[Geowanking] GML Simple Features instance documents for PostGIS
norman.barker at gmail.com
Tue Jan 2 11:12:29 PST 2007
Hi Chris, all,
thanks for the reply, and it was particularly good to hear from
GeoServer which I used to use a lot (I will be moving this discussion
to the PostGIS list when it gets going Java-wise), at the moment I am
interesting in discussing the GML 3 profile, and possible requirements
for a GML loader, any 'stateful' requirements (the main benefit of a
clustered app server) etc.
My comments inline.
On 1/2/07, cholmes at openplans.org <cholmes at openplans.org> wrote:
> On Sun, December 31, 2006 6:41 am, Norman Barker wrote:
> > Hi,
> > (sent again, as it hasn't appeared on the list in the last 24 hrs)
> > I was surprised how easy it was to make castor (http://www.castor.org)
> > bind to the simple features profile gmlsf.xsd document and create java
> > classes (it fails on the main schema gml.xsd).
> > Does anyone have sample document of any of the example schemas at
> > http://schemas.opengis.net/gml/3.1.1/profiles/gmlsfProfile/1.0.0/examples/
> > that they would be prepared to share with me? The hydrography schema
> > looks particularly interesting.
> > I wrote a tutorial on using EJB3 annotations with PostGIS (check out
> > the postgis wiki for more) and the plan is to use castor to load the gmlsf
> > document, and then use the geometry annotation to persist this to postgis
> > (without having to create the tables as this is handled for
> > you). As everything is transactional in the middle tier (or can simply be
> > made that way) I think it would give you a nice and simple postgis WFS-T,
> > though no where near as complete as GeoServer it would get round the GPL
> > issue, and give the acceptance level to management who are happy with
> > EJBs.
> Note we are contemplating going to change at least the core of GeoServer
> to LGPL, and I would be very interested to hear thoughts from anyone about
> license concerns, issues, hopes and dreams, ect.
I think GPL is just seen as bad for commercial organisations (speaking
from my own viewpoint) certainly for projects this year I have had to
reject GeoServer, GeoNetwork solely on its license, and have looked to
code up by hand instead. This is more a legacy issue really, I don't
have problem giving code away, but it is a larger wall for others. I
find the PostGIS license GPL, yet JDBC driver being LGPL an
interesting one as well.
Perhaps for everyone you could summarise the roadmap to LGPL, I spoke
to the GeoServer group at the last OGC meet, and keep in regular
contact with Simone G, and was to understand that if I wrote a plugin
I could keep the code private (though in most cases I wouldn't),
however the plugin architecture licensing issue was a bit woolly.
> As for using something like castor, I think one of the very early
> implementations of GeoServer actually made use of Castor. It didn't work
> out, wasn't flexible enough, but this was also years ago - so the tech's
> probably much better now. And yeah, if you just support one data format
> then it should be much easier to code up.
I also found Castor didn't work well before, but rather than the
sophisticated parser for Geotools I think I need something simpler and
rigid so I can annotate the base classes for persistence.
Certainly the quick test gave good results.
> > Just a thought, have to get Castor to bind to a sample document first,
> > does anyone have any largish real documents they would be prepared to
> > share?
> Not at the moment. But we're finishing up GML 3.1.1 simple feature output
> in GeoServer, so when that's done I can probably generate one for you?
GML 3.1.1 is supposed to encompass the GeoRSS and point profiles
right, perhaps there are some large documents of these around? It
would be nice however to go at least one layer deep in the GML if
Just finding it a little frustrating at the moment, OS Mastermap is
GML 2x, as are the cite tests, yet GML 3 has the more interesting
profiles (GMLJP2 annotations for example). I could generate a GML 3.1
document using XML Spy, but I a real life document would be a great
start (chicken and egg?!!).
any thoughts gratefully received, if anyone has a particular use case
for a J2EE (so clustering, replication, security etc) I would also
would be interested to hear.
> > http://lists.burri.to/mailman/listinfo/geowanking
> > !DSPAM:1003,4597a2c3200722223018498!
> Geowanking mailing list
> Geowanking at lists.burri.to
More information about the Geowanking