[raster] ST_Intersects doc and behaviour are out of synch

[pracine]

The doc state that "If no band is specified band 1 is assumed" for ST_Intersects(geometry geommin, raster rast, integer nband=NULL) but nband default to null which make the operator to ignore nodata values.

I guess the desired behaviour is to default to 1.

[dustymugs]

When thinking about the correct behavior of not specifying nband, I have to ask myself...

If a user doesn't provide a value for nband when nband is a possible parameter, what area should be considered? The area covered by the raster? Or the area covered by band 1 of the raster?

Personally, I'd lead towards the area covered by the raster. Since the user didn't provide an explicit value for nband, the response should not be explicit.

I'm for changing the docs to correctly reflect the behavior.

[pracine]

+1

[dustymugs]

Fixed in r9936 for 2.0. Fixed in r9946 for trunk.