RE: [dita] Attempt to summarize position on nested sections

["France Baril" <France.Baril@ixiasoft.com>]

Hi Paul, I see your point. 

I'm resistant to your argumentation because I'm trying to protect people
against themselves, which in turn might force them into solutions that
are not always the most natural.

I guess we're back to defining what the smallest reusable unit of
information is in this example, which is kind of hard  after the fact.
It mostly depends on what the intention was when the project was
created. 

Yet the article example fails to convince me. If we agree that the
Information section is task-like and could be re-used. We're already
splitting the example into 2 topics. In this case, it doesn't cause any
problem because you can add a topic at the end of another one. What
would happen if the task-like section (or any reusable section) was in
the middle of an article? Isn't it easier to make the article from
topics, even if it means some are not reusable? (Could they be written
as reusable units?) 

Allowing nested sections might means that some information that could be
reused will not.

As far as polluting the searchspace with topics is concerned. Most good
search-engine will allow you to seach for reusable topics only, if a
reusable=true/false attribute is made available in the topics
themselves.
 
France

-----Original Message-----
From: Paul Prescod [mailto:paul.prescod@blastradius.com] 
Sent: 14 novembre 2005 06:04
To: France Baril; Michael Priestley
Cc: dita@lists.oasis-open.org
Subject: [dita] Attempt to summarize position on nested sections

Sorry for the delay on the response to this. 

I wouldn't agree with your characterization, exactly. There are a few
words that I think are overstated. First, we aren't really talking about
making _everything_ techncially available. We're talking about a
particular narrow issue of sections within sections. Sections within
paragraphs, within lists, etc. would still be disallowed.

Second, your second alternative conflates two ideas that I think are
separate. We all agree on the goal of keeping "semantically independent
topics separate." The question is whether specializers can be trusted to
define for themselves what is a semantically independent topic or
whether this decision will be forced upon them by baked-in limitations
of the framework.

Third, I disagree that "more topics" always equates with "better
usability". Consider the following topic:

http://support.microsoft.com/kb/906294 

Look at the second section, which has two sections itself. (let's put
aside the third section which is task-like)

In my opinion, this article has perfectly fine usability, and splitting
it into three or four topics would not improve it at all.

Fourth: DITA already allows topics nesting and therefore allows a whole
novel or website to be written in a single topic.

So I disagree that usability is an issue here at all. Rather, the issue
is about _re_usability. Michael feels that forcing people to put content
into sub-topics rather than sub-sections makes their information more
_re_usable. So, for example, he argues that the "resolution" section
should be its own topic in case it is reused anywhere. I agree to a
point: if an analysis undertaken by Microsoft suggests that "resolution
sections" are typically reused rather than authored for a specific
context then they should be topics (and content-managed objects or flat
files). 

On the other hand, if the analysis suggests that they are seldom or
never reused, then it would actually be incorrect and destructive to
model it as a topic: "just in case." If you do, you are polluting the
search-space for topics unnecessarily. You're also making your
processing much more complicated because you need to trick your DITA
implementation into giving section-like behaviour to a topic: "don't'
split this into a separate output file." "Don't use the authored title:
always use the title RESOLUION." 

My point is that one must make the topic-ness decision based upon
usability, content analysis and modelling, not based upon the fact that
the section happens to contain sub-sections. That is one of several
indicators that might _suggest_ that a section-like thing should be a
topic. But it doesn't _prove_ that a section-like thing should be a
topic. I think that this topic is one example where it is not.

> -----Original Message-----
> From: France Baril [mailto:France.Baril@ixiasoft.com]
> Sent: Tuesday, November 08, 2005 10:39 AM
> To: Paul Prescod; Michael Priestley
> Cc: dita@lists.oasis-open.org
> Subject: RE: [dita] Two proposals for nested sections
> 
> Paul, Michael, would you please restate the problem as you perceive 
> it?
> 
> My perception is that we are discussing making everything technically 
> available vs following a topic-based philosophy where a topic has the 
> purpose of keeping semantically independent topics separated in a 
> structure that reinforces a specific usability criteria about nesting.
> 
> France