forrest-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "J.Pietschmann" <j3322...@yahoo.de>
Subject Re: FAQ DTD changes #2
Date Wed, 07 May 2003 19:11:41 GMT
Jeff Turner wrote:
> What is out-of-band about <elaboration> in the FAQ?

Well, given the following

[inlining elaborations]
> That's a feature not a bug ;P 

I think I got the intention a bit wrong. If you actually want
to inline elaborations, your approach is ok.
This raises the question why you wrote you needed an xsl:if.
I'd do
   <xsl:template match="elaboration" mode="index"/>
   <xsl:template match="elaboration">
    <!-- normal operation -->
    <xsl:apply-templates/>
   </xsl:template>
Am I missing something?

It might be an issue that <question> is expected to result
in a single text block. If <elaboration> may suddenly introduce
block-like stuff, like your <pre> above, things may become
interesting.

An alternative idea would be to keep the full info in <question>
and resort to a separate optional <title> for the index:
   <qa>
     <title>NullPointerException at Forrest startup</title>
     <question>
      I am trying to run Forrest v0.4 on a W2K box running JDK 1.3.1
      and keep getting a NullPointerException on startup. The
      stacktrace I get is:
        <pre>
        ....
        </pre>
      What am I doing wrong?
    </question>
    <answer>
      ...

[snip]
> Optional metadata of course. If metadata has some obvious benefit, people
> will use it.
The problem is that the metadata is often only really beneficial
if it is consistently used. Take your example of the FAQ creation
date, mostly for the benefit of flagging new FAQs. Someone adds
an important FAQ in a hurry without supplying a date, or mistyping
it grossly, or the patch with a proper adding date wasn't applied
until the "new" period passed. The FAQ is not flagged. People get
upset, but probably no action is taken. The next FAQ is also added
without date. There is no longer an obvious benefit in providing an
adding date in patches. The feature falls into disuse.
Well, the demise is not inevitable, but I've often seen a small
fluktuation in available ressources causing it ("...will do later...").
Once a certain amount of data with unmaintained metadata accumulated,
not even the most dedicated fellows will bother to fix it.

If you want metadata to be useful:
1. Make it *mandatory*, not optional.
2. Provide automated plausibility checks.
3. Preferably provide some automation tools.
4. Fix all the old data.
5. Sell its advantages aggressively.

>>> <meta name="..."> ... </meta> tag allowed inside <faq>. 
>>*Cough*! Do you want people asking why the encoding they
>>added to a FAQ "doesn't work"?
> Character encoding?  Unsupported tags inside <meta>?

The possible problem I see is that <meta> for documents goes more
or less unchanged into the HTML (I may be wrong here), which might
cause people to think the FAQ <meta> has a very similar purpose,
and that the possible <meta> keys are basically to the ones often
seen in HTML.

This could be an interesting application for namespaces:
The document/XHTML schema defines stuff for general purpose layout
and closely associated stuff, like <xhtml:meta> or <doc:meta> for
general purpose document level metadata.
The FAQ schema could use <faq:meta> for FAQ metadata and faq:section
for structuring FAQs, compared to <xhtml:section> or <doc:section>
for structuring general purpose documents.

J.Pietschmann


Mime
View raw message