Re: Possibly defaulted (was RE: Minutes for Thurs 7th Feb 2002 Telcon)

On Wed, 13 Feb 2002 15:07:21 +0100 (CET), in soap you wrote:

> Simon, 
> In our opinion this is consistent with the handling of 
>mustUnderstand and actor/role attributes.
> Back to itemType: an implementation IMO violates nothing if it
>does not include/process the itemType attribute. Do you agree?

yes

> On the other hand, if an implementation does not use a WSDL (or
>other) description of messages, the information coming in the
>itemType attribute may be very handy. I think the worth of
>itemType is the same as the worth of the size in the "leftmost"  
>dimension of the array, which too can be discovered by inspecting
>the whole array. Do you agree with this?

yes

> If you agree with both my statements, I don't see why you would 
>want the removal of itemType attribute. 

My gut feeling is that making it optional is fine in theory, but in
practice will just lead to complications [after all, a similar
situation gave us the position today, where a large number of section
5 toolkits put xsi:type's on everything]

Now that sparse & PT arrays are gone, the conformance is just a
performance optimization [at least for single dimension arrays] for
consistency should that be optional too ? (i.e. it defaults to *)

Cheers
Simon

>On Tue, 12 Feb 2002, Simon Fell wrote:
>
> > Wasn't there a recent discussion on issues to do with defaulting
> > mustUnderstand & actor [i need to check the archives], is this
> > consistent with the results of that ?
> > on a related note, IMHO this makes the case for itemType weaker, I'm
> > still of the opinion that itemType is redundant, and results in
> > array's being a special case that force you to include type
> > information on the wire.
> > 
> > Cheers
> > Simon
> > www.pocketsoap.com
> > 
> > On Tue, 12 Feb 2002 11:34:43 +0100 (CET), in soap you wrote:
> > 
> > > Simon, (and maybe Andrew 8-) )
> > > we agree that the words "(possibly defaulted)" in rule 2 in 
> > >Encoding are offensive. 8-)
> > > The Encoding task force suggests that we resolve this editorial
> > >issue by removing these offending parenthesized words. The
> > >situation would become equal to that with the mustUnderstand
> > >attribute - effectively it has the default value of "false", even
> > >though this default value would not show in the infoset that the
> > >SOAP Node receives; the node must act as if the value was there
> > >as "false".
> > > Same here, if we're in an array and there is no itemType 
> > >attribute present, the Encoding processor must act as if it were 
> > >present with the value {xml-schema-namespace}anyType.
> > > Is this satisfactory?
> > >
> > >                   Jacek Kopecky
> > >
> > >                   Senior Architect, Systinet (formerly Idoox)
> > >                   http://www.systinet.com/
> > >
> > >
> > >
> > >On Fri, 8 Feb 2002, Andrew Layman wrote:
> > >
> > > > Re
> > > > 
> > > > 6) Encoding use of default attributes, see item 5 in [3]
> > > > Agreed that text in rule 2 is confusing.
> > > > NEW ACTION: MJH to remove "(possibly defaulted)" from rule 2. NEW
> > > > ACTION: JK to contact originator with proposed resolution.
> > > > 
> > > > I might be that originator.  :-)
> > > > 
> > > > 
> >

Received on Wednesday, 13 February 2002 12:07:19 UTC