RE: [dita] otherprops syntax - should we specify?

[Michael Priestley <mpriestl@ca.ibm.com>]

This is an excellent discussion, and
I think the onus is now on me to provide some background on why we've taken
the approach we have, since it does differ from what we've all seen in
other architectures. This difference is in fact not an oversight, but is
based on extensive experience with full boolean logic conditionality, and
is designed to be an improvement on traditional conditional processing
architectures.

There are two factors that make DITA's
approach a bit different from traditional architectures:

1) Maps. Singlesourcing in DITA is primarily
accomplished by using different maps; each map can include a mix of shared
and unique topics without requiring communication or coordination between
the reusers. 

Because maps provide the main mechanism
for reuse and singlesourcing, conditionality is only required within relatively
contained and limited contexts, like a single topic. If the conditionality
gets too complicated to maintain easily, the problem topic can be refactored
into two or more separate topics that share common elements through conref,
instead of having a single topic with complex metadata. This is covered
in detail in briefing 5 of the DITA deepdive series.

2) Semantic groupings of values into
separate attributes, so you have multiple axes of metadata rather than
an undifferentiated list of values. This semantic grouping allows implicit
boolean logic to be applied at build time, following the rules laid out
in the spec (ie all the values in one attribute must be excluded for the
element to be excluded).

The existing attributes are not a complete
set of metadata axes, however; hence the need to provide semantic groupings
within the one attribute that does not have a specific semantic already
associated with it (otherprops).

There are several advantages of semantic
groupings over boolean logic:

1) they can be easily understood by
new writers (eg when a document's ownership is transferred or shared).
2) as a result, they can accomodate
longer lists of values (ie heavier reuse) without becoming unmaintainable.

3) because they are simple metadata
assertions, they can be used for more than just filtering (eg flagging
and search).

Adding boolean logic support would compromise
all of these benefits. There is a cost associated with maintaining complex
logic that can drastically undermine the cost efficiencies of singlesourcing.
I've seen many projects hit this wall with traditional approaches, which
is why we took a different approach, supporting coordinated strategies
based on conref, maps, and metadata rather than using condition lists with
boolean logic.

A specific note about NOT values:
- a NOT value cannot be easily used
for anything except filtering (eg hard to flag).
- a NOT value can be rendered invalid
when new platforms/products are added (does the paragraph still apply to
every platform except Linux once you add other flavours of Unix?)
- if a NOT value absolutely must be
used, it can be handled without explicit boolean support by creating a
new unitary value; eg instead of "NOT Linux" use "NOTLinux".


Thanks for the discussion, I hope this
background is useful.

Michael Priestley
mpriestl@ca.ibm.com

"Rob Frankland" <robf@rascalsoftware.com>

01/10/2005 06:08 PM

To	"'W. Eliot Kimber'" <ekimber@innodata-isogen.com>, "'Christopher Wong'" <cwong@idiominc.com>
cc	"'DITA TC list'" <dita@lists.oasis-open.org>
Subject	RE: [dita] otherprops syntax - should we specify?

I'll add my two cents worth here, although I have
not coded anything using
DITA. Users want conditional that are both easy to use and powerful. I
cannot tell from the discussion if DITA conditionals support any type of
Boolean logic but they will need to in version 2.0, if not now. At the
simplest, users want to define three conditions and be able to make a
selection of two which seems possible currently. However, making a condition
based on a NOT does not seem to be accommodated. If I am way off base,
I
apologize in advance.

Rob

-----Original Message-----
From: W. Eliot Kimber [mailto:ekimber@innodata-isogen.com] 
Sent: Monday, January 10, 2005 1:56 PM
To: Christopher Wong
Cc: DITA TC list
Subject: Re: [dita] otherprops syntax - should we specify?

Christopher Wong wrote:

> Does anyone else know what other people do with otherprops?

I would think that any non-trivial use of DITA that needs to account for
business-specific conditionality would quickly find the DITA-supplied 
conditions insufficient. Therefore I would assume that almost *any* use
of DITA that needed conditionality would require additional condition 
types and valuds.

This would certainly be the case for any client I'm likely to have that
wanted to implement a DITA-based XML application.

Cheers,

E.
-- 
W. Eliot Kimber
Professional Services
Innodata Isogen
9390 Research Blvd, #410
Austin, TX 78759
(512) 372-8122

ekimber@innodata-isogen.com
www.innodata-isogen.com