RE: Re[2]: Clark's commentary

So how about we do a survey of what people believe is being used or
imminently profileable in an xml 2.0?  I've put my best guesses for Chris,
Tim Bray, Norm, James.  Obviously, correct me where I'm wrong.  And it would
be great if others could add to the list

N = namespaces
F = infoset
B = Xml base
D = XML 1.0 DTDs (include entities and IDs)
E = entities from DTDs
I = IDs (from either DTDs or Schema)
S = Schema
P = PIs in general (includes PIs for stylesheets)
T = PIs for stylesheets (And a replacement for stylesheet PI would remove
need for PIs)
C = XInclude specifically as a replacement for DTD entities

The letter is in scope, (-) is not in scope, and (?) is unsure or I guessed
Chris: NFB-EI--T-
DaveO: NFB--?S-TC
TimBr: NF-----PT-

James: NFB-?--PT?
Total: 5541221351

My list for James is based upon the sentence "His vision of a much-improved
XML 2.0 included adding in XML Namespaces, XML Base, and the Infoset, while
subtracting DTDs and dealing with the problem of character entities."

I'd like to be very clear on my position for schema.  To me, XML 2.0 should
support XML Schema.  But XML 2.0 should have 2 levels, same as XML 1.0 has
dtdless and dtd levels.  So RelaxNG could use well-formed XML 2.0.

There is early consensus on at least adding namespaces, infoset, and having
the functionality of stylesheet PIs.  Base almost makes the cut.  It is
fascinating that there is little support early for XML Schema.  We'll have
to see what the 5 other TAG members say.


> -----Original Message-----
> From: Chris Lilley []
> Sent: Tuesday, January 08, 2002 2:58 PM
> To: David Orchard
> Cc: 'Tim Bray';
> Subject: Re[2]: Clark's commentary
> On Monday, January 07, 2002, 8:02:52 PM, David wrote:
> DO> Here we get into the excellent discussion of what
> features are being used.
> DO> Sorry Norm, but your iotas don't quite match my iotas.
> Which isn't
> DO> surprising though ;-)
> Which was Norms point about the *reason* for no other changes..
> DO> A nice facet of xml (2.0 = 1.0 - DTDs - PIs + namespaces
> + infoset + xml
> DO> base) is that I think it more closely mimics standard
> practice, for example
> DO> SOAP 1.2.
> Well, (as the XML processing workshop clearly demonstrated) it depends
> o *whose* definitio of standard practice, commonly used features, etc.
> DO> This is an excellent example of architectural refactoring
> that often happens
> DO> in software.  SOAP 1.2 had to invent the equivalent of
> XML 2.0 for what it
> DO> needed.  Now it turns out that other people could use the
> same definitions.
> DO> So let's refactor the XML 2.0 stuff into a coherent
> piece, then SOAP WG
> DO> doesn't have to document it/maintain it.  And other specs
> can use it rather
> DO> than copying the verbage from soap 1.2.
> Which, as you observed, gets into the excellent discussion of what
> features are being used. Speaking for what *I* see as 'everyone using'
> it includes some of DTDs (for ID and for entities - not just character
> entities either) - PI (one, for linking stylesheets since that is the
> one and only approved method), namespaces (plus multitudionous hacks
> to pretend that DTDs understand namespaces) and xml base. And XLink.
> Although I do agree that the proper place for infoset is the XML
> specification.
> If PIs went, would there be an XLink arcrole defined for stylesheet
> linkage and the assorted title, media etc attributes (and if so in
> what namespace) or would there be attempt n+1 to resurrect packaging
> rather than hanging everything in a compound document off one,
> priveledged (or should that be burdened) document?
> Bringing this back to the topic of discussion, it does clearly show
> that factoring a problem into small easy pieces moves, rather than
> removes, the effort of figuring out how all the pieces work together.
> It also demonstrates that changes in one piece affect the importance
> or content of other pieces.
> --
>  Chris                  

Received on Tuesday, 8 January 2002 18:53:08 UTC