bm, I also can't reproduce this issue. CGU can reproduce it on his jds in m78
and m80. I suspekt its a jds only problem. Maybe the GTK on the stack is a hint?

Created attachment 22990 [details]
crashreporter stack

Created attachment 23018 [details]
Stack trace obtained when it crashed on Suse 9.3 beta

Now, I also got a stack (starting with a niceness of -11 helped):

#0  0xb67bff6f in OutlinerView::GetSelection (this=0x0) at
/work/dev/impress36/SRC680/svx/source/outliner/outlvw.cxx:1416
#1  0xad655855 in sd::Outliner::GetNextSpellSentence (this=0xaba21710) at
/work/dev/impress36/SRC680/sd/source/ui/view/Outliner.cxx:504
#2  0xad70e703 in sd::SpellDialogChildWindow::GetNextWrongSentence
(this=0xac640be8) at
/work/dev/impress36/SRC680/sd/source/ui/dlg/SpellDialogChildWindow.cxx:141
#3  0xab98990f in svx::SpellDialog::GetNextSentence_Impl (this=0xaba42f28,
bUseSavedSentence=false) at _uninitialized.h:63
#4  0xab989c57 in svx::SpellDialog::SpellContinue_Impl (this=0xaba42f28,
bUseSavedSentence=false) at SpellDialog.hxx:180
#5  0xab98a970 in svx::SpellDialog::ChangeHdl (this=0xaba42f28) at
/net/jumbo.germany/sol1/SRC680/src/svx/source/dialog/SpellDialog.cxx:563
[...]

As you can see the view returned by GetView(0) of the outliner is NULL. Well, I
suspected this, but now I have a starting point for debugging ...

->AF: I haven't found out much more. When setting breakpoints in
Outliner::InsertView() and RemoveView() I never get the crash. It would be
interesting which code calls the last RemoveView(0) just before the crash. I
have added some code to my svx/.../outliner.cxx that prints the parameters and a
stack trace for these functions. If you like I can give you the patched
libraries, or I try to get an output for a crash and attach it to this issue.
(Maybe you have an idea what happens with the help of the above backtrace.)

Accepted.

I applied the simple fix: In GetNextSpellSentence() the OutlinerView is checked
for being NULL.  When so then it is not accessed.

Why the view is NULL in the first place I do not yet know.

Back to QA for verification.

re-open issue and reassign to cgu@openoffice.org

reassign to cgu@openoffice.org

reset resolution to FIXED

*** Issue 44687 has been marked as a duplicate of this issue. ***

Now I don't get a crash but the corrected word would not be replaced.
also linux only

Back to af

I am setting the resolution again to FIXED because a) the crash does not occur
anymore and b) the problem with the text not being substituted is duplicate to
(or at least has the same root cause as) issue 44399.

Back to QA.

re-open issue and reassign to cgu@openoffice.org

reassign to cgu@openoffice.org

reset resolution to FIXED

Verified in cws impress37

*** Issue 45916 has been marked as a duplicate of this issue. ***

Tested in master m91. CLosed.