[OpenLayers-Dev] WMSManager beta version

["RIVIERE Robert - CETE Méditerr./DI/ETER"]

Hi Lorenzo,

I'm very busy this week so I'm sorry I won't be able to participate to a real discussion on the subject.

> I'm actually studying CSW2 and hope that Ticheler will have time for me 
> next week to choose the simplest syntax from KVP encoded request to ask 
> for a set of WMS servers (or layers) and which minimum set of metadata 
> make sense to respond from a CSW2 point of view.

That's funny, that's exactly what I'm doing these days for our cataloguing project : I'm working with a bigger one, the "french national catalogue", www.geocatalogue.fr.

Basically, our catalogue will be a community one (french ministery of transports) and we'll send general information to the geocatalogue.
We've decided to do so with CSW2. You probably already know that CSW2 has to be implemented in a "profile". There are two competing profiles, ebRIM and ISO.
For our exchanges, we have decided to use the ISO Profile the same way as you're thinking about, i.e. the simplest KVP syntax for a limited set of use cases.
I think it's easier for us than for you because we already have XML 19139 metadata generating/parsing mechanisms and the problem is just about transporting these metadata.

Here are some things I've learned when  doing this :
* You'll have to implement at least to operators :
- GetCapabilities (the CSW one, not the WMS one. You'll want to have metadata about the CSW server you're talking with)
- Get Records to ask for specific metadata from the catalog

* GetRecords: in ISO Profile, it has 19 parameters. Some of them can be ignored, but if you want a real query function you'll have to implement the CONSTRAINT attribute, expressed in CQL or OGCFilter.

* The answer will be an ISO 19115 metadata (at least in ISO profile). As you're requesting for web services, there probably should be some 19119 parts in it.
There is a big question on how to map WMS capabilities to 19115+19119 semantics. I really don't know.

* The metadata will be in 19139 XML. Their parsing can be pretty hard (it's designed to be extensible, polymorphic, etc...). 

My personal advice is that all this will be quite complex. I'd be glad to hear that someone has found a simpler way to be CSW2 compliant (geonetwork guys or someone else ?).

Ciao

Robert