[Geodata] Fwd: ebrim ICEDS examples

Jo Walsh jo at frot.org
Thu Mar 15 13:34:50 EDT 2007


dear Norman, thanks for cc'ing us on this experiment. 
You are a brave soul and my mind is stretching to try and encompass
the idea of an ebRIM success story... ;)
On Thu, Mar 15, 2007 at 09:59:20AM -0600, Norman Barker wrote:
> really appreciate if anyone could look at my ebXML document for ICEDS
> (attached) and check it makes sense in particular

Well, it validates... it translates the capabilities... and i suppose

<rim:Association id="urn:uiid:B9B32114-D280-11DB-AB4C-A08156D89593"
associationType="urn:x-ogc:specification:csw-ebrim:AssociationType:OperatesOn"
targetObject="urn:uuid:6BAB3C60-D274-11DB-B611-236E56D89593"
sourceObject="urn:x-ogc:specification:csw-ebrim:Service:OGC-CSW:ebRIM:ICEDS"/>

Presumably makes sense to someone who spends all day looking at ebRIM
and has trained their brain to render it human-inspectable.  ;)
To me, that looks like an RDF triple with an identifier. 
And people complain about RDF/XML!
But if you are getting real use case value of out of using this as a
format for machine-machine query then i am interested... 

> I couldn't get GML
> types in for the spatial bbox they wouldn't validate in XMLSpy for the
> value element

Could you unpack this a bit more, what you lose by missing GML types?
I see 
 <rim:Slot name="http://www.opengis.net/gml/Envelope" 
    slotType="gml:EnvelopeType">
               <rim:ValueList>
                    <rim:Value>-30,-35 65,80</rim:Value>
               </rim:ValueList>
 </rim:Slot>
This is the moral equivalent of a georss:box?

> 
> http://ebxmlrr.sourceforge.net/wiki/index.php/Community/ogc
> 
> Must admit that despite my own FUD about ebRIM I am sold, as soon as I
> did a search for LANDSAT in the ebxmlrr query interface saw all of its
> metadata, found the associated service which was classified as a WCS -
> it is everything I dreamed of.  It would be interesting to map the
> GRASS WPS so we can search for datasets, and then processing services
> using standard taxonomies.

Hm, but you already knew what you were looking for, and what you
expected to find as a result, searching a very volume limited registry?
Which standard taxonomies for processing services did you have in mind?

> Am I right in thinking that MapBuilder is a CSW client, could this be
> an ajax map client to ebxmlrr?

Cameron Shorter would know for sure, but I don't think is on this
list, though you must have caught him in your list crossfire ;) 

I see this thread, over a year old, started by geodan who were early
SME supporters of core mapbuilder development, worth asking them:
http://www.arcknowledge.com/gmane.comp.gis.mapbuilder.devel/2005-09/msg00237.html

For instant client gratification you are probably better off looking
at gvSIG or at uDig. But we are still in 'canned demo world' here. 
 
AFAIK CSW2 is partial in subversion, not in the formal release yet,
for GeoNetwork. I saw a quick demo yesterday of a catalogue management
UI based on deegree's CSW component -
http://www.deegree.org/docs/csw/deegree_csw_configuration.html
The XML samples of CSWCapabilities make a lot more sense to me than
ebRIM, of which i don't see signs in this ecology yet. 

I would love to see a 'walkthrough' of how you expect client
interaction with a 'registry service' to behave - essentially a
map-based initial search interface, draw bounding box, refine query
through form interaction, finally haul any services discovered into
the client?

How appropriate is a web client going to be for raw coverage data?
Is this why you subsequently want to direct to processing services?
What is the sense of doing this on a single-client, single-purpose,
direct client-service basis? E.g. I want to be able to republish the
results of the query set i just did, and the trail i made to get
there, for others to reuse, to feed back into making search services
more useful and to lower the overhead on costly processing services.

But Will It Scale? I see a missing middle layer in the architecture,
addressing the same quality of problemspace as Tile Map Service -
single popular services aren't going to want or be able to support the
load... important to get a core of this stuff out into the wild. 

What are you getting from this that a combination of W*S capabilities 
and Dublin Core/GeoRSS based properties describing services, won't
provide you with? 

> Appreciate I am spamming a bit, once I have received all the input on
> this I suggest restricting this to the geodata and ebxmlrr groups, so
> if you are interested please sign up to these.

Apologies if this is a bit rambling, I've been in permaconference
mode, and wanted to offer *some* response because this has really
piqued my interest. My approach has been quite the inverse recently,
trying to boil down a minimal model of what's needed to represent and
collect metadata about online sources and services, then working back
out to the more common XML representations from there. If we are
living with CSW, then i would want to explore alternative profiles
than ebRIM, or ISO19139, to send over it. 

cheers, and do keep us in the loop,


jo
--


More information about the Geodata mailing list