[postgis-users] Vacuum analyse ERROR: Operation on two GEOMETRIESwith different SRIDs

[strk at refractions.net]

On Wed, Jun 15, 2005 at 11:09:10AM +0100, Mark Cave-Ayland wrote:
> 
> > -----Original Message-----
> > From: Sergey Karin [mailto:sergey.karin at gmail.com] 
> > Sent: 15 June 2005 10:50
> > To: Mark Cave-Ayland (External)
> > Subject: Re: [postgis-users] Vacuum analyse ERROR: Operation 
> > on two GEOMETRIESwith different SRIDs

(cut)

> > INFO:  analyzing "public.workspace"
> > INFO:  "workspace": 1 pages, 15 rows sampled, 15 estimated total rows
> > ERROR:  Operation on two GEOMETRIES with different SRIDs

I've verified this myself.
Yes, it's btree ops involved. Even in 1.x series.
analyze verbose can show the corrupted table, but question is:
should a table containing mixed-SRID geometries be really considered
corrupted ?

The btree ops are really just place-holders to allow DISTINCT
and ORDER BY operations, we could change their
semantics to include SRID in their checks (compare SRID number,
upper-left X and Y values, lower-right X and Y values).

What do you think ?

--strk;