44773 2010-08-27 09:43:22 -0700 REGRESSION: some tests no longer painting 2010-08-30 09:17:47 -0700 1 1 1 Unclassified WebKit Tools / Tests 528+ (Nightly build) PC OS X 10.5 RESOLVED FIXED P2 Normal --- 1 tony zimmermann abarth eric paroga webkit.review.bot zimmermann oldest_to_newest 270548 0 tony 2010-08-27 09:43:22 -0700 After http://trac.webkit.org/changeset/66223, three tests started failing pixel tests on the Chromium bots: fast/canvas/canvas-as-image.html fast/canvas/canvas-as-image-incremental-repaint.html fast/gradients/generated-gradients.html It seems that they're not painting the background at all. I'm building webkit-mac to see if this also happens in Safari. 270554 1 tony 2010-08-27 09:46:43 -0700 Yup, repros in Safari too. Here are the pixel results: http://trac.webkit.org/export/66223/trunk/LayoutTests/platform/mac/fast/canvas/canvas-as-image-expected.png http://build.chromium.org/buildbot/layout_test_results/webkit-rel-webkit-org/results/layout-test-results/fast/canvas/canvas-as-image-actual.png http://trac.webkit.org/export/66223/trunk/LayoutTests/platform/mac/fast/canvas/canvas-as-image-incremental-repaint-expected.png http://build.chromium.org/buildbot/layout_test_results/webkit-rel-webkit-org/results/layout-test-results/fast/canvas/canvas-as-image-incremental-repaint-actual.png http://trac.webkit.org/export/66223/trunk/LayoutTests/platform/mac/fast/gradients/generated-gradients-expected.png http://build.chromium.org/buildbot/layout_test_results/webkit-rel-webkit-org/results/layout-test-results/fast/gradients/generated-gradients-actual.png 270686 2 zimmermann 2010-08-27 13:04:24 -0700 Thanks for the report, I'll investigate tomorrow, about to go to bed. Are you sure these are the only tests which are broken? 270704 3 tony 2010-08-27 13:32:23 -0700 (In reply to comment #2) > Thanks for the report, I'll investigate tomorrow, about to go to bed. > Are you sure these are the only tests which are broken? These are the only layout tests that regressed on the chromium bots (across windows, mac and linux). It's possible for there to be a test that chromium was already failing, in which case a regression would not be noticed. 270883 4 tony 2010-08-27 18:44:23 -0700 I suspect fast/canvas/webgl/css-webkit-canvas-repaint.html and fast/canvas/webgl/css-webkit-canvas.html are failing for the same reason. I bet the pixel results were generated before r66223 was landed. 270937 5 65813 zimmermann 2010-08-27 23:53:53 -0700 Created attachment 65813 Patch 270938 6 65813 krit 2010-08-27 23:56:25 -0700 Comment on attachment 65813 Patch r=me 270939 7 zimmermann 2010-08-27 23:57:03 -0700 Landed in r66299. Tony, can you verify the regression is gone? All tests you mentioned work again for me in Safari. 270945 8 webkit.review.bot 2010-08-28 00:30:54 -0700 http://trac.webkit.org/changeset/66299 might have broken Qt Linux Release 271460 9 tony 2010-08-30 09:17:47 -0700 (In reply to comment #7) > Landed in r66299. > > Tony, can you verify the regression is gone? > All tests you mentioned work again for me in Safari. Yes, the regression is gone. Thanks! 