Re: DOCBOOK: Re: objection to docbook.dcl

[Adam Di Carlo <adam@onshore.com>]

Karl Eichwalder <keichwa@gmx.net> writes:

> Norman Walsh <ndw@nwalsh.com> writes:
> 
> > / Adam Di Carlo <adam@onshore.com> was heard to say:
> > | accept.  The unnecessarily broad divergance of the shipped Docbook
> > | declaration puts a burden on document engineers using DocBook.
> 
> > This whole problem is probably the result of documentation errors on
> > my part. The declaration shipped with DocBook is advisory and was
> > never intended to be normative: the documentation should state that
> > clearly.
> > 
> > There's no reason why you should use it if your software behaves
> > better with a different declaration.
> 
> Yes, for authoring purposes.  Nevertheless you'll have to distribute a
> SGML declaration.  Otherwise it's impossible to say what's a valid DocBook
> SGML document and what's not.  IIRC, the "defaults" are rather
> restrictive (I'll have to re-read the Handbook, though).

Yes -- I'm not against a declaratation.  I'm just against a
declaration which is unnecessarily restrictive, with the consequence
that a non-trival number (perhaps 30%?  more?) of the docbook SGML
documents out there will fail to conform to it.

I don't know -- maybe there's a good reason why for interchange
OMITTAG must be off.  Do some tools out there not understand that?

Maybe's there's a good reason the NAMELEN is so tight, but I don't
know it -- it should match the reference concrete syntax (which I
guess I don't have a copy of -- I muddled it with the implied SP
declaration).
 
> > The docbook.dcl file was designed with the goal that any document that
> > validated against it could be safely exchanged.
> 
> Yes and that's a point to respect.  For more info please read the
> documentation coming with TEI.
> 
> [BTW, that's why a centralized CATALOG will not work as long as DTDDECL
> or DELEGATE isn't in wide spread use and supported by software.  For
> every single SGML document (TEI, DocBook, HTML) you've to use the right
> SGML declaration.]

Well, OpenJade *does* support DTDDECL, and that's what led me to
examine this issue. (Whether OpenJade is widespread is another issue,
but it is better than Jade IMHO).  

Since I was (but am no longer) shipping the Debian package catalog
files with the DTDDECLs intact, as provided to me from OASIS, I
suddenly had a situation where openjade couldn't cope with my DocBook
SGML files.

I have no philosophical problem with a strict DECL for interchange.

However, that leads me to wonder whether perhaps we should establish a
special SGML FPI with the more restrictive declaration in effect, and
leave the implied declaration for normal files.  Some kind of "DocBook
Strict" perhaps?

-- 
.....Adam Di Carlo....adam@onshore.com.....<URL:http://www.onshored.com/>