Bug report #777
crashes with python reference to destroyed objects (e.g exportToWkt() crashes when geom has been deleted)
Status: | Closed | ||
---|---|---|---|
Priority: | High | ||
Assignee: | - | ||
Category: | Python plugins | ||
Affected QGIS version: | master | Regression?: | No |
Operating System: | All | Easy fix?: | No |
Pull Request or Patch supplied: | No | Resolution: | |
Crashes QGIS or corrupts data: | Yes | Copied to github as #: | 10836 |
Description
When a feature is destroyed, if it has ownership over a geometry, it deletes that geometry. After that point, if one tries to call functions on the geometry, it will return nulls (Linux) or cause a KERN_PROTECTION_FAILURE (OS X).
This can be reproduced by selecting a feature, then typing the following into the Python console:
iface.activeLayer().selectedFeatures()r0.geometry().exportToWkt()
(Sometimes it requires calling it twice to reproduce the crash -- however, it will never return the correct answer.)
The reason for this appears to be that the feature created when it is pulled out of the list is then destoryed before exportToWkt() is called, taking the geometry with it.
A workaround is to instead call geometryAndOwnership(), which tells the feature to not destroy the geometry.
It seems like this problem may actually be exportGeosToWkb -- functions like wkbType() fail in the same way. It's possible that the mGeometry check at the beginning of these functions needs to move before teh exportGeosToWkb(), or that the exportGeosToWkb() needs to more resilient against being deleted.
Related issues
Associated revisions
Refine QgsFeature geometry getters/setters
All pointer based methods have been removed.
Now we have only:
void setGeometry( const QgsGeometry& geom )
and
QgsGeometry geometry() const
Benefits include avoiding a whole lot of tricky pointer lifetime
issues, potential memory leaks, and finally closing #777, which
has survived for over 9 years!...
Impacts on PyQGIS code:
- no more need for the messy
g = QgsGeometry( feature.geometry() )
workaround, just use g = feature.geometry() instead
- IMPORTANT: you can no longer test whether a feature has geometry
using `if f.geometry():`, since QgsFeature::geometry() will
always return an object. Instead, use
`if not f.geometry().isEmpty():`, or preferably the new method
`if not f.hasGeometry():`
Fix #777
History
#1 Updated by Martin Dobias about 17 years ago
The problem here is in python bindings because the scenario seems to be like this:
1. get feature
2. store geometry's reference in Python
3. feature is deleted (together with geometry)
4. reference in Python still exists, but the object it's pointing to doesn't
I'm trying to find out how to cope with this correctly...
Martin
#2 Updated by Jürgen Fischer over 16 years ago
see also #1248
#3 Updated by Giovanni Manghi over 15 years ago
Hi,
what is the status of this issue?
cheers
#4 Updated by Jürgen Fischer about 15 years ago
see also #2173
#5 Updated by Paolo Cavallini over 14 years ago
Still true?
#6 Updated by Martin Dobias over 14 years ago
This haven't been fixed yet
#7 Updated by Giuseppe Sucameli over 14 years ago
No crashes in my Ubuntu 9.04, but I never get the correct results.
I tried wkbType() on the same selected geometry a lot of times and I get different (and also strange) results:
0, 16777216, 7, 187101, 92, 143587, ...
#8 Updated by Giovanni Manghi about 13 years ago
- Target version changed from Version 1.7.0 to Version 1.7.4
#9 Updated by Giovanni Manghi almost 13 years ago
- Pull Request or Patch supplied set to No
- Crashes QGIS or corrupts data set to Yes
- Affected QGIS version set to master
#10 Updated by Paolo Cavallini over 12 years ago
- Target version changed from Version 1.7.4 to Version 1.8.0
#11 Updated by Paolo Cavallini over 12 years ago
- Target version changed from Version 1.8.0 to Version 2.0.0
#12 Updated by Giovanni Manghi almost 12 years ago
- Priority changed from Low to High
#13 Updated by Matthias Kuhn almost 12 years ago
Fix in pull request #436
Not sure if it's the best way to do it.
Maybe a reference counter or the like could also help to overcome this problem. But I'm not sure how easy it is to do mixed ref-counting between python and C++?
#14 Updated by Paolo Cavallini almost 11 years ago
- Target version changed from Version 2.0.0 to Future Release - High Priority
#15 Updated by Jürgen Fischer over 8 years ago
- Assignee deleted (
Martin Dobias)
#16 Updated by Nyall Dawson over 8 years ago
- Status changed from Open to Closed
Fixed in changeset bd7d913379b68a8104608b1afab4d380e4edc26b.