User-Agent: Mozilla/5.0 (X11; U; Linux ppc; en-US; rv:1.7) Gecko/20040823 Firefox/0.9.3 Build Identifier: Mozilla/5.0 (X11; U; Linux ppc; en-US; rv:1.7) Gecko/20040823 Firefox/0.9.3 This bug contains the widget/src/qt patch. Reproducible: Always Steps to Reproduce:
Created attachment 159277 [details] widget/src/qt dir Updated the patch to work with HEAD.
ok, so once I implemented nsFontEnumeratorQt::GetDefaultFont with a no-op impl, I got past the unresolved symbol; but now I crash at startup: (gdb) bt #0 0x419e273b in QApplication::construct () from /usr/lib/libqt.so.3 #1 0x419e2b65 in QApplication::QApplication () from /usr/lib/libqt.so.3 #2 0x41782e5a in nsAppShell::Create (this=0x8142e70, argc=0x0, argv=0x0) at /home/chb/mozilla/widget/src/qt/nsAppShell.cpp:73 #3 0x4173f4e9 in nsAppShellService::Initialize (this=0x8142dc8, aCmdLineService=0x8142830, aNativeAppSupportOrSplashScreen=0x0) at /home/chb/mozilla/xpfe/appshell/src/nsAppShellService.cpp:166 #4 0x0804dcb0 in main1 (argc=1, argv=0xbfffee94, nativeApp=0x0) at /home/chb/mozilla/xpfe/bootstrap/nsAppRunner.cpp:1209 #5 0x0804ef93 in main (argc=1, argv=0xbfffee94) at /home/chb/mozilla/xpfe/bootstrap/nsAppRunner.cpp:1799 I suppose Qt doesn't like null argc/argv. The reason this is null is: #3 0x4173f4e9 in nsAppShellService::Initialize (this=0x8142dc8, aCmdLineService=0x8142830, aNativeAppSupportOrSplashScreen=0x0) at /home/chb/mozilla/xpfe/appshell/src/nsAppShellService.cpp:166 166 rv = mAppShell->Create(0, nsnull); My qt version: qt3-non-mt-3.3.1-41.14 qt3-devel-3.3.1-33 qt3-devel-tools-3.3.1-32 qt3-3.3.1-36.16
Created attachment 159368 [details] Fixes the crash
Created attachment 160091 [details] Adapt to current CVS Adding new patch to make it compile with current CVS
Thanks, but it won't make too much difference. I had this locally for a while but I'm just rewritting lots of it so it won't make a difference. nsCommonWidget can't be the one receiving the events, the only object which needs to propagate Qt events is the nsWindow and I'm exporting the dispatcher there. Also it seems like the drawing widget needs to be different than our window widget since otherwise we get event storms (e.g. a resize event is received and Resize is called -> we resize internal widget -> resize signal from Qt is received -> we dispatch resize event to Mozilla -> Resize is called... ). Of course until this lands in CVS, and it doesn't look like it will happen anytime soon, everyone will have to deal with my busy schedule at the moment :(
Created attachment 160166 [details] updated widget/src/qt directory Some updated. Use with the new GFX patch.
Code is now in CVS.
vrfy fixed. qt is now in the tree.