[postgis-devel] Should EMPTY be spatially equal to self ?

['Sandro Santilli']

On Sun, Jan 15, 2012 at 04:32:15PM -0500, Paragon Corporation wrote:

> I'm not talking about that.  I'm talking about tools that have to support
> 1.3, 1.4, 1.5, 2.0.0 (if you use 1.3 it means goemetric quality, if you use
> 1.4-1.5 it means
> geometric equality or bounding box equality, if you use 2.0.0 it may or may
> not use an index (and right now since we 
> don't have any tagged releases -- we can't even tell people how they need to
> upgrade).
> 
> You get my frustration?

Yes, but really the only problem I see is 1.3.
The cases in which an index is not used to me are to be considered bugs.

We have 1.4 and 1.5 branches that we are maintaining so I don't really
care about 1.3. IMHO the best thing to do is ~= means bbox equality
matching back to 2 minor releases (1.4 and 1.5) and now finally using
an index.

Trying to support all deploys done from development snapshot will always
give you high frustration so I suggest you stop caring about those.

> I meant to say EXCLUSION constraints -- and yes ticketed at:
> http://trac.osgeo.org/postgis/ticket/1471
> 
> Note we probably could easily squeeze this in 2.0, but I don't care cause I
> have no need for it at the moment, but I know several on the list
> have expressed interest mostly for point data which ~= doesn't quite work
> for because of the float / double issue you can get points really close that
> are not the same that would be flagged wrong.

The float/double issue is exactly the same you get with all the other operators.
Get over it. Boxes are possibly slightly larger than the geometry they contain.

--strk; 

  ()   Free GIS & Flash consultant/developer
  /\   http://strk.keybit.net/services.html