I don't think resizing window is the most important part of the interactivity. But it's a great way to stress layout code.
Created attachment 152148 [details] Patch
Created attachment 152735 [details] Patch
Comment on attachment 152735 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=152735&action=review > Tools/Scripts/webkitpy/performance_tests/perftest.py:338 > + src_text = open(src_filename).read() > + dst_text = re.sub("<!-- WEBKIT_INTERACTIVE_TEST_TEMPLATE -->", Please spell out source and destination. I'd prefer calling it something like original & edited though. > Tools/Scripts/webkitpy/performance_tests/perftest.py:349 > + src_png = os.path.join(test_root, "Parser", "resources", "greenbox.png") > + dst_png = os.path.join(test_root, "Interactive", "resources", "greenbox.png") > + shutil.copy(src_png, dst_png) > + src_html = os.path.join(test_root, "Parser", "resources", "html5.html") > + dst_html = os.path.join(test_root, "Interactive", "resources", "html5.html") Ditto. > PerformanceTests/Parser/resources/html5.html:557 > + <!-- WEBKIT_INTERACTIVE_TEST_TEMPLATE --> Can we instead add a script element along the line of: <script> if (window.parent.interactiveTest) window.parent.interactiveTest(); </script> That'll avoid python change, and I don't think it'll add a statistically significant impact on the test result given the size of the page.
Created attachment 152906 [details] Patch
Ryosuke, thanks for taking look at this! I updated the patch to get rid of the modify-and-copy approach. It looks becoming much simpler.
Comment on attachment 152906 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=152906&action=review > PerformanceTests/Interactive/window-resize.html:4 > +window.location = "../Parser/resources/html5.html?interactiveTest=exerciseWindowResize"; Please make sure this technique works on Qt, Chromium, & Mac. > PerformanceTests/Parser/resources/html5.html:557 > + <script><!-- Do we really need <!-- ~ -->? > PerformanceTests/Parser/resources/html5.html:558 > + if (0 <= window.location.search.indexOf("interactiveTest")) { Maybe we can make it more genetic by specifying the path instead? e.g. html5.html?inject=Interactive/resources/driver.js and you do: document.write("<script src='../../" + injectedScript + "'></script>");
Created attachment 152968 [details] Patch for landing
(In reply to comment #6) > (From update of attachment 152906 [details]) > View in context: https://bugs.webkit.org/attachment.cgi?id=152906&action=review > > > PerformanceTests/Interactive/window-resize.html:4 > > +window.location = "../Parser/resources/html5.html?interactiveTest=exerciseWindowResize"; > > Please make sure this technique works on Qt, Chromium, & Mac. > > > PerformanceTests/Parser/resources/html5.html:557 > > + <script><!-- > > Do we really need <!-- ~ -->? > yeah, looks like... > > PerformanceTests/Parser/resources/html5.html:558 > > + if (0 <= window.location.search.indexOf("interactiveTest")) { > > Maybe we can make it more genetic by specifying the path instead? > e.g. html5.html?inject=Interactive/resources/driver.js > and you do: > document.write("<script src='../../" + injectedScript + "'></script>"); Good idea. did it.
Comment on attachment 152968 [details] Patch for landing Rejecting attachment 152968 [details] from commit-queue. Failed to run "['/mnt/git/webkit-commit-queue/Tools/Scripts/webkit-patch', '--status-host=queues.webkit.org', '-..." exit_code: 1 ERROR: /mnt/git/webkit-commit-queue/PerformanceTests/ChangeLog neither lists a valid reviewer nor contains the string "Unreviewed" or "Rubber stamp" (case insensitive). Full output: http://queues.webkit.org/results/13275491
Created attachment 153149 [details] Patch for landing
Comment on attachment 153149 [details] Patch for landing Clearing flags on attachment: 153149 Committed r123063: <http://trac.webkit.org/changeset/123063>
All reviewed patches have been landed. Closing bug.