[QGIS-Developer] pyqgis QgsGeometrySnapper does not snap to a source created inside a function

[Germán Carrillo]

*> The underlying C++ code expects a pointer to the reference layer. If the
python wrapper just passes the pointer of the object that was created in
the call, the object will be out of scope after the function call returns
and will be garbage collected. Someone who actually knows the code details
can verify or contradict this.*

This makes sense to me.
In fact, the original code makes QGIS crash (using v3.28.6) with a
Segmentation Fault, indicating that the underlying C++ object has been
deleted.
That's why you should first create a Python variable to make it clear that
the layer object should persist in the outer scope.

Regards,

Germán


El vie, 20 oct 2023 a las 5:30, David Strip (<qgis-user at stripfamily.net>)
escribió:

> On 10/19/2023 10:32 AM, Gabriel De Luca via QGIS-Developer wrote:
>
> Hi German,
>
>   Thanks. Yes, it works that way. Do you know why?
>
>
> Regards,
> Gabriel
>
>
> Since nobody who really knows the code has responded yet, I'll conjecture
> it's a scoping issue. The underlying C++ code expects a pointer to the
> reference layer. If the python wrapper just passes the pointer of the
> object that was created in the call, the object will be out of scope after
> the function call returns and will be garbage collected. Someone who
> actually knows the code details can verify or contradict this.
>


-- 
-----------
   |\__
(:>__)(
   |/
Soluciones Geoinformáticas Libres
http://geotux.tuxfamily.org/
https://twitter.com/GeoTux2 <http://twitter.com/GeoTux2>

<http://gis.stackexchange.com/users/4972/germ%c3%a1n-carrillo>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/qgis-developer/attachments/20231020/6733d6ee/attachment.htm>