Tobias Reif wrote:
>> But suppose I could,
>> how would you propose to formally describe the structures that are
>> valid. What constitutes valid is not an academic question, it has
>> direct bearing on how tools work.
>
> I'm not sure what I meant back then, but I do think that there should be
> one normative schema included in the spec (eg a normative/official RNG).
> I guess what I meant was that tools processing DocBook documents should
> not be required to support any specific schema language (eg requiring
> WXS+PSVI support for conformance). DTD support is required through the
> XML spec, but ideally the DBX spec should avoid building on this type of
> dependency.
Having formal syntax description of DocBook in some language (e.g.
RelaxNG) as normative part of DocBook standard doesn't mean that tools
must be RelaxNG aware. Many RFCs uses EBNF syntax diagrams to precisely
define syntax of protocols but that doesn't imply that you must use some
sort of EBNF code generator when programming your own implementation of
that protocol.
I definitively think that DocBook standard should formally, precisely
and unambiguously define DocBook grammar and this can't be done without
some sort of formal language. AFAIK the best language for this is RNG at
these days.
Jirka
--
-----------------------------------------------------------------
Jirka Kosek
e-mail:
http://www.kosek.cz
