85631 2012-05-04 09:09:28 -0700 [Qt] REGRESSION: 350 tests started to fail with newer Qt5-WK1 2012-10-09 01:28:03 -0700 1 1 1 Unclassified WebKit Tools / Tests 528+ (Nightly build) All All RESOLVED FIXED Qt, QtTriaged P1 Critical --- 79666 1 ossy pierre.rossi hausmann ossy oldest_to_newest 616276 0 ossy 2012-05-04 09:09:28 -0700 After this update 360 tests started to fail with Qt5-WK1: https://lists.webkit.org/pipermail/webkit-qt/2012-May/002715.html The root of the problem is something near font handling. a159ca80d73f07d617d6edd9b3e59bda25449a1c in qtbase made almost all tests (5000+) fail on Qt5-WK1. (But pass on Qt4-WK1 and Qt5-WK2) Pierre's following patch in testfonts fixed the problem more or less, but after this fix there are still 360 failing tests with Qt-WK1. commit f552eb4e2ef2773f03a63075683a360b5095c667 Author: Pierre Rossi <pierre.rossi@nokia.com> Date: Thu May 3 13:39:56 2012 +0200 Add one more aliasing rule to not mess with the Qt4 baseline. The approach consisting in hard coding font families in QFont::defaultFont resulted in interesting results. While the default font was always intended to be a serif font, it returned "Times" which somehow resolved to "Nimbus Sans L". Since QFont::defaultFont now returns "Nimbus Roman No9 L" early on, we need this additional alias so that it later on gets the same test font. diff --git a/fonts-qt5-wk1.conf b/fonts-qt5-wk1.conf index 07e7f18..332bcdd 100644 --- a/fonts-qt5-wk1.conf +++ b/fonts-qt5-wk1.conf @@ -7,6 +7,15 @@ --> <match target="pattern"> <test qual="any" name="family"> + <string>Nimbus Roman No9 L</string> + </test> + <edit name="family" mode="assign"> + <string>Times</string> + </edit> + </match> + + <match target="pattern"> + <test qual="any" name="family"> <string>serif</string> </test> <edit name="family" mode="assign"> 616304 1 ossy 2012-05-04 09:33:55 -0700 I skipped them - http://trac.webkit.org/changeset/116109 737528 2 ossy 2012-10-09 01:28:03 -0700 The reamining tests needed only rebase - r130735