[postgis-users] RE: [postgis-devel] GML and XY,XYM,XYZ,XYZM

[Martin Daly]

strk,

> I've introduced support for XYM.
> How should XYM geometries be rendered in GML ?
> And what aboux XYZM ?

Excellent questions, with no easy answers.

GML 2.x <coord> elements are hard-wired to have elements <X>, <Y> and
<Z>, so that is a non-starter.  GML 2.x <coordinates> elements just
contains tuples of co-ordinates, so that is a possibility.  I very much
doubt that any existing GML 2.x parsers have any knowledge of linear
reference systems, and instead assume that the tuples have at most three
ordinates, and that they are X, Y and  optionally Z.  Other usage is
valid, but is likely to be restricted to a particular vendor, or
industry, or whatever, and is not likely to be particularly
interoperable.

GML 3.x has more support for specifying the CRS of the tuples, including
linear reference systems, but using that will reduce the number of
available clients for the data, because GML 3 support is not widespread.

Not only no easy answers, but no answers at all!  Would it be tolerable
to not support AsGML for XYM and XYZM, or to strip the M?

Regards,
Martin