xml-dev - Re: Arbitrary Infoset boundaries (was Re: Common XML -FinalReviewDraft)

[Rick JELLIFFE <ricko@geotempo.com>]

"Simon St.Laurent" wrote:
> 
> At 09:32 AM 8/3/00 +0800, Rick JELLIFFE wrote:
> >This agrees with XML's premise that it is a resolved document format,
> >rather than a format for documents in progress.
> 
> Is that premise formally stated anywhere?

I doubt it. Probably there is some decision of the WG that comments on
it, in some minutes.  It is a practical necessity: SGML-on-the-Web has
to be resolved SGML. 

It is the reason why three things were removed from XML: SDATA entities
was one, IGNORE/INCLUDE marked sections in instances was another, and
just I forgot the third before typing it.  (That DTDs are optional also
shows that strongly-typed read-write considerations did not have
primacy.)

XML 1.0 gives syntax not explaination: it handle almost all explaination
by defering to SGML (ISO 8879). As the first paragraph of XML 1.0 says, 
"XML is an application profile or restricted form of SGML, the Standard
Generalized Markup Language [ISO 8879]. By construction, XML documents
          are conforming SGML documents."

So, apart from going through archives or talking to people, the premises
of XML can be found by figuring out what XML substracted from SGML.  I
leave this as an exercise for the reader.  XML substracted features
needed for unresolved documents, because they were not what XML was
about (at the time).  (XML was initially a delivery mechanism for data
kept or maintained in SGML'86 or other structured formats, in the minds
of many I think. )  

Since XML was attempting to shield people from the complexity/richness
of SGML, there is no reason for the XML Spec to state what it lacks (or
lots of other technical decisions).  

(I should note that there is another reason possibly constributing to
getting rid of SDATA entities: some people thought that entities
should/would be superceded by URIs and links. In this view, entities
were an 80/20 syntax and XML would shortly have something better...so
just provide enough entity capability for removing constants to headers
and leave the rest till later.  I am sure there are different
perspectives on this.  The view that XML was for web distribution was
very strongly held by some: one of the arguments against the encoding
header was based on consistency--that XML was a distribution format
which should use MIME headers and a feature made to overcome the
weaknesses of operating systems was outside-the-requirements. However,
one still needs to store resolved documents,
so there is no contradiction.)

Rick Jelliffe