Which style first ? Re: Associating DSSSL style sheets with documents

[Len Bullard]

Jon Bosak wrote:
> 
> [Kazumi Saito:]
> 
> | Many of DSSSL style-sheet will be depend on DTD, I think. I want to
> | put style-sheet entity declearation on DTD.
> 
> I think that it's a mistake to think of stylesheets as having a
> one-to-one relationship to DTDs.  It is possible to make one
> stylesheet work with many DTDs (though this is probably not a good
> practice), 

I don't see why this is the case.  In practice, where stylesheets are 
in use, the opposite has been the case.  Many DTDs share a lot 
of structures and can in fact, share stylesheets.  It is an issue 
of the degree of overlap.  Where content tagging is practiced, 
it is often convenient to take another format-oriented stylesheet 
such as one might provide for HTML and use the style information 
with different GIs.  Where DTDs vary only in degree, the same 
stylesheet is used.  

One very useful side effect is organizations quickly 
realize how many non-useful variations they have in their document 
structures and start looking for ways to winnow these out of 
their practices.  One way to do that cheaply is to parse against a 
DTD the organization provides.  The DTD becomes a corporate 
policy and a repository of corporate memory.  This 
is useful when attempting large rehosting or conversion projects.
The last thing I want when converting documents is a large 
collection of well-formed but inscrutable markup.

> | About well-formed XML document which has no DTD, it is not so bad that
> | such document has style-sheet or pointer to style-sheet for
> | portability, I think.
> 
> For XML this will be the typical case.  Typical XML documents
> transmitted over the Web will not have DTDs and will have to point to
> (or include) one or more stylesheets that are intended to be used with
> them.

I'm not sure the common HTML web experience to date will be the most 
predictive model for sound practice with stylesheets or DTDs.  The one
example 
we have, HTML does have a DTD for whatever use is made of it.
Our experience with DTDless processing was that people quickly found 
it necessary or convenient to create them although they don't transmit 
them often as you point out. 

As people who did not formerly practice SGML will learn, unvalidated 
markup is a nuisance, having a DTD is the best way to find out 
what was intended by the originator of a marked up instance, 
and is a rigorous expression of policy.

len bullard

xml-dev: A list for W3C XML Developers
Archived as: http://www.lists.ic.ac.uk/hypermail/xml-dev/
To unsubscribe, send to majordomo at ic.ac.uk the following message;
unsubscribe xml-dev
List coordinator, Henry Rzepa (rzepa at ic.ac.uk)