Version: 1.3.0 (using KDE 3.2.3, compiled sources) Compiler: gcc version 3.3.3 OS: Linux (i686) release 2.4.26 1. Load Umbrello, C++ must be already selected 2. Select an old style code generator like PHP 3. Click in the toolbar on new document 4. Change active language to C++ -> Crash Maybe this is also the reason for bug 77645. But if I delete all datatypes after step 1, it doesn't crash. backtrace: Program received signal SIGSEGV, Segmentation fault. 0x40eab96b in QString::QString () from /usr/lib/qt/lib/libqt-mt.so.3 (gdb) bt #0 0x40eab96b in QString::QString () from /usr/lib/qt/lib/libqt-mt.so.3 #1 0x08180a17 in UMLObject::getName (this=0x84c2240) at umlobject.cpp:82 #2 0x0825a46c in CPPSourceCodeDocument::syncNamesToParent (this=0x84ca738) at cppsourcecodedocument.cpp:73 #3 0x0810388b in ClassifierCodeDocument::synchronize (this=0x84ca738) at classifiercodedocument.cpp:393 #4 0x0825791c in CPPCodeGenerator::syncCodeToDocument (this=0x84c8d28) at cppcodegenerator.cpp:219 #5 0x0825846c in CPPCodeGenerator::qt_invoke (this=0x84c8d28, _id=7, _o=0xbfffe6b0) at cppcodegenerator.moc:94 #6 0x40be3eac in QObject::activate_signal () from /usr/lib/qt/lib/libqt-mt.so.3 #7 0x40be3ce4 in QObject::activate_signal () from /usr/lib/qt/lib/libqt-mt.so.3 #8 0x08114c1b in CodeGenerationPolicy::modifiedCodeContent (this=0x84c8858) at qmetaobject.h:261 #9 0x0826f629 in CPPCodeGenerationPolicy::setDefaults (this=0x84c8858, clone=0x0, emitUpdateSignal=true) at cppcodegenerationpolicy.cpp:376 #10 0x0815addd in UMLApp::setGenerator (this=0x83e17c0, gen=0x84c8d28, giveWarning=102) at uml.cpp:1120 #11 0x0815b380 in UMLApp::setActiveLanguage (this=0x83e17c0, id=-80) at uml.cpp:1213
I mark all bugs as high, which we should try to fix prior KDE 3.3.
On Thu, 2004-07-29 at 15:21, Sebastian Stein wrote: > [...] > Priority|NOR |HI > > ------- Additional Comments From seb_stein gmx de 2004-07-29 15:21 ------- > I mark all bugs as high, which we should try to fix prior KDE 3.3. > Lieber Sebastian, Ich werde froh sein wenn ich 71969 noch rechtzeitig eintuete und sehe kaum eine Chance dass ich diese weiteren HI-prio bugs bis 1.3 fixe... Das nur am Rande. Sieht man sich in Ludwigsburg? Viele Gruesse, Oliver
I got a different backtrace for the same steps... here it seems the codegenerator is a null pointer again in CPPSourceCodeDocument::updateContent(), so gen->findObjectsRelated() causes the crash. Again, a simple fix would be to test for 'null' here, but I think you have a general problem with not available code generators... Ciao, Tobias
This seems to be fixed in HEAD but not in KDE 3.3 branch. Anyone know what fixed it and therefor what should be backported?
Jonathan Riddell <jr@jriddell.org> [040825 19:39]: > This seems to be fixed in HEAD but not in KDE 3.3 branch. Anyone know > what fixed it and therefor what should be backported? I can't reproduce it either, but I don't know how it was fixed. I think we shouldn't close the report till it is fixed in 3.3 as well. Sebastian
I believe this to be fixed now on KDE_3_3_BRANCH, could you check? Thanks.
Oliver Kellogg <okellogg@users.sourceforge.net> [040826 22:27]: > I believe this to be fixed now on KDE_3_3_BRANCH, > could you check? Thanks. Yes, it works. Nevertheless, I think it creates memory leaks. For 3.3 this is ok, for 3.4 I hope we could start fixing this leaks. Sebastian