[TriLUG] Fwd: Re: DocBook and DITA

Mark Johnson mrj at debian.org
Sat Apr 16 09:41:55 EDT 2005


A nice summary of the distinctions between DocBook & DITA...

Cheers,
Mark

----- Forwarded message from Nancy P Harrison <nancyph at us.ibm.com> -----
    Date: Sat, 16 Apr 2005 06:54:11 -0400
    From: Nancy P Harrison <nancyph at us.ibm.com>
Reply-To: Nancy_P_Harrison/Lexington/IBM <nancyph at us.ibm.com>
 Subject: Re: DocBook and DITA
      To: Rajal Shah <rajal at meshsoftware.com>

[In the interest of full disclosure, it should be noted that the writer is 
on both the DocBook and the DITA TCs at OASIS, was involved in the 
original DocBook development, and is currently one of the DITA architects 
at IBM.]

Dear Rajal,

While I wasn't at the CMS conference, I'm sorry, and a bit surprised, to 
hear that you came away with the impressions you describe.  My responses 
to your questions are below; I hope they're useful.

"Rajal Shah" <rajal at meshsoftware.com> wrote on 04/15/2005 12:57:00 PM:

> I was amazed and frustrated to hear people downplaying Docbook over 
> DITA at a conference I attended this week on XML documentation and CMS.
> 
> I figure this group would be the ideal place to pose this question:
> 
> 1. What is the opinion on DITA?

As someone involved with both DITA and DocBook, and having used both, I 
don't see incompatibility between them. Rather, I see two XML-based 
architectures developed independently to meet different objectives. 

DocBook was developed to meet the needs of technical book publishers, for 
information designed around a hierarchical and linear model, hence the 
'book' part of the name. 

DITA, on the other hand, was designed around a topic-based, authoring 
model focused on reuse of information at the topic level.

So, if you're authoring a book, with the book structure that implies, 
you're probably going to want to use DocBook; it supports a complete 
processing tool stream for authoring and publishing books in multiple 
formats.

If you're authoring topic-based information centers, especially where you 
need to reuse and reorganize your information for different audiences or 
information subsets, DITA is a better fit for that; it was designed for 
that use.  And if you have a need to extend the information models to meet 
your specific purposes, DITA is also designed to enable that, while 
allowing reuse of your processing stream.

 
> 2. Does DITA share inline elements at least with Docbook? (para, 
> bullets etc.) What is the point in redefining those in DITA again if
> they were already done in DOCBOOK and since both of these standards 
> are part of OASIS.

Many of the DITA inline element names were derived from the corresponding 
elements in HTML rather than DocBook; HTML is a standard familiar to a 
very large number of authors, and this reuse of HTML tags allows for an 
easier learning curve for authors going from HTML to DITA authoring. 

The DITA community is interested in getting XSLT transforms created 
between DITA and DocBook, to enable interoperability of content created in 
either format.  With good transforms, there could be some very useful 
'hybrid' solutions; for example, maintaining a book's front matter and 
back matter in DocBook, while populating the body of the book, or even the 
body of individual chapters, with DITA topics nested and sequenced via a 
DITA map.

 
> 3. DITA has this inheritance concept and base their stylesheet 
> templates of the class attribute instead of the element name in the 
> source.. That way you can share the XSl and XSL-FO of the parent 
> element even if you customize/specialize the class. That does sound 
> like a good concept.. What do people think about it? Does DOCBOOK 
> have anything similar to it?

DocBook does not have anything similar to this. If you customize DocBook 
by adding - or changing the names of - elements, you need to customize 
your processing to get your changes processed. If your customizations are 
changes to attribute names or values, the standard DocBook processing may 
work, depending on the kinds of changes you make.

> 
> 4. DITA has limited support for XSDs and no suppport for RelaxNG.. 
> They are still in the DTD world. I've initiated the schema 
> technology thread a couple of months back on this mailing list and 
> we all universally agreed that DOCBOOK recommends the newer 
> technologies.. DITA was strongly advocating just the opposite at the
> conference.. Any comments?

I'm sorry to hear that you came from the conference with a 'DITA is 
anti-schema' impression. DITA has full support for schemas, and the Open 
DITA toolkit (available from http://sourceforge.net/projects/dita-ot/) 
contains both DTD and XSD implementations of the DITA architecture. 

Hopefully, there will even better schema support soon. Recognizing the 
growing importance of schemas relative to DTDs, the DITA community is 
hoping to add RelaxNG support to the toolkit in the near future, as well 
as examples of document instances using both XSD and RelaxNG schemas.


> 
> Regards.
> --
> Rajal
> 


Regards,

Nancy


____
Nancy Harrison
IBM Rational Software
Phone: 781-676-2535
nancyph at us.ibm.com


----- End forwarded message -----


-- 
____________________________________________________________
Mark Johnson      <mrj at debian.org>
Debian XML/SGML:  <http://debian-xml-sgml.alioth.debian.org>
Home Page:        <http://linux.duke.edu/~mark/>
GPG fp: DBEA FA3C C46A 70B5 F120  568B 89D5 4F61 C07D E242



More information about the TriLUG mailing list