QGIS » QGIS Application

Also on Debian

I cannot reproduce this crash on my kubuntu 8.10 system, both with 1.0 branch and head. Is it possible that you attach a small example dataset and a little step-by-step tutorial to get the crash?

thanks,
Marco

Step-by-step:

using a shape of the administrative boundaries of Belgium (you can download it on launch qgis[[BR)
]

add a vector layer: for example BEL0.shp

open the map composer doing file/print

in the map composer, do "add a new map" and draw the map on the canvas

then "add scalebar"; QGIS crashes when I click on the canvas to draw the scalebar

you see?

I'm terribly sorry Moritz, I didn't realise that setting the ticket to worksforme, closed the ticket (though in hindsight this appears to be quite obvious).

In my case, I compiled QGIS myself (svn trunk and 1.0.0 branch), and I'm running Qt 4.4.3.

Apologies for the confusion, hopefully we can find a solution here ;-)

-carson

Replying to [comment:9 cfarmer]:

I'm terribly sorry Moritz, I didn't realise that setting the ticket to worksforme, closed the ticket (though in hindsight this appears to be quite obvious).

No offense taken ;-)

In my case, I compiled QGIS myself (svn trunk and 1.0.0 branch), and I'm running Qt 4.4.3.

Well, I've tried the gfoss packages, compiling via dpkg-buildpackage and via cmake+make and all three give the same result, also running Qt 4.4.3...

Checking with ldd on qgis and the different libgis components, my non-expert eyes don't see anything obvious.

I'll try the stable dev branch and trunk to see if there's any difference.

Apologies for the confusion, hopefully we can find a solution here ;-)

I hope so, too. Apart this bug report and one mention of the issue on the dev list r1 without any response, this just doesn't seem to be an issue for most people...

One thing we could try is to identify the relevant differences between Ubuntu 8.04 and 8.10, but I'm not sure what elements to look at. Are there no qgis devs listening that could help us here ?

Moritz

r1 http://lists.osgeo.org/pipermail/qgis-developer/2009-January/005620.html

Moritz

I cannot reproduce this problem on any of the systems I have access to (which of course does not mean it does not exist). To help trace down the problem, you could paste a debugger backtrace. Just run gdb with qgis:
run 'gdb qgis' in a terminal
and after the crash, type 'backtrace' and paste the output here.

If you have a qgis compiled with debugging output enabled, the output will be even better, containing line numbers of the relevant source files.

Marco

Replying to [comment:11 mhugent]:

I cannot reproduce this problem on any of the systems I have access to (which of course does not mean it does not exist).

I've been able to reproduce it with 1.0.0 release and with 1.0 dev-branch, but not with trunk, so something seems to have changed between the two...if someone could just give me a hint as to where to look in the code (i.e. where is the print dialog and process), this would help me avoid having to search for it (I'm not familiar at all with the qgis code).

To help trace down the problem, you could paste a debugger backtrace. Just run gdb with qgis:
run 'gdb qgis' in a terminal
and after the crash, type 'backtrace' and paste the output here.

If you have a qgis compiled with debugging output enabled, the output will be even better, containing line numbers of the relevant source files.

Ok, I'll try this as well, but the fact that it works in trunk is already a hint. I hope it's just something that needs to be backported.

Moritz

Replying to [comment:12 mlennert]:

Replying to [comment:11 mhugent]:

To help trace down the problem, you could paste a debugger backtrace. Just run gdb with qgis:
run 'gdb qgis' in a terminal
and after the crash, type 'backtrace' and paste the output here.

If you have a qgis compiled with debugging output enabled, the output will be even better, containing line numbers of the relevant source files.

Ok, I'll try this as well, but the fact that it works in trunk is already a hint. I hope it's just something that needs to be backported.

Ok, here's the debug message:

Debug: /home/mlennert/SRC/QGIS/Version-1_0/src/core/composer/qgscomposeritem.cpp: 84: (setSelected) entered.
Fatal: ASSERT failure in QList<T>::at: "index out of range", file /usr/include/qt4/QtCore/qlist.h, line 387

And I'll attach the backtrace of gdb.

Moritz

bump...

Could anyone knowledgable in these things have a look at the gdb output ?
And can anyone else reproduce this bug ?

For us it's somewhat of a showstopper because teaching students how to make good maps is a bit difficult when they can't even add a scalebar... ;-)
(Our current best solution seems to be to use the QuickPrint plugin and to import the resulting pdf into inkscape for editing...not ideal, but at least we can stick with QGIS.)

Moritz

I could reproduce it both on Windows 2000 and Mac OSX (10.5.5)

Alice

Note that you can avoid the problem by working with the english version:
in the QGIS options, "locale" tab ("paramètres du lieu" in french), tick the box "override system locale" ("forcer la nationalité du système" in french) and choose english(US), then restart QGIS.

Milestone Version 1.0.2 deleted