[
Lists Home |
Date Index |
Thread Index
]
Norman Walsh wrote:
> / Elliotte Harold <elharo@metalab.unc.edu> was heard to say:
> | I don't know about W3C schemas, but the XML spec is not at all clear that=
> a
> | non-deterministic content model in a DTD is an error. Indeed some version=
> s of
> | Xerces have not so flagged it.
>
> From=203.2.1 of XML 1.0 3e, it seems pretty clear to me:
>
> For compatibility, it is an error if the content model allows an
> element to match more than one occurrence of an element type in the
> content model.
>
> Granted, "for compatibility" says, in short, we didn't have to do it
> this way, but we did for SGML compatibility. I don't think that
> negates the normative weight of the rest of the sentence though.
But even SGML doesn't require conforming systems to report
this condition; [4.267] says an ambiguous content model is not
a "reportable markup error".
(That's probably because nobody knew *how* to detect this condition
until 1992 when Anne Bruegemann-Klein figured it out, but still...)
Anyway, AIUI the consensus in the SGML world was that
the prohibition against ambiguous content models was
overly restrictive and unnecessary. That W3C XML Schema
retained the restriction was a big mistake, IMO, one
that could easily have been avoided by not conflating
validation with annotation.
--Joe English
jenglish@flightlab.com
|
