Open MTRNord opened 5 months ago
Hi Marcel. This is a difficult one, as your distribution is not the most common.
I do have one question:
In any case, I would suggest the following:
@Jake-Moss , do you use Fedora by any chance? Anything you could add here?
Hi :) It is Python 3.12.3
.
Qgis luckily has qgis --noplugins
which gets me back into the qgis UI. I will give a reinstall a try and do a manual install. I am also meanwhile trying to run it in gdb (Downloading all the debug info is taking a moment right now) and see if maybe that gives a usable backtrace as well. :)
I would be curious to see what you find out! Installing the dependencies manually may indeed be your best bet for now indeed!
So it seems the crash originates actually from pyproj's Measure type :thinking: which triggers a INTERNAL_SYSCALL_ERROR_P
/a pthread kill.
The backtrace is massive: https://gist.github.com/MTRNord/6b31540232fa9f8963121c655679e425 and this is even without all the debug info. I will try the manual package install next now instead and also will check if proj maybe has a known issue about this.
It seems like manually installing the dependencies doesn't fix the issue here. :/
@Jake-Moss , do you use Fedora by any chance? Anything you could add here?
Not on my main system but my laptop is running Fedora. I'll be able to give this a test next week
The dependency installation already seemed spotty.
We're unfortunately aware of just how spotty this is. Hopefully [1] will provide a fix for that, in the mean time @pedrocamargo qpip [2] might be a good alternative if we want to shift away from managing our own dependency installation.
I'm going through the motions of setting up my laptop for qgis now.
[1] https://github.com/qgis/QGIS-Enhancement-Proposals/issues/202
[2] https://github.com/opengisch/qpip
Laptop is finally done updating. I can reproduce this on both wayland and X so I don't believe it's display server related.
Running under GDB doesn't yield anything useful. What makes matters worse is that with a couple of symlink hacks and the git history I can reproduce this on all compatible versions of qaequilibrae (all the way back to 0.4). I'll aim to make this little test more robust when I dig into the plugin code tomorrow.
Apologies I was unable to get to this today. I'm not able to reproduce this particular crash when using QGIS from a flatpak, however that raises other issues like dependency installation which I've previously been able to get working.
BEFORE REPORTING A BUG
Describe the bug On activating or opening qgis with the plugin enabled it throws:
To Reproduce Steps to reproduce the behavior:
ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.
error)Expected behavior It works
Screenshots Not applicable
Desktop (please complete the following information):
Additional context The dependency installation already seemed spotty. This may be related to the issue. I havent found a way to get more debug logs out of qgis however I am happy to provide them if there are commands to run or places to look for