[Qgis-developer] 2.10 segfaults on startup

[Even Rouault]

Le mardi 30 juin 2015 13:40:33, Radim Blazek a écrit :
> On Tue, Jun 30, 2015 at 1:22 PM, Even Rouault
> 
> <even.rouault at spatialys.com> wrote:
> > Le mardi 30 juin 2015 13:17:50, Paolo Cavallini a écrit :
> >> Il 30/06/2015 11:15, Richard Duivenvoorde ha scritto:
> >> > Yep, confirmed here on a fresh Debian self compiled version.
> > 
> > Might be related to the issue discussed at
> > http://lists.osgeo.org/pipermail/gdal-dev/2015-May/041796.html
> > and
> > http://lists.osgeo.org/pipermail/qgis-developer/2015-May/037820.html
> 
> Thanks Even,
> 
> we can probably get around this using mutex in QgsGdalLayerItem if
> Spatialite is < 4.2.
> Can we get somehow SPATIALITE_412_OR_LATER or bSpatialiteGlobalLoaded
> from GDAL API or its headers  ?

Radim,

I can't think a way of getting those internal details from the outside.

However if the (external) spatialite version available when compiling QGIS is 
< 4.2, then it is very likely that the one used when compiling GDAL is also < 
4.2

The configure test used by GDAL is:
AC_CHECK_LIB(spatialite,spatialite_target_cpu,SPATIALITE_412_OR_LATER=yes,SPATIALITE_412_OR_LATER=no)

You could also fallback to runtime checks with SELECT spatialite_version()

This assumes a setup where not too odd things happen, i.e. compiling with a 
version and updating it between compilation of various software components.

Even

-- 
Spatialys - Geospatial professional services
http://www.spatialys.com