[postgis-devel] RE: WKTRaster news from the Toronto CodeSprint?

[Mateusz Loskot]

Stephen Frost wrote:
> Mateusz Loskot (Mateusz.Loskot at cadcorp.com) wrote:
> > > That would be much nicer than having yet another table in the
> > > public
> > > schema that doesn't really belong.  I hope to do the same for the
> > > geometry_column some day..
> >
> > If you present OGC guys with good reasons, perhaps it will happen one day.
> 
> There appears to be some confusion here.  geometry_columns would become
> a view instead of something which has to be updated through a complex
> set of magic functions that do things they really shouldn't (like add
> columns to tables).


Ah, Got it.

> This would be for PostGIS, and of course we
> wouldn't break the OGC spec requirements.  I don't mind the
> raster_columns table from a specification standpoint, but from an
> implementation perspective (I got the impression you were handling
> both) that setup is just plain horrible.

I agree, thought I'd call it "unfortunate" :-)

> > Anyway, discussions about it belong to the world of Sci-Fi, IMHO.
> 
> This was discussed quite a bit previously on this list.  The only
> possible issue is if there is too much information to be encoded in the
> 32-bit integer that PostgreSQL currently provides for typmod.  Based on
> the prior discussion on this list, it looked like it would be possible
> to do for the geometry_columns data.


OK

--
Mateusz Loskot
Senior Programmer, Cadcorp
http://www.cadcorp.com