65813 2010-08-27 23:53:53 -0700 2010-08-27 23:56:24 -0700 Patch FixImage.diff text/plain 2147 zimmermann SW5kZXg6IFdlYkNvcmUvQ2hhbmdlTG9nCj09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT0KLS0tIFdlYkNvcmUvQ2hhbmdlTG9n CShyZXZpc2lvbiA2NjI5OCkKKysrIFdlYkNvcmUvQ2hhbmdlTG9nCSh3b3JraW5nIGNvcHkpCkBA IC0xLDMgKzEsMTggQEAKKzIwMTAtMDgtMjcgIE5pa29sYXMgWmltbWVybWFubiAgPG56aW1tZXJt YW5uQHJpbS5jb20+CisKKyAgICAgICAgUmV2aWV3ZWQgYnkgTk9CT0RZIChPT1BTISkuCisKKyAg ICAgICAgUkVHUkVTU0lPTjogc29tZSB0ZXN0cyBubyBsb25nZXIgcGFpbnRpbmcKKyAgICAgICAg aHR0cHM6Ly9idWdzLndlYmtpdC5vcmcvc2hvd19idWcuY2dpP2lkPTQ0NzczCisKKyAgICAgICAg Rml4IHJlZ3Jlc3Npb24gZnJvbSB0aGUgUmVuZGVySW1hZ2VSZXNvdXJjZSBpbnRyb2R1Y3Rpb24s IGZhc3QvY2FudmFzL2NhbnZhcy1hcy1pbWFnZS5odG1sCisgICAgICAgIGFuZCBmYXN0L2NhbnZh cy9jYW52YXMtYXMtaW1hZ2UtaW5jcmVtZW50YWwtcmVwYWludC5odG1sIHBpeGVsIHRlc3RzIGRp ZCBub3Qgd29yayBhbnltb3JlLgorCisgICAgICAgICogcmVuZGVyaW5nL1JlbmRlckltYWdlUmVz b3VyY2UuaDoKKyAgICAgICAgKFdlYkNvcmU6OlJlbmRlckltYWdlUmVzb3VyY2U6Omhhc0ltYWdl KTogTWFrZSBoYXNJbWFnZSB2aXJ0dWFsLgorICAgICAgICAqIHJlbmRlcmluZy9SZW5kZXJJbWFn ZVJlc291cmNlU3R5bGVJbWFnZS5oOgorICAgICAgICAoV2ViQ29yZTo6UmVuZGVySW1hZ2VSZXNv dXJjZVN0eWxlSW1hZ2U6Omhhc0ltYWdlKTogQWx3YXlzIHJldHVybiB0cnVlLCBqdXN0IGxpa2Ug UmVuZGVySW1hZ2VHZW5lcmF0ZWRDb250ZW50IGRpZC4KKwogMjAxMC0wOC0yNyAgS2ltbW8gS2lu bnVuZW4gIDxraW1tby50Lmtpbm51bmVuQG5va2lhLmNvbT4KIAogICAgICAgICBSZXZpZXdlZCBi eSBLZW5uZXRoIFJvaGRlIENocmlzdGlhbnNlbi4KSW5kZXg6IFdlYkNvcmUvcmVuZGVyaW5nL1Jl bmRlckltYWdlUmVzb3VyY2UuaAo9PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09Ci0tLSBXZWJDb3JlL3JlbmRlcmluZy9SZW5k ZXJJbWFnZVJlc291cmNlLmgJKHJldmlzaW9uIDY2MjIzKQorKysgV2ViQ29yZS9yZW5kZXJpbmcv UmVuZGVySW1hZ2VSZXNvdXJjZS5oCSh3b3JraW5nIGNvcHkpCkBAIC00OSw3ICs0OSw3IEBAIHB1 YmxpYzoKIAogICAgIHZvaWQgc2V0Q2FjaGVkSW1hZ2UoQ2FjaGVkSW1hZ2UqKTsKICAgICBDYWNo ZWRJbWFnZSogY2FjaGVkSW1hZ2UoKSBjb25zdCB7IHJldHVybiBtX2NhY2hlZEltYWdlLmdldCgp OyB9Ci0gICAgYm9vbCBoYXNJbWFnZSgpIGNvbnN0IHsgcmV0dXJuIG1fY2FjaGVkSW1hZ2U7IH0K KyAgICB2aXJ0dWFsIGJvb2wgaGFzSW1hZ2UoKSBjb25zdCB7IHJldHVybiBtX2NhY2hlZEltYWdl OyB9CiAKICAgICB2b2lkIHJlc2V0QW5pbWF0aW9uKCk7CiAKSW5kZXg6IFdlYkNvcmUvcmVuZGVy aW5nL1JlbmRlckltYWdlUmVzb3VyY2VTdHlsZUltYWdlLmgKPT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PQotLS0gV2ViQ29y ZS9yZW5kZXJpbmcvUmVuZGVySW1hZ2VSZXNvdXJjZVN0eWxlSW1hZ2UuaAkocmV2aXNpb24gNjYy MjMpCisrKyBXZWJDb3JlL3JlbmRlcmluZy9SZW5kZXJJbWFnZVJlc291cmNlU3R5bGVJbWFnZS5o CSh3b3JraW5nIGNvcHkpCkBAIC00NSw2ICs0NSw3IEBAIHB1YmxpYzoKICAgICB2aXJ0dWFsIHZv aWQgaW5pdGlhbGl6ZShSZW5kZXJPYmplY3QqKTsKICAgICB2aXJ0dWFsIHZvaWQgc2h1dGRvd24o KTsKIAorICAgIHZpcnR1YWwgYm9vbCBoYXNJbWFnZSgpIGNvbnN0IHsgcmV0dXJuIHRydWU7IH0K ICAgICB2aXJ0dWFsIEltYWdlKiBpbWFnZShpbnQgd2lkdGggPSAwLCBpbnQgaGVpZ2h0ID0gMCkg eyByZXR1cm4gbV9zdHlsZUltYWdlLT5pbWFnZShtX3JlbmRlcmVyLCBJbnRTaXplKHdpZHRoLCBo ZWlnaHQpKTsgfQogICAgIHZpcnR1YWwgYm9vbCBlcnJvck9jY3VycmVkKCkgY29uc3QgeyByZXR1 cm4gbV9zdHlsZUltYWdlLT5lcnJvck9jY3VycmVkKCk7IH0KIAo=