Re: why no doc type declaration and PIs in SOAP?

I think that the status quo (a SOAP message MUST NOT have a DTD
or PIs) should be preserved. The issue is what to do about
them if a (non-compliant) SOAP client sends a message that
includes either.

As I stated at the f2f, I think that they should be ignored.
We shouldn't be imposing that a SOAP server MUST check for
these and return a Fault as this adds unnecessary complexity,
especially for PIs. 

What we could say is that a SOAP Receiver SHOULD ignore a DTD
or PI in a SOAP message and that they MAY send a Fault (which should
be defined should we go down this path). 

A SOAP implementation that can not handle DTDs can exclude
them relatively easily. Whether it chooses to send a fault would
be implementation dependent.

Cheers,

Chris

Marc Hadley wrote:
> 
> Jacek Kopecky wrote:
> >
> >  I couldn't find it in archives (because the search engine
> > returned nothing at all), so I'll ask:
> >  What are the reasons for disallowing document type declaration
> > and processing instructions in SOAP? Will we keep the
> > restrictions in SOAP version 1.2? As it is now, SOAP grammar is a
> > subset of XML.
> >
> We discussed this at the recent F2F in relation to issue 4 which raises
> the question of what a receiver should do on receipt of a message
> containing a PI or DTD. I have an pending action to re-raise this issue
> - consider it raised !
> 
> There seems to be two opinions on the subject of DTDs and PIs:
> 
> (i) Allowing them increases complexity and doesn't bring any particular
> benefit, the only compelling argument for allowing PIs was so that a
> stylesheet could be associated with a message for human viewing.
> 
> (ii) Adding them doesn't add much to the complexity, they are part of
> XML so we should allow them.
> 
> My original suggestion for resolution of issue 4 was to retain their
> current status (i.e. not allowed) and add text requiring a SOAP
> processor to generate a fault when a message containing one was received.
> 
> Others felt that if present they should be ignored but this might prove
> difficult in the case of DTDs with current parsers.
> 
> Discussion ?
> 
> Regards,
> Marc.
> 
> --
> Marc Hadley <marc.hadley@sun.com>
> XML Technology Centre, Sun Microsystems.

Received on Thursday, 20 September 2001 14:16:16 UTC