Gramps crashes after some minutes

[riny]

Binary package hint: gramps

After some minutes Gramps crashes. If I want to start it direct after the crash it wil crash direct after restart. If I wait 5 minutes I have some minutes to work again.

ProblemType: Bug
Architecture: i386
CheckboxSubmission: f2dfd2e95cbf3a4787fe86bee47110bd
CheckboxSystem: 6ce041aeed0a2c17b3343b66d157175d
Date: Sun Oct 11 14:45:25 2009
DistroRelease: Ubuntu 9.10
NonfreeKernelModules: nvidia
Package: gramps 3.1.2-1
PackageArchitecture: all
ProcEnviron:
 LANG=nl_NL.UTF-8
 SHELL=/bin/bash
ProcVersionSignature: Ubuntu 2.6.31-12.41-generic
SourcePackage: gramps
Uname: Linux 2.6.31-12-generic i686

[riny (rinyfeller)]

Architecture: i386
CheckboxSubmission: f2dfd2e95cbf3a4787fe86bee47110bd
CheckboxSystem: 6ce041aeed0a2c17b3343b66d157175d
DistroRelease: Ubuntu 9.10
NonfreeKernelModules: nvidia
Package: gramps 3.1.2-1
PackageArchitecture: all
ProcEnviron:
 SHELL=/bin/bash
 LANG=nl_NL.UTF-8
ProcVersionSignature: Ubuntu 2.6.31-12.41-generic
Uname: Linux 2.6.31-12-generic i686
UserGroups: adm admin cdrom dialout lpadmin plugdev sambashare

[merovius (merovius)]

I seem to have this same issue under Karmic. I would appear to have fixed it by uninstalling GRAMPS with synaptic and then reinstalling from the website by downloading the .deb file. Did loose the last bit of work I did a few weeks ago.

[merovius (merovius)]

I forgot to mention that I did $mv ~/.gramps ~/.gramps.backup before I uninstalled and reinstalled via the .deb file. I have been advised that this technique did not work for another user. I am not sure if they used the $mv ~/.gramps ~/.gramps.backup command first when they tried the reinstall.

[Duncan Lithgow (duncan-lithgow)]

@riny: have you run the 'Check and Repair Database' tool? Does that report any problems? What about starting GRAMPS from the command line, that might give some error messages you can paste here.
@merovius: it's probably a better idea to export the databases you want to keep to the GRAMPS XML format before upgrading. Then re-import them.

[merovius (merovius)]

Update: The other user (sweevo) has reported his second attempt using $mv ~/.gramps ~/.gramps.backup worked and he was able to close and copy his old .gramps file back into the new one and all was well with his GRAMPS install.

The process is in this forum http://ubuntuforums.org/showthread.php?t=1288475

I had exported my file but did not import it. I just left the original in my home folder.

I did run from terminal and copied the errors to the above forum. Hope that is of some use.

I may try copying my old.gramps folder over to my new one and see if the data I seem to have lost returns. I have fairly recent backups so no worry over major data loss.

[Duncan Lithgow (duncan-lithgow)]

I understand that this is now working for each of you, but was the original problem a bug or something you did wrong. I'd like to either close this or forward it to the gramps developers.

[merovius (merovius)]

I could not say either way honestly. It was the first time I tried to use it after upgrading to Karmic Beta. It would crash almost immediately. After doing as suggested on the Ubuntu forum and backing up the .gramps folder and reinstalling it worked. The downloaded .deb worked but not reinstalling with synaptic. Not sure what the difference (if any) would be. I then filed a bug and found this same one already filed. Sorry if thats not much help. Thanks for your efforts in any event.

[Petr Hejl (hejlp)]

This bug is perfectly valid. It happens very often when changes are going to be saved to database (OK button).

I was using 3.1.2 provided by gramps download page in Ubuntu 9.04. Now I've upgraded to 9.10. The package was replaced by the package in repository. Now the application is crashing with Segmentation fault all the time.

