Re: IDREF vs HREF for graph edges in SOAP encoding

 Gudge,
 as I expressed before it is my opinion that you don't require
XML Schema processing if you just specify that every application
built according to the Encoding spec will treat the attributes
with local names id and idref (or whatever, just 'ref' will be
proposed in the rewrite for IDREF) as typed ID and IDREF. You
see, the DTD/Schema equivalent may be built in and therefore no
actual DTD/Schema processing is required.
 It is a problem of the current XML libraries that they require 
DTD/Schema to be able to assign types to attributes and act 
accordingly, but this does not matter for the function of the 
application (it's just the minor inconvenience of being unable to 
call getElementByID() or something).
 Best regards,

                   Jacek Kopecky

                   Senior Architect, Systinet (formerly Idoox)
                   http://www.systinet.com/



On Wed, 16 Jan 2002, Martin Gudgin wrote:

 > 
 > ----- Original Message -----
 > From: "Christopher Ferris" <chris.ferris@sun.com>
 > To: "Martin Gudgin" <marting@develop.com>
 > Cc: "Marc Hadley" <marc.hadley@sun.com>; "XML dist app"
 > <xml-dist-app@w3c.org>
 > Sent: Wednesday, January 16, 2002 2:01 PM
 > Subject: Re: IDREF vs HREF for graph edges in SOAP encoding
 > 
 > 
 > > Only if one wanted to leverage the internal subset,
 > > other than that, you could treat them in the same
 > > manner as href and id.
 > 
 > Sorry, this may be the context I'm missing. Are we saying that we will use
 > attributes with local names of ID and IDREF rather than attributes with type
 > of ID and IDREF? If the former then we don't need DTD/schema processing but
 > at the same time I guess I'm not entirely sure what the difference is
 > between ID/IDREF and id/href. If the latter then surely we need DTD/schema
 > processing to determine which attributes are of type ID/IDREF
 > 
 > > It would certainly be much
 > > more convenient for implementations that did choose
 > > to leverage DTD processing.
 > 
 > This leads me to think we're talking about type rather than local name
 > 
 > > Given that we're talking
 > > about encoding, which leverages XML Schema types, it
 > > is pretty clear to me that we're also imposing schema
 > > processing anyway, no?
 > 
 > My understanding is that our spec specifically states that schema processing
 > MUST NOT be required.
 > 
 > Gudge
 > 
 >

Received on Friday, 18 January 2002 04:13:31 UTC