[gdal-dev] GRASS gdal and ogr plugins

[Tom Russo]

On Fri, Aug 22, 2014 at 07:17:15PM +0200, we recorded a bogon-computron collision of the <even.rouault at spatialys.com> flavor, containing:
> Tom,
> 
> Should be fixed by r27597. This was a side effect of the work done for 
> http://trac.osgeo.org/gdal/wiki/rfc46_gdal_ogr_unification

Indeed, it is fixed by r27597.  So it *was* just a matter of changing the name. 
Thank you.

> > I've just svn updated gdal to r27596 and rebuilt it, grass, and the grass
> > plugin and installed them.  I had not previously updated gdal since about
> > May 14.
> > 
> > Now gdalinfo and ogrinfo  are reporting:
> > > gdalinfo --formats
> > 
> > ERROR 1: A non OGR driver is registered with the same name: GRASS
> > Supported Formats:
> >   GRASS -raster- (ro): GRASS Database Rasters (5.7+)
> > ...
> > 
> > and
> > 
> > > ogrinfo --formats
> > 
> > ERROR 1: A non OGR driver is registered with the same name: GRASS
> > Supported Formats:
> >   PCIDSK -raster,vector- (rw+v): PCIDSK Database File
> > ...
> > 
> > GRASS does not appear here, so clearly the registration of the OGR grass
> > plugin is not happening.  The code does not exit abnormally, just prints
> > the error message and clearly doesn't have the OGR grass plugin loaded.
> > 
> > The same happens from applications that are built against the gdal library.
> > Normal operation, just no access to GRASS data through OGR.
> > 
> > Looking at ogr/ogrsf_frmts/generic/ogrsfdriverregistrar.cpp, I'm wondering
> > if that's because both the gdal grass driver and ogr driver are returning
> > "GRASS" as their name, and the "GRASS" provided by gdal is confusing the
> > registrar?  Should the OGR driver maybe have a different name?  Or am I
> > completely off the mark?
> > 
> > I do note that the change in ogrsfdriverregistrar.cpp that introduced the
> > error message was dated 24 May 2014, just 10 days after I had last built
> > gdal on this machine.  The previous build would report "GRASS" as an
> > available format from OGR.
> > 
> > I'm using grass off of the "develbranch_6" branch, updated right before I
> > built the gdal plugin.
> > 
> > I build the gdal plugin so:
> >   build and install gdal without GRASS support
> >   build GRASS and install it
> >   build the plugin:
> >   cd /.../gdal/frmts/grass/pkg
> >   ln -s ../grass57dataset.cpp .
> >   ln -s ../../../ogr/ogrsf_frmts/grass/ogrgrassdriver.cpp .
> >   ln -s ../../../ogr/ogrsf_frmts/grass/ogrgrassdatasource.cpp .
> >   ln -s ../../../ogr/ogrsf_frmts/grass/ogrgrasslayer.cpp .
> >   ln -s ../../../ogr/ogrsf_frmts/grass/ogrgrass.h .
> >   ./configure --with-gdal=/usr/local/bin/gdal-config
> > --with-grass=/usr/local/grass-6.5.svn
> > --with-autoload=/usr/local/lib/gdalplugins gmake
> >   sudo gmake install
> > 
> > The result is gdal_GRASS.so and ogr_GRASS.so in /usr/local/lib/gdalplugins.
> > 
> > Am I simply building the plugin wrong, or is this a bug?
> 
> -- 
> Spatialys - Geospatial professional services
> http://www.spatialys.com

-- 
Tom Russo    KM5VY   SAR502   DM64ux          http://www.swcp.com/~russo/
Tijeras, NM  QRPL#1592 K2#398  SOC#236        http://kevan.org/brain.cgi?DDTNM
 echo "prpv_a'rfg_cnf_har_cvcr" | sed -e 's/_/ /g' | tr [a-m][n-z] [n-z][a-m]