Re: [egov] Tactics on data alignment / standardization

["Carl Reed" <creed@opengis.org>]

David -

Yes, initially getting into and understanding GML can be quite daunting!
Honestly, some our members question the need for such completeness. Why not
some simple schema for just point, line, polygon coordinate sets with
attributes? Well, this is exactly where we started with GML about 4 years
ago. Then, we worked with simple, 2d geometry. The model for geometry we
used is grounded in the ISO TC 211 document 19107 - Spatial Schema. We did
this in order to insure agreement and member consensus  on how to express
geometry. This was relatively straight forward, but still stressed XML at
that time :-) But geometry is only part of the equation. Without some
context, then the geometry is basically worthless to a consumer (client).
You must have at a minimum some metadata and a coordinate reference system
(CRS). And geometry itself is OK if you only want to display raw point,
lines, polygons etc on a display device. Most applications need attributes -
and so now one needs to be able to express "features" in which geometry is
only a property. So now we have a much richer contextual environment (as
Eliot C. points out in a later email) that meets the needs of a certain
level of market requirement for spatial data transport interoperability.

But wait, we do not live in a 2d world. What about 3d? And we need to make
sure we properly express units of measurement. And what about expressing
topology? Without topology, GML would be useless to most of the worlds major
mapping organizations (NIMA, CENSUS, USGS, Ordnance Survey). And the list of
requirements goes on. It sure would be nice if spatial data were as simple
as say financial data!

So the result of 4 years of very hard work by OGC members is the current
version of GML (version 3) - a specification of over 500 pages.

So, GML can appear daunting. However, there is an increasing number of aids
available for learning and using GML. For example, GML Dev Days is going on
in Vancouver this week. Training seminars and so forth. Tutorials are being
written. A new book on GML with hundreds of examples is soon to be released.

And, at the end of the day, one can actually write a very simple GML
application schema for the point, line, polygon case - just like when we
started GML :-) And the members are very aware that GML appears complex and
continue to work toward a healthy balance of functional
capability,contextual richness and simplicity.

As one of our members who has implemented a generic engine for "consuming"
any GML application schema says:

"One of the most powerful aspects of GML is the freedom it gives users to
define their own custom application schemas. While this capability provides
extraordinary flexibility to data modelers, it also creates significant
challenges . . ."

Sorry for the lengthy discourse, but, as with GML, context is important!

Regards

Carl



----- Original Message -----
From: David RR Webber - XML ebusiness <Gnosis_@compuserve.com>
To: Carl Reed <creed@opengis.org>
Cc: <egov@lists.oasis-open.org>; Thomas Lee <ytlee@cecid.hku.hk>
Sent: Wednesday, July 23, 2003 7:46 PM
Subject: Re: [egov] Tactics on data alignment / standardization


> Carl,
>
> On quick inspection you deserve an award for the most complex
> XML application I have seen, and also for using the most
> complex parts of the W3C specifications.
>
> You guys certainly do not let fear hold you back!
>
> DW.
> ====================================================
> Message text written by "Carl Reed"
> >
> Thomas -
>
> You might be interest to learn that the UK Ordnance Survey now provides
> their Land Line (vector) data as a GML (Geography Markup Language)
> encoding.
> GML is an adopted OpenGIS Specification. GML is based on the
> eXtensibleMarkup Language (XML), is an open and non-proprietary
> specification used for the transport and storage of geographic
information.
>
> Kind regards
>
> Carl Reed
> Executive Director
> OpenGIS Consortium
> <
>