Yas, you are on record as favoring the attribute proposal for
implementation purposes. I'm wondering why the existing proposal is hard
to implement. They are all XML, after all, so internally you would just
reduce them to some ID.
And
Both reduce to (inventing my own pseudo-code):
index-range-start(foo:bar)
The test would be to make sure index-range-end(foo:bar) also exists. Why
is one syntax harder than the other? The FO plugin for the DITA Open
Toolkit already detects mismatched index page ranges today.
The other issue that both you and Paul G are eminently qualified to
address is the usability issue. The proposed change to index ranges is
essentially a switch from element/content-based authoring to an
element/@attribute approach. You essentially require that the author
come up with an ID to assign to an attribute. Existing XML authoring
tools simply don't make attribute editing/viewing easy. Everything I've
heard about user friendly XML authoring says to avoid authoring
attributes (which are usually invisible), let alone coming up with IDs
for those invisible attributes, let alone authoring matching pairs of
those invented IDs for those invisible attributes. Won't this approach
be unacceptably unfriendly to your users?
Chris
-----Original Message-----
From: Yas Etessam [mailto:yas.etessam@xmetal.com]
Sent: Tuesday, August 08, 2006 9:26 PM
To: dita@lists.oasis-open.org
Subject: RE: [dita] some index-range-* issues
We can avoid cross-topic ranges by having some guidelines that users
shouldn't be creating orphaned range tags into their topics. Paul G's
new markup proposal will create XML with enough information that
implementors could warn users when they've got orphaned tags. Even
though the DTD can't prevent the orphaned tags, XML editors could
theoretically warn users about that scenario.
In terms of what we want to support, it doesn't make much sense to
modify DITA to accommodate a 'poor indexing practice' if the best
practice is to simply show the start page. In terms of technology, it
wouldn't be difficult to add some type of "range" attribute on an
indexterm within topicmeta to indicate to some output process to include
both the start/end pages but the real question is "should we"? JoAnn
seems to be suggesting that we should try to encode best practices
within our data model as opposed to being overly accommodating.
- Yas Etessam
________________________________
From: JoAnn Hackos [mailto:joann.hackos@comtech-serv.com]
Sent: Tuesday, August 08, 2006 2:47 PM
To: Dana Spradley; Grosso, Paul
Cc: dita@lists.oasis-open.org
Subject: RE: [dita] some index-range-* issues
Dana,
You're echoing my thoughts, reflected in this earlier memo.
Perhaps the additional confusion here is moving across topics. It would
seem better to avoid cross-topic indexing ranges completely. Would that
still be an option?
I think the current state of the proposals in both cases tries to
accommodate poor indexing practices that ignore the usability of an
index for actual readers. The simplest method is to give the page number
only for the first page of a longer item, letting the reader decide when
he has had enough. Some indexers use ff (folios or numbers of pages)to
indicate a longer discussion, beginning on a page, such as 356ff with
the ff in italic. There is something problematic, it seems, to have page
ranges that span topics, given our case for the standalone nature of a
topic.
Anyway -- just a few thoughts on the philosophy behind the technical
debate.
JoAnn
JoAnn T. Hackos, PhD
President
Comtech Services, Inc.
710 Kipling Street, Suite 400
Denver, CO 80215
303-232-7586
joann.hackos@comtech-serv.com