Apache OpenOffice (AOO) Bugzilla – Issue 32314
crash when window new and make a reload.
Last modified: 2005-01-24 10:44:27 UTC
- occures in the cws impress15ea - start an Impress / Draw - click on 'Window/New' - save the (new) doc - create an object (e.g. a rectangle) - make a reload - click on Yes - you get a crash. - was ok in src680m40
Accepted.
change prio
The reason for the crash is maybe an invalid controller: The stacked shells are created asynchronously for a new ViewShellBase. This is also true for the controller. In the short time between creation of ViewShellBase and that of ViewShell (and with it the controller) a default SfxBaseController is used. For some reason this seems to be used as SdUnoDrawView and thus crashes. If this is true, then to fix this bug we have to make the creation of the initial stacked shells synchronous. This has to be done anyway for making Draw and Impress as OLE objects work.
Seems like issue 35524 fixed this crash. What remains are assertions probably caused by other bugs (see issue 35524 for details.) To be sure I have to speak with MBA about them.
The other problems have to be solved under their own issues.
Step 1/3 of sending issue to QA.
Step 2/3 of sending issue to QA.
Step 3/3 of sending issue to QA.
Verified in CWS.
Reopened.
Crashes master m63. Back to AF.
This new crash is related to but not the same as the original crash. It still is an invalid controller that causes the crash. But this time it is the ViewShellBase::UpdateController() method that is called once too often. It is called from ViewShell::Init() and thus right after the creation of a new ViewShellBase object. This is unnecessary since the SFX code that creates the ViewShellBase also registers the new controller (which is done in UpdateController()). In the case of the reload it is not just unnecessary but dangerous and causes the crash. Can be fixed by simply remove the call to UpdateController() from ViewShell::Init(). There is still the call from the PaneManager.
back to AF
Fixed in CWS impress21beta by removing the call to UpdateController() as described above.
See issue 38341 for a related issue that may even cover this one.
Reopening...
...reassigning...
...resolving
According to WG the crash is fixed on Windows and Linux but still occurs on Solaris.
The crash that still occurs on Solaris has the same root cause as issue 38341. *** This issue has been marked as a duplicate of 38341 ***
According to AF I close this issue as duplicate.