[Petr Hejl (hejlp)]

Crashes also in other situations not just save. I checked the database with integrated tool and it is ok.

[Duncan Lithgow (duncan-lithgow)]

@Petr: could you please check if you have 'Assistive Technologies' enabled?

[Petr Hejl (hejlp)]

I'm not sure what is that and where to check it, but apport report is uploading right now.

In meantime I get this from gdb/core dump, so perhaps there is something wrong in libwebkit.

#0 0x00007fdbd61382d1 in ?? () from /usr/lib/libwebkit-1.0.so.2
#1 0x00007fdbe42dcaf8 in ?? () from /usr/lib/libgio-2.0.so.0
#2 0x00007fdbe617cbbe in g_main_context_dispatch () from /lib/libglib-2.0.so.0
#3 0x00007fdbe6180588 in ?? () from /lib/libglib-2.0.so.0
#4 0x00007fdbe61806b0 in g_main_context_iteration () from /lib/libglib-2.0.so.0
#5 0x00007fdbe518efbd in gtk_main_iteration_do () from /usr/lib/libgtk-x11-2.0.so.0
#6 0x00007fdbe57f6da7 in ?? () from /usr/lib/pymodules/python2.6/gtk-2.0/gtk/_gtk.so
#7 0x00000000004a290d in PyEval_EvalFrameEx ()
#8 0x00000000004a40e0 in PyEval_EvalCodeEx ()
#9 0x00000000004a245f in PyEval_EvalFrameEx ()
#10 0x00000000004a2e47 in PyEval_EvalFrameEx ()
#11 0x00000000004a2e47 in PyEval_EvalFrameEx ()
#12 0x00000000004a40e0 in PyEval_EvalCodeEx ()
#13 0x000000000052be30 in ?? ()
#14 0x000000000041d6e7 in PyObject_Call ()
#15 0x00000000004a0f31 in PyEval_EvalFrameEx ()
#16 0x00000000004a40e0 in PyEval_EvalCodeEx ()
#17 0x00000000004a245f in PyEval_EvalFrameEx ()
#18 0x00000000004a2e47 in PyEval_EvalFrameEx ()
#19 0x00000000004a40e0 in PyEval_EvalCodeEx ()
#20 0x000000000052be30 in ?? ()
#21 0x000000000041d6e7 in PyObject_Call ()
#22 0x00000000004254ff in ?? ()
#23 0x000000000041d6e7 in PyObject_Call ()
#24 0x00000000004a199c in PyEval_EvalFrameEx ()
#25 0x00000000004a2e47 in PyEval_EvalFrameEx ()
#26 0x00000000004a2e47 in PyEval_EvalFrameEx ()
#27 0x00000000004a40e0 in PyEval_EvalCodeEx ()
#28 0x000000000052be30 in ?? ()
#29 0x000000000041d6e7 in PyObject_Call ()
#30 0x00000000004254ff in ?? ()
#31 0x000000000041d6e7 in PyObject_Call ()
#32 0x000000000049c623 in PyEval_CallObjectWithKeywords ()
#33 0x00007fdbe5b00a4d in ?? () from /usr/lib/pymodules/python2.6/gtk-2.0/gobject/_gobject.so
#34 0x00007fdbe68225ae in g_closure_invoke () from /usr/lib/libgobject-2.0.so.0
#35 0x00007fdbe6837983 in ?? () from /usr/lib/libgobject-2.0.so.0
#36 0x00007fdbe6838d39 in g_signal_emit_valist () from /usr/lib/libgobject-2.0.so.0
#37 0x00007fdbe6839283 in g_signal_emit () from /usr/lib/libgobject-2.0.so.0
#38 0x00007fdbe50e7af5 in ?? () from /usr/lib/libgtk-x11-2.0.so.0
#39 0x00007fdbe68225ae in g_closure_invoke () from /usr/lib/libgobject-2.0.so.0
---Type <return> to continue, or q <return> to quit---
#40 0x00007fdbe683729c in ?? () from /usr/lib/libgobject-2.0.so.0
#41 0x00007fdbe6838d39 in g_signal_emit_valist () from /usr/lib/libgobject-2.0.so.0
#42 0x00007fdbe6839283 in g_signal_emit () from /usr/lib/libgobject-2.0.so.0
#43 0x00007fdbe50e67ad in ?? () from /usr/lib/libgtk-x11-2.0.so.0
#44 0x00007fdbe5195728 in ?? () from /usr/lib/libgtk-x11-2.0.so.0
#45 0x00007fdbe68225ae in g_closure_invoke () from /usr/lib/libgobject-2.0.so.0
#46 0x00007fdbe683764d in ?? () from /usr/lib/libgobject-2.0.so.0
#47 0x00007fdbe6838bcc in g_signal_emit_valist () from /usr/lib/libgobject-2.0.so.0
#48 0x00007fdbe6839283 in g_signal_emit () from /usr/lib/libgobject-2.0.so.0
#49 0x00007fdbe529c72f in ?? () from /usr/lib/libgtk-x11-2.0.so.0
#50 0x00007fd...

