Lists Home |
Date Index |
Elliotte Harold wrote:
> I do wonder if the attribute order issue suggests a possible way out of
> the conundrum though. Attribute order in XML is not significant, period.
> This was required not by the original first edition XML spec but by the
> SGML spec which XML 1.0, first edition, incorporates by reference.
> Does this same SGML spec place any other requirements on SGML and by
> extension XML processors? In particular does it mandate anything that an
> SGML processor is expected to return to client applications? Do any of
> the SGML gurus on the list happen to know that?
ISO 8879:1986 didn't specify what data a parser was
required to return to the application; that was added
later in the Element Structure Information Set (ESIS)
attachment. I don't remember if 8879 was ever amended
to include the ESIS attachment, but it was pretty much
There's a copy here: http://xml.coverpages.org/WG8-n931a.html
It's more or less equivalent to the items listed in the
XML Information Set, minus the stuff that's new in XML
(like namespaces), plus some stuff that was removed
(like link rules).
The ESIS spec introduced an important idea, the difference
between "structure-controlled" and "markup-sensitive"
| 1. A structure-controlled SGML application operates
| only on the element structure that is described by
| SGML markup, never on the markup itself.
| 2. A markup-sensitive SGML application can act on the
| actual SGML markup and can act on element structure
| information as well. Examples include SGML-sensitive
| editors and markup validators.
Later, DSSSL and HyTime TC1 introduced the notion of
groves and grove plans, along with a modularized
definition of the full SGML property set. This
spec is much more concrete than the ESIS, and is
extensible -- it's good for describing things
like the structure of parsed microcontent (e.g., SVG)
or augmentation processes (e.g., a WXS PSVI).