Apache OpenOffice (AOO) Bugzilla – Issue 61515
c_updt.bas fails in tToolsXMLFilterSettings
Last modified: 2013-08-07 15:06:28 UTC
attached is resultfile; Owner of testcase tToolsXMLFilterSettings is JSI c_updt.bas owner is PJ Bad about that is: the testcase specification doesn't run to its end.
Created attachment 33786 [details] resultfile with error
JSI takes it
JSI -> SUS: The testcase "tToolsXMLFilterSettings" is freezing (only in Calc, only Linux) the office. I was nbot be able to do it manually but we have the defined test scenario in c_updt.bas
cc'ing PJ, JSI: JSI->PJ: I will exclude the testcase from your Calc update test include file that the complete test can run again. SRC680m150 was the last successfull complete run. So, there seems to be a change from m151 on.
#8 0x4194096c in __pthread_sighandler () from /lib/i686/libpthread.so.0 #9 <signal handler called> #10 0x41b7fb71 in kill () from /lib/i686/libc.so.6 [.. 3 libc frames omitted ] #14 0x41b80e8c in abort () from /lib/i686/libc.so.6 #15 0x4cc64515 in os::abort(int) () from /usr/java/jre1.5.0_03/lib/i386/client/libjvm.so #16 0x4ccecc54 in VMError::report_and_die() () from /usr/java/jre1.5.0_03/lib/i386/client/libjvm.so #17 0x4cc68982 in JVM_handle_linux_signal () from /usr/java/jre1.5.0_03/lib/i386/client/libjvm.so #18 0x4cc660b4 in signalHandler(int, siginfo*, void*) () from /usr/java/jre1.5.0_03/lib/i386/client/libjvm.so #19 0x41940a05 in __pthread_sighandler_rt () from /lib/i686/libpthread.so.0 #20 <signal handler called> #21 0xbfffbb34 in ?? () Previous frame inner to this frame (corrupt stack?) (gdb) quit looks like a crash in Java, so probably in the dialog. It hangs because Java locks in shutdown.
Now, in m154, we ahve it on all platforms. Onliest difference: The other test runs running after detecting that error.... Linux freezes the test.
SUS->PL: When I took a look through the pp2 CWS with changes on the filter project since m152, I recognized your fixes/enhancements on the xmfiltersettings dialog in VCL49. There seem to be some regression when being tested by the testtool. We created a stacktrace #1211427 by testing on my PC. 00ead328 603f32e2 xsltdlg680mi!XMLFilterTestDialog::~XMLFilterTestDialog( void )+0x5f [o:\src680\src.m152\filter\source\xsltdialog\xmlfiltertestdialog.cxx @ 276] There you see, that it is most likely that your changes brought up some existing problem, calling 'delete' twice on a reference. I workaround this and tested the changed DLL. Although both work manually, the testtool fails without an fix. You might want to take a look on it.. BTW was your CWS tested by the testtool?
Created attachment 33882 [details] Diff of fix for testtool
pl->sus: if that patch fixes the problem, then it's obvious that someone is destroying your XMLFilterTestDialog twice. That should rather be fixed than surviving the destructor on an already dead object in my opinion. Now the changes in xmlfiltersettingsdialog have not changed the lifetime of the testdialog at all, so i fail to see (it's only used as a stack object, this was not changed), how this would be able to affect the test dialog at all is beyond me.
Evaluation has been finished: The behaviour is different from the past-VCL49 MASTER (=>see: #i52570#. A XSLT document which is being tested can not be closed anymore if the 'Test XSLT' dialog is opened. The test failed also in the VCL49 CWS but has been ignored an manually it is impossible to close the imported XSLT document.
SUS->PL, JSI: The problem occurs, when only an untitled/empty document is open, when the XML FILTER SETTINGS dialog is opened. When a Test Dialog is opened and an document is opend (for the same appication as the open document) by browse, the empty document is exchange by the test document (a new view in an old frame is loaded). By this the window hierachy is changed, which is a problem for modal dialog. Please discuss if to change the modal dialog or the automated test.
take it back to me to close it, the test have to be changed...
The test have to be changed, after discussion with LO and TPF.
closed
reopen
reassign
discussed with jsi, sus: we will backout the "fix" (actually a workaround the Sunray XServer) for issue 52570, since it a) introduces a usability issue by making a dialog modal that wasn't before b) creates a crash in the test tool Issue 52570 will be reopened and set to 2.0.3 respectively, in which timeframe i will seek another workaround for that
fixed in CWS fix61515
please verify in CWS fix61515 re-open issue and reassign to jsi@openoffice.org
reassign to jsi@openoffice.org
reset resolution to FIXED
verified in CWS fix61515
verified in MWS m162