RE: [docbook-apps] Collecting glossentries in glossary database

["David Cramer \(Tech Pubs\)" <dcramer@motive.com>]

The downsides of using linkends and ids are:

1) You must include the glossary in the doc so that it will validate (or
modify the DTD so it won't consider linkends on glossterms IDREFs).
2) It's extra work for the writer to add the linkend to every glossterm,
and 
3) If you're editor helps you out by populating other IDREF attributes
with all the IDs in the doc, all the glossentry ids will clutter that
list of candidate IDs (which would otherwise consist mostly of chapter
and sections).

We do use linkends and ids and some glossary generation xslts of our own
that predate the ones in the stock xsls. We have a system that actually
modifies the DTD so that the linkend attribute on glossterm is an
enumeration of all the ids in our master glossary (i.e. that bit of the
dtd is dynamically generated). That solves #1 and #2. #3 we solve by
customizing the editor. It all ends up working, but took effort to set
up.

David

> -----Original Message-----
> From: Georges Schmitz [mailto:georges.schmitz@heitec.de] 
> Sent: Thursday, July 07, 2005 9:14 AM
> To: docbook-apps
> Subject: Re: [docbook-apps] Collecting glossentries in 
> glossary database
> 
> Using text content of glossterms, one drawback comes into my 
> mind: the existing mechansim cannot deal with 2 acronyms 
> spelled in the same way, but meaning different things. This 
> may exactly happen in the aviation licensing domain; you have 
> international and national acronyms that intersect.
> 
> If I split national and international glossterms in different 
> glossdivs, I cannot even be sure, that linking gets me to the 
> right page.
> 
> Just a thought...
> Georges
> 
> > ...<glossterm linkend="glo-jar">JAR</glossterm>...
> > I'm asking myself why the stylesheets are only relying on the text 
> > node content of glossterm and firstterm (or the |baseform| 
> attribute), 
> > and why the stylesheet will not collect the glossary 
> entries by using 
> > the linkend attribute (wouldn't this be safer anyway)? Is there a 
> > special reason for this?
> >
>