Re: DOCBOOK-APPS: SGML vs XML

[Bob Stayton <bobs@caldera.com>]

On Tue, Jul 30, 2002 at 02:48:08PM -0700, Paul Grosso wrote:
> At 14:42 2002 07 30 -0700, Bob Stayton wrote:
> >On Tue, Jul 30, 2002 at 01:43:54PM -0700, Paul Grosso wrote:
> >> At 17:35 2002 07 30 +0200, Jirka Kosek wrote:
> >> >Trevor Jenkins wrote:
> >> >
> >> >> > There's a suggested workaround here [4], [5].
> >> >> > [4] http://www.dpawson.co.uk/docbook/styling/profiling.html
> >> >> > [5] http://www.kosek.cz/xml/dboscon/profiling/frames.html
> >> >> 
> >> >> Thanks for those links. I lost them in a recent system failure. But this
> >> >> "profiling" technique is, I believe, an obfuscation and that's something
> >> >> that working on SGML has taught me to eschew. It has it's place but not in
> >> >> the depths of a 500+ page manual.
> >> >
> >> >It works for documents with virtually any length. And there is also one
> >> >advantage -- if you want conditional processing you are not forced to
> >> >learn new syntax (SGML conditional sections) you just use attribute with
> >> >some value.
> >> 
> >> I admit I haven't read the refs, but relative to marked sections, there
> >> are limitations to using elements/attributes for profiling.
> >> 
> >> If the DTD requires exactly one title per chapter, there is no way using
> >> element/attribute "profiling" that you can have two alternatives for the
> >> title depending on the profiling, because that won't be allowed by the DTD.
> >> 
> >> With marked sections (organized so that only one of the marked section
> >> keyword parameters evaluate to INCLUDE at any given time), you don't have
> >> this problem.
> >
> >Actually, there is a workaround for this in XML using
> >the <phrase> element.   You have one title element
> >but put two phrase elements in it, which are profiled. 
> >
> ><title>
> ><phrase role="foo">My Foo Product</phrase>
> ><phrase role="bar">My Bar Product</phrase>
> ></title>
> 
> Bob,
> 
> That's an interesting solution for titles in DocBook, but you aren't
> saying that their is always going to be a workaround to this general
> issue, are you?
> 
> I believe it is still the case that there is no element/attribute
> solution to the general profiling scenario that works in all cases
> that marked sections work.  Am I wrong?
 
No, you are right.  Marked sections are much more
powerful because they operate at a 'higher' level
than elements.  You can have duplicate elements and ID
attributes, and marked sections can also cross element
boundaries, as long as the thing validates in the end.

Are you going to lobby to put marked sections back
in XML?  8^)

-- 

Bob Stayton                                 400 Encinal Street
Publications Architect                      Santa Cruz, CA  95060
Technical Publications                      voice: (831) 427-7796
Caldera International, Inc.                 fax:   (831) 429-1887
                                            email: bobs@caldera.com