[gdal-dev] troubles with OVERRIDE_PROJ_DATUM_WITH_TOWGS84

[Even Rouault]

Le mardi 22 mai 2012 21:01:26, Margherita Di Leo a écrit :
> Hi folks,
> 
> looks like this bug ticket [1] finds its roots into the gdal code. From a
> discussion [2] in the gfoss.it [3] ML (the italian OSGeo local chapter)
> emerged that gdal 1.8.0 introduced a further option named
> OVERRIDE_PROJ_DATUM_WITH_TOWGS84, allowing to replace the +datum definition
> with a corresponding +towgs84.
> Such option is not supposed to be the default behavior when generating the
> standard EPSG file for Proj4:
> 
> epsg_tr.py --config OVERRIDE_PROJ_DATUM_WITH_TOWGS84 FALSE -proj4 -skip
> -list gcs.csv > epsg
> 
> Anyway, from direct testing emerged that setting the above mentioned option
> to YES, NO, TRUE or FALSE has absolutely no effect at all; the +datum
> definitions are unconditionally overwritten by +towgs84 in all cases. And,
> accordindgly to this, the EPSG file distributed by the most recent Proj4
> 4.8.0 introduces many unexpected changes (e.g. affecting the Italian Rome
> 1940 SRS), thus causing many troubles due to this abrupt compatibility
> break.

 --config OVERRIDE_PROJ_DATUM_WITH_TOWGS84 FALSE definitely works. I've just 
verified it, and I see that the epsg file delivered with proj 4.8.0 takes into 
account http://trac.osgeo.org/proj/changeset/2172 . The effect of that is just 
to avoid expanding +datum definitions.

Your issue with EPSG:3003 is of a different kind, and related to another change 
where the new logic now peaks up the EPSG preferred transformation whereas it 
didn't before. I think this is related to 
http://trac.osgeo.org/geotiff/changeset/1819/ and 
http://trac.osgeo.org/geotiff/changeset/1829/

> 
> Could anyone please verify that, and, in case, I'll open a ticket.
> Thanks!
> 
> madi
> 
> 
> [1] http://trac.osgeo.org/proj/ticket/122
> [2] http://lists.gfoss.it/pipermail/gfoss/2012-May/023071.html
> [3] www.gfoss.it