Re: [dita] Thoughts on Issue #20

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

re 1:  there is a req to strip
values from output - you don't want internal rev numbers or values that
reference unreleased products being included in output, for example. So
there do have to be two actions - include content while stripping values,
and include content while preserving values. Don't much care what we call
them.

re 2: dunno.

re 3: since this part was your idea,
I'm happy to take suggestions :-) In the previous/existing behavior, only
the rev attribute could be flagged with color, so this situation never
came up.

re 4: not sure I understand your grounds
for objection fully, but in any case the proposal does describe how to
change the default behavior for a particular ditaval file, but I'm reluctant
to change the default for every ditaval file because, regardless of the
merits of your argument, it would be backwards incompatible (ie every document
out there created based on existing default behavior would break with your
change). Hence my desire to make the change an opt-in rather than global.

Michael Priestley
IBM DITA Architect and Classification Schema PDT Lead
mpriestl@ca.ibm.com
http://dita.xml.org/blog/25

"Paul Prescod" <paul.prescod@blastradius.com>

04/04/2006 10:52 AM

To	<dita@lists.oasis-open.org>
cc
Subject	[dita] Thoughts on Issue #20

Apologies for not getting this out sooner. Better
late than never.

#1. Is it important to have "include" and "passthrough"
be distinct
values? Or can we merge their behaviour and say that include always
attempts to "passthrough" if the output format supports it.

#2. Is "printchar" really implementable in most output formats?
E.g.
HTML, HTML Help? Is it easy or hard to implement in XSL-FO?

#3. What happens if two attributes on the same element suggest
conflicting colours? Or an inherited attribute suggests one colour and
an inner one suggests another.

#4. The logic for action is not intuitive because it is not the same as
in-document. In the document, we say that if an attribute is not
mentioned at all, that means "include". If it is mentioned with
a
particular postive value, then it means "exclude everything except
this
value". In the ditaval file, "include" is a no-op and "exclude"
means
"exclude if this value is set." This is non-intuitive both because
it is
inconsistent but also because it means that you end up listing
everything you DON'T want to handle. People are more comfortable with
positive logic than negative logic.

My proposal: include means "Include if this value (or another included
value) is in the list. Exclude otherwise."

In that model, "Exclude" would be the no-op.

 Paul Prescod