[Petr Hejl (hejlp)]

Update of libwebkit via ppa (webkit-team) to 1.1.16-2 didn't fix it.

[merovius (merovius)]

I Have done a clean install of Karmic and had no problems so far. I think there is a problem with Gramps and upgrading to Karmic. Gramps ran OK after I had fixed it as posted previously in my upgrade attempt. I am not running Assistive Technologies. I have not tested it extensively. But NO crashes so far.

[Serge Noiraud (serge-noiraud-laposte)]

I'm the gramps developper which include GeoView in gramps.
Geoview try to load libwebkit or gtkmozembed to show the maps.
If we have none of these kits, Geoview is not loaded.
If we have one kit, we load the kit available
and if we have the two kits, webkit is used.
If you don't need the maps for the moment, you can disallow geoview in the .ini file
$HOME/.gramps/gramps.ini or gramps32.ini
You remove 'GeoView' in the data-views variable.

This is for the work around. Normally, gramps must run correctly.

Now, for your problem, I think there is a problem during libwebkit and python-webkit installation.
During the karmic development, I already had this problem one time.
To correct this, I only did :
sudo aptitude remove python-webkit
then
sudo aptitude install python-webkit
and normally, the problem should disapear ...

[Petr Hejl (hejlp)]

Thanks Serge. The first think I tried was to disable GeoView by removing GeoView from data-views in keys.ini. It didn't help.

What helped is:
removing GeoView from data-views in keys.ini
removing geoview=True from keys.ini
removing temp/GeoView

[Serge Noiraud (serge-noiraud-laposte)]

I can reproduce the problem too. If it's the same problem than the bug 3341 on gramps-project bug list,
http://www.gramps-project.org/bugs/view.php?id=3341

** (gramps.py:5858): DEBUG: NP_Initialize
** (gramps.py:5858): DEBUG: NP_Initialize succeeded
 --- modulename: GeoView, funcname: page_loaded
GeoView.py(210): self.set_button_sensitivity()
 --- modulename: GeoView, funcname: set_button_sensitivity
GeoView.py(187): self.fct()
 --- modulename: GeoView, funcname: set_button_sensitivity
GeoView.py(420): self.forward_action.set_sensitive(self.renderer.can_go_forward())
 --- modulename: GeoView, funcname: can_go_forward
GeoView.py(169): return self.window.can_go_forward()
GeoView.py(421): self.back_action.set_sensitive(self.renderer.can_go_back())
 --- modulename: GeoView, funcname: can_go_back
GeoView.py(157): return self.window.can_go_back()
Segmentation fault (core dumped)

This is a libwebkit or a python-webkit problem :
I ask to webkit if I can go back ( go to the prior html page )
It should answer False because at this moment, we have no html page loaded.

As you can see, the can_go_forward is ok. This is used to have the back and forward button active or not.
self.window = webkit.webview()

If I use : python -m trace --count when I launch gramps, sometimes it work, sometimes it doesn't,
So The problem appears ( I suppose ) when webkit is not fully initialized.

