Hi Jukka and Even After all, I agree with Even that OGR behaves correctly in this case. And I think "-nlt guess" is a good proposal. I'm also aware what Jukka says that mixed geometries are allowd. But I always wondered which current implementation can cope with this - except Spatialite/SQLite (which is a (nice) bucket anyway).
Yours, Stefan 2013/4/22 Jukka Rahkonen <jukka.rahko...@mmmtike.fi> > Even Rouault <even.rouault <at> mines-paris.org> writes: > > > What could be done is to have a "-nlt guess" mode in ogr2ogr, that, in > case > > the source layer has an unknown geometry type, would retrieve the first > feature > > of the source layer to find the geometry type and create the target layer > > geometry type. > > I guess that it would work for most cases. However, I know that in real > life > the geometry of the first feature can be empty. It is also possible that > WFS > feature type really contains mixed geometries. For handling them > automatically I would perhaps use Spatialite as an interim target and sort > the features by geometry types before moving them further. > > -Jukka Rahkonen- > > _______________________________________________ > gdal-dev mailing list > gdal-dev@lists.osgeo.org > http://lists.osgeo.org/mailman/listinfo/gdal-dev >
_______________________________________________ gdal-dev mailing list gdal-dev@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/gdal-dev
