[gdal-dev] Hidden regression in GDAL 1.11.0 - gdal_rasterize with postgis silently fails to render many polygons.

[Graeme B. Bell]

Excellent work Even.

Thank you very much for your kind and speedy help with this! :-)

Hoping to see it in 1.11.2, as we prefer to take RPMS from the main repos for our builds here rather than rolling our own. 

For the local workaround I have removed the tables flag for the moment and will accept the speed penalty. 

Graeme. 

On 09 Feb 2015, at 19:23, Even Rouault <even.rouault at spatialys.com> wrote:

> Graeme,
> 
> I confirm. This is the result of a subtile interaction of tables= , SQL request 
> and SRS resolving. The PG driver has become quite subtile...
> Traced in http://trac.osgeo.org/gdal/ticket/5837 and now fixed.
> 
> As a workaround you could use a view that does the the "select * from 
> world_diamonds where st_intersects(wkb_geometry, 
> ST_GeometryFromText('POLYGON((-180 -90,-180 90,180 90,180 -90,-180 
> -90))',4326))" and use the view name with the -l parameter of gdal_rasterize 
> (he bug doesn't occur if using a straight layer name)
> 
> gdal_rasterize could/should be improved as well to set a spatial filter 
> automatically. That way you could just use the layer name.
> 
> Anyway I guess this will be a good reason to issue a 1.11.2RC2 with the fix.
> 
> Even