[udig-devel] Re: Can uDig plot PostGIS data from an SQL view?

[Cory Horner]

Gary Lucas wrote:

>I hadn't thought of this before my last email, but later it
>occured to me to insert an entry into the geometry_columns table
>so that it matched the record from the original table on which the
>view is based (with the name of the view appearing in
>the "f_table_name" column).  Still nothing
>plotted.  It did make a difference in that when I ran uDig and 
>brought up the "Resource Summary" it now shows the 
>correct value for the Coordinate Reference System (CRS) which,
>in this case, is just EPSG:WGS84 (SRID=4326). On the other hand, the 
>values in the Bounds entry are rather suspicious being
>(-494.9, -79.3), (483.1,144.3)    They seem to change everytime
>I run the program.
>  
>
I suspect that the GeoTools PostGIS implementation currently does not 
know to treat views differently from tables.  The estimate_extents 
function likely fails on the view, and an approximation is made based on 
a handful of features (if it fails it should check if it is a view and 
ask the table -- if available -- what its bounds are).  We should 
investigate further to ensure that indexes are properly exploited.

Your view is the only layer in your map?  Also, you may need to remove 
your PostGIS database from the uDig catalog and re-import it, as the 
bounds and state will be cached.

>Could there be another table beside geometry_columns that
>needs an additional entry?
>
>Also, I didn't name my geometry column "the_geom" which seems
>to be a prefered default in some of the GeoTools code.
>  
>
The entry in the geometry_columns table for your view should be 
identical to the one for the original table except for the table name.

If you don't have any luck, could you send me your table schema and 
geometry columns entries so I might see if there is a case we aren't 
handling correctly?

Thanks,
Cory.