Description of problem: With a Postscript file loaded into evince hitting "Ctrl-R" or picking "Reload" from a menu most likely will bring to a screen "crash alert", although this is somewhat unpredictable, and various messages will show up in a terminal window if you are running from a terminal. What will show up is not exactly deterministic but something like that seems to be the most popular: (evince:4560): GLib-GObject-WARNING **: invalid uninstantiatable type `(null)' in cast to `EvDocument' (evince:4560): GLib-GObject-WARNING **: invalid uninstantiatable type `(null)' in cast to `GObject' (evince:4560): GLib-GObject-WARNING **: instance of invalid non-instantiatable type `(null)' (evince:4560): GLib-GObject-CRITICAL **: g_signal_emit_valist: assertion `G_TYPE_CHECK_INSTANCE (instance)' failed I have seen also messages like: GLib-GObject-WARNING **: invalid uninstantiatable type `GEnum' in cast to `EvDocument' and like the first three quoted but with "type `(null)'" replaced by "type `<invalid>'". It may happen also that nothing is printed but basically evince is dead after such experiment regardless of what was displayed. This happens equally well on x86 and x86_64 installations. Version-Release number of selected component (if applicable): evince-0.1.5-1 How reproducible: 100%
I guess that the following can be added to the list above: (evince:5962): Gdk-CRITICAL **: gdk_region_destroy: assertion `r != NULL' failed
Fixed upstream, will be in the next release.
Created attachment 113577 [details] a log from an attempt to "reload" something in evince This bug got status CLOSED RAWHIDE but with evince-0.2.0-1 with a build date "Wed 06 Apr 2005 04:07:46 AM MDT" the situation is actually even worse than it used to be. Either hitting Ctrl-R or picking up "Reload" from a menu scrambles display and effectively completely crashes evince. This attachment includes a record from such attempt.
Is 0.2.1 better?
Created attachment 113754 [details] another backtrace from evince; this time 0.2.1-1 > Is 0.2.1 better? I would not say so. I see either alerts "application quit unexpectedly" and/or backtraces of that sort as attached. It is similar but not really the same as the one from the previous attachment so here is another sample from 0.2.1.
Created attachment 116274 [details] yet another traceback, this one for v. 0.3.1
Comment on attachment 116274 [details] yet another traceback, this one for v. 0.3.1 This problem persists in Fedora Core 4, which ships with evince 0.3.1. I have attached yet another backtrace. The bug status still is NEEDINFO, what further information would be helpful?
I compiled evince 0.3.2, and this bug appears to have been fixed.
I get only ** (evince:14727): CRITICAL **: ps_document_next_page: assertion `gs->busy != TRUE' failed on start and each reload, and no problems in GUI. Using evince 0.3.1 on FC4 up to date. Would you please attach your problematic .ps file or post a link to it?
With evince-0.4.0-1 I can hit Ctrl-R, or pick up "Reload" from a menu, many times without causing crashes. OTOH each such operation results in: (evince:28032): GLib-GObject-CRITICAL **: g_object_unref: assertion `G_IS_OBJECT (object)' failed for a change.
I can't reproduce this. I tried with evince-0.5.5 and a bunch of different ps files. If this is still a problem, please reopen, but do attach the problematic ps file in that case. thanks, Kristian
> I can't reproduce this. Well, a date on the original report was "2005-02-28". :-) Also I did not bother with evince for quite a while.