exclude raster_columns from being listed in geometry_columns

[robe]

Since raster_columns table has a geometry column, it is appearing automagically in our new spanking geometry_columns view.

It really should be treated as a system table, so we should automagically exclude it.

[pracine]

I remember there was another ticket about that…

[robe]

Yah it was about the populate_geometry_columns message, but that ticket has been closed and was before geometry_columns was a view and I had said I wouldn't deal with the raster issue until geometry_columns became a view.

[strk]

I think it is useful to have the possibility to look at system tables with a GIS client, why making it any harder than it is ?

Do topology tables annoy you as well, btw ?

[robe]

no topology tables do not annoy me. Those actually make sense to be listed in geometry_columns because they really are geometry columns you can query like any other and are real bonafied geometry tables. raster_columns is a view now so there is doubly no point in having it listed.

The main issue with raster_columns why it is different and useless to list is that its a mixed bag of mbrs so there is no consistency in the srid. It will always be unknown. I can't think of a reason why a gis client would even care about this column because if they cared about raster, they'd go straight for the raster_columns system catalog and be displaying the extent.

[strk]

Convinced ! r8889 does it and regress-tests it.