[Scribus] Scribus Cashes ???

[Craig Ringer]

Roger wrote:
> Hi all
> Had the dreaded 'Scribus Crashes with fault 11" with 1.3.3.10svn and now this 
> afternoon with 1.3.4

A signal 11, ie SIGSEGV or segmentation fault, indicates that Scribus
has to access memory that it has not allocated. This can be caused by
reading a NULL pointer or a pointer to certain types of freed or never
allocated memory.

It practially always indicates a bug or failure to check for an error
condition somewhere. Unfortunately without a backtrace, preferably with
debug symbols enabled, or a core dump, there's not much that can be
done. If you can perform the same steps and it often crashes when you
perform those steps, that's something that can be investigated usefully
if you can describe the steps in tedious detail and provide any fonts,
images, and Scribus documents used in the process.

Note that 1.3.4 is an UNSTABLE DEVELOPMENT VERSION, essentially a sort
of alpha-quality preview snapshot. It is not maintained; all future work
is going into 1.3.5svn . A lot of people do understandably find the
versioning somewhat confusing (1.3.3.x stable, 1.3.x development) but
this is a historical accident that can hopefully be corrected soonish.

> I already lost some work through not saving every few minutes but I cant take 
> blame for this as I have been lulled into a deserved sense of security by 
> Scribus not faulting until yesterday.
> The only upgrade on my machine is cups all else is standard installation from 
> the cd.
> 
> I'll test with valgrind, etc.

Scribus will be unbearably, impossibly slow under Valgrind. It's usable
for a few things, but not much.

The best thing you can do is rebuild Scribus with debugging turned on,
run it under gdb, and get a backtrace. However, honestly, you might be
wasting your time doing this with 1.3.4 rather than 1.3.5svn ; I don't
know. The up-to-date dev team members can tell you that for sure.

--
Craig Ringer