Re: [docbook] Ruminations on the future of DocBook

[Tobias Reif <tobiasreif@pinkjuice.com>]

Jirka Kosek wrote:

 > Tobias Reif wrote:
 >
 >> I mean that no feature of DocBook should rely on any feature from any
 >> specific schema lang, and that no single specific schema lang should
 >> be normatively referenced in any DocBook spec.
 >
 > Relax NG is only tool which can be used to formally and precisely
 > describe syntax of DocBookNG.

Yes.

 > Of course you can describe document type and all its content models in
 > prose, in human readable text. But this will be very verbose and very
 > hard to use description.

I think you misunderstood me. Sure there must be at least one normative 
schema in addition to the human lang spec. The XHTML 2 working group for 
example plans to supply the schema in three versions, DTD, RNG, WXS 
(XSD), all of them will be normative (AFAIK).

 > As Relax NG doesn't have features like attribute defaulting its usage
 > has no sideeffects

I know, that's one of it's design goals.
(I was talking about general aspects, not just RNG.)

 > and you don't need process RelaxNG grammar in order
 > to process DocBook

Yes, that is actually one symptom of my suggested goal.

 > (however this is not true for DTD).

And it could be false for yet another schema language. Any such 
dependence should be avoided.

 > But there
 > definitively should be formal description of DocBookNG grammar, this
 > formal description should be normative

Yes.

 > and Relax NG is the most
 > suitable tool for this task at these days.

It will satisfy most of the requirements, yes.
(Note that Norm lists one shortcoming:
http://norman.walsh.name/2003/05/21/docbook
"A future version of RELAX NG might give us back our exclusions.")

None of your points contradicts anything I said.

Let me repeat: IMHO, no feature of DocBook should rely on any feature 
from any specific schema lang, and no single specific schema lang should 
be normatively referenced in any DocBook spec [added for clarification:] 
  as required for a conforming implementation.

 > P.S. I notices that I'm using DocBookNG instead Norm's DocBook V.next
 > label. But I mean the same.

I also thought that "DocBook NG" would be a good name, but what will the 
next version set (major backwards-incompatible refactoring) after that 
be called then?

Tobi

-- 
http://www.pinkjuice.com/