[postgis-users] Migrating PostGis: pointn, geometryn functions

[strk at refractions.net]

Guido, the 1-indexing requirement is from 
OGC Simple Feature Specification for SQL

--strk;

On Fri, Aug 26, 2005 at 02:09:16PM +0200, Guido Lemoine wrote:
> Hi all PostGisers,
> 
> Being new to the list, let me start by saying that you Refractions guys 
> and girls
> are all doing a great service to the spatial community...
> 
> I've decided to file a few observations that you may want to have a 
> closer look
> at.
> 
> We recently migrated to postgresql 8 and postgis 1.0.3. Most of our
> application code is embedded in Java and some in PLPGSQL functions.
> 
> During migration, we ran into some "backward compatibility" problems.
> 
> One is about the postgis functions that index multi-dimensional 
> structures, which
> now start at index 1 instead of 0. For example, whereas pre-1.0.x 
> postgis functions
> pointn(geometry, 0) used to return the first point of the geometry (a 
> linestring, or
> polygon, etc.),  this now returns NULL and needs to be replaced by 
> pointn(geometry, 1).
> Same applies for geometryn(geom, 0/1) for a MULTIgeometry.
> 
> I found some obscure reference to the intention to move to 1-based indexing
> from 2003, but nothing in the release notes. The move is apparently 
> conform some
> (OpenGIS?) standard (probably defined by a geographer rather than a 
> programmer).
> 
> I tried to find some clues to this in the this mail-list, but that's 
> hard, because
> I can't search it for keyword[s]. Maybe it's worth putting it somewhere 
> in the
> list of upgrades.
> 
> I have another, more annoying one, which I will detail in my next message,
> 
> Otherwise, keep up the good work,
> 
> Guido Lemoine
> 
> _______________________________________________
> postgis-users mailing list
> postgis-users at postgis.refractions.net
> http://postgis.refractions.net/mailman/listinfo/postgis-users