Does it help ?
is it the same bug than https://bugs.launchpad.net/ubuntu/+source/gramps/+bug/388815 ?

[Serge Noiraud (serge-noiraud-laposte)]

You can use gtkmozembed in place of webkit to have a workaround
sudo aptitude remove python-webkit
sudo aptitude install python-gtkmozembed
Then run gramps.
You should not have this problem anymore.
If the problem always occurs, it's a gramps problem.
If it doesn't, it's a webkit problem.

[Serge Noiraud (serge-noiraud-laposte)]

As we call some webkit function earlier ( webkit problem )
I tried to solve this.
I made a patch : It works for me. can you test it ?

[Serge Noiraud (serge-noiraud-laposte)]

I don't know how to add a patch, so I copy it :
--- /usr/share/gramps/DataViews/GeoView.py.org 2009-11-25 10:52:12.000000000 +0100
+++ /usr/share/gramps/DataViews/GeoView.py 2009-11-25 10:53:15.000000000 +0100
@@ -417,8 +417,9 @@
         """
         Set the backward and forward button in accordance to the browser.
         """
- self.forward_action.set_sensitive(self.renderer.can_go_forward())
- self.back_action.set_sensitive(self.renderer.can_go_back())
+ pass
+ #self.forward_action.set_sensitive(self.renderer.can_go_forward())
+ #self.back_action.set_sensitive(self.renderer.can_go_back())

     def open(self, url):
         """
@@ -698,6 +699,11 @@
         """
         Specifies the UIManager XML code that defines the menus and buttons
         associated with the interface.
+ <placeholder name="CommonNavigation">
+ <toolitem action="Back"/>
+ <toolitem action="Forward"/>
+ <toolitem action="Refresh"/>
+ </placeholder>
         """
         if Config.get(Config.GEOVIEW_GOOGLEMAPS):
             alternate_map = "GoogleMaps"
@@ -709,11 +715,6 @@
             alternate_map = "MicrosoftMaps"
         return '''<ui>
           <toolbar name="ToolBar">
- <placeholder name="CommonNavigation">
- <toolitem action="Back"/>
- <toolitem action="Forward"/>
- <toolitem action="Refresh"/>
- </placeholder>
             <placeholder name="CommonEdit">
               <toolitem action="OpenStreetMap"/>
               <toolitem action="%s"/>
@@ -731,9 +732,9 @@
         Required define_actions function for PageView. Builds the action
         group information required.
         """
- HtmlView._define_actions_fw_bw(self)
- self.forward_action.set_sensitive(False)
- self.back_action.set_sensitive(False)
+ #HtmlView._define_actions_fw_bw(self)
+ #self.forward_action.set_sensitive(False)
+ #self.back_action.set_sensitive(False)
         self._add_action('OpenStreetMap', 'gramps-openstreetmap',
                          _('_OpenStreetMap'),
                          callback=self.select_openstreetmap_map,

[blaisemarion (info-blaisemarion)]

I got the same problem with Karmic and Gramps. It happens right after it is opened and the database is unlocked to load. Did anyone find a fix yet?

Thanks.

[Duncan Lithgow (duncan-lithgow)]

I haven't had this bug since I installed 3.1.3

Duncan

[nemarona (nemarona)]

python-webkit comes automatically installed with Ubuntu. This means that uninstalling it will also uninstall the ubuntu-desktop package, whose presence is crucial when upgrading to a new Ubuntu version. If you follow Serge's advice and remove and reinstall python-webkit, I suggest you do
sudo apt-get remove python-webkit
sudo apt-get install ubuntu-desktop
By the way, I tried this to no avail.
Any solution / fix / workaround to this bug?

[Duncan Lithgow (duncan-lithgow)]

@Benny: please don't mark as Fix Released without some explanation. Clearly asking users to uninstall webkit is not a sensible solution as nemarona explained above.