Inspector. However, they're skipped now.
Created attachment 142434 [details] patch for landing
Committed r117410: <http://trac.webkit.org/changeset/117410>
(In reply to comment #15) > Committed r117410: <http://trac.webkit.org/changeset/117410> perf test is still broken, but with another error message: Uploaded JSON but got a bad response: The payload doesn't contain results or results are malformed
It seems that we need to fix the server side too. Just a moment...
Created attachment 142446 [details] Patch
Comment on attachment 142446 [details] Patch LGTM, r=me.
Committed r117422: <http://trac.webkit.org/changeset/117422>
And I realized that unit tests are bleeding too: http://build.webkit.org/builders/Qt%20Linux%20Release/builds/47128/steps/webkitpy-test/logs/stdio
(In reply to comment #22) > And I realized that unit tests are bleeding too: http://build.webkit.org/builders/Qt%20Linux%20Release/builds/47128/steps/webkitpy-test/logs/stdio Is this failure still happening even after landing r117422? Now 'results' does not have 'description'.
(In reply to comment #23) > (In reply to comment #22) > > And I realized that unit tests are bleeding too: http://build.webkit.org/builders/Qt%20Linux%20Release/builds/47128/steps/webkitpy-test/logs/stdio > > Is this failure still happening even after landing r117422? Now 'results' does not have 'description'. Oh, sorry for the noise r117422 fixed all of them, but bots are in late.