[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Ecrit] Profiles (was: Consensus Call) - namespaces

This concrete example is helpful.

Since applications such as LoST don't typically do schema validation, maybe it's more helpful to look at what an application would actually do, such as using SAX and DOM.

Clearly, the application only has a limited repertoire of XML tags that it recognizes and then parses. It will presumably build up its internal data structure from those elements.

There are two cases:

(1) Senders can only generate certain orderings of elements, such as Point followed by civicAddress, even though the schema allows any ordering. A 'profile' specifies this.

(2) Any ordering permitted by the schema can appear in the XML document.

Doing (1) simplifies processing, since there are fewer elements that can occur at any given time, but with an event-based or DOM-based model, it doesn't seem to make a whole lot of difference.

We obviously do want to restrict the GML elements that can occur in the XML sent, but that's a standardization issue, not a labeling issue. Thus, I'm all in favor of defining profiles for the sender.

I just don't see how a profile label (beyond the namespace indications) would simplify the processing or improve error processing.

Henning

On Apr 26, 2007, at 11:54 PM, Andrew Newton wrote:

Perhaps we do need to restart the discussion. I was never arguing for anything as strong as XML validity. That was one of your arguments. However, if you take a schema that has the following definition (pasted from 4119):

    <xs:sequence>
      <xs:any namespace="##other" processContents="lax" minOccurs="0"
         maxOccurs="unbounded"/>
    </xs:sequence>

Now take a look at the following from draft-ietf-geopriv-pdif-lo- profile:

         <gp:geopriv>
           <gp:location-info>
             <gml:Point srsName="urn:ogc:def:crs:EPSG::4326"
                <gml:pos>-43.5723 153.21760</gml:pos>
             </gml:Point>
             <cl:civicAddress>
               <cl:FLR>2</cl:FLR>
             </cl:civicAddress>
           </gp:location-info>
           <gp:usage-rules/>
         </gp:geopriv>

What is there to say that <cl:civicAddress> can legally follow <gml:Point>? Nothing. From an XML processors perspective, it could be GML followed by some other stuff that can be ignored (look at processContents="lax"). What's the cue to the application that it shouldn't ignore <cl:civicAddress>? How does it know that is important information, if even it may not have the schema in its catalog?

Either LoST knows about a civicAddress or it doesn't. Just having the schema in the catalog doesn't make a difference, as it has to understand the semantics, e.g., to do matching and caching.




-andy

On Apr 26, 2007, at 11:26 PM, Henning Schulzrinne wrote: 


_______________________________________________
Ecrit mailing list
Ecrit at ietf.org
https://www1.ietf.org/mailman/listinfo/ecrit