Compilation on Ubuntu 22.04, gcc 11.3 about fields2cover HOT 15 CLOSED

is it segfaulting? can you provide a bit more detail on the nature of the crash? I ask because I think I saw something similar.

from fields2cover.

It's pretty hard because it dies really silently. For example, when I was debugging it just to hover the mouse over the points list, the process simply exited.

from fields2cover.

Thanks for noticing it @jfulem.
I'm working on it. I've also noticed on the Python interface that I cannot concatenate some operations because it crash, so it's needed to do it step by step.
I'll come later with a solution or more tests (and a headache).

from fields2cover.

@Gonzalo-Mier, perfect. Thank you.

from fields2cover.

@jfulem I haven't been able to reproduce the error. I would like to ask you which GDAL and SWIG version are you using?
According to SWIG docs, version 4.1.0 solves some memory leaks on Python. You could try to improve the swig version, and check if this error still happens. Unfortunately, a few python tests fail on that version. I'll continue trying to fix it.

On the other hand, are you able to create a test on a Docker that replicates this issue? On the add_gdal_version_support branch I've update the Dockerfile to be able to modify the GDAL version at the beginning.

from fields2cover.

I used swing 4.0.2 and GDAL 3.4.3. I can make the test on a Docker, yes.

from fields2cover.

I wasn't able to replicate this issue: https://github.com/Fields2Cover/Fields2Cover/blob/fix_issue_47/tests/python/types/Path_test.py#L211-L212

I also created a docker with the same versions of your system and it works well. @jfulem, could you try to run the python tests of the branch fix_issue_47 to see if the test covers this issue correctly?

from fields2cover.

I will try this evening. My path contains approx. 6500 points.

from fields2cover.

I've updated the path on the test and it seems to also work without failure. Now it contains 173554 points. I don't know if we are missing something. Thanks for taking the time.

from fields2cover.

@Gonzalo-Mier, I checked the test. It looks like it's working fine. I will test it in my project using F2C for initial harvesters' path planning. But the test works fine.

from fields2cover.

@jfulem were you able to test it? Should I close this issue?

from fields2cover.

I was able to test it and it is still causing the issues. I'm trying to replicate it in the test. The problem is that I have an object which reads the field and creates the paths and the points. Then accessing X and Y from the created points from another python object basically crashes. To avoid the crash I have to do something like this:

states = list(path.states)
self.points = [item.point for item in states]
xp = [p.getX() for p in self.points] # to avoid crash
yp = [p.getY() for p in self.points] # to avoid crash

Then I can retrieve the points list in another class/object and I can iterate through the points there.

from fields2cover.

Hi @jfulem,
Please, try to make a unit test. I'm not able to debug it with this information. Thanks for your time

from fields2cover.

@Gonzalo-Mier , I tried to mimic our structure to reproduce the issue: https://github.com/jfulem/Fields2Cover/tree/fix_issue_47
But it looks like it's working just fine in the tests. So, I think the issue is somewhere else and you can probably close this issue.

from fields2cover.

I'll close it then. Please, feel free to re-open this issue if you find it again.

from fields2cover.