John L. Clark wrote:
> I'm not exactly sure what algorithm the DocBook-XSL distribution uses,
> but I think Jan's idea was that if an xref is pointing to a glossentry,
> then the text representing that xref should be the glossterm that that
> glossentry is defining. I've come up with my own version which makes
> this explicit and which handles cross-referencing to bibliographic
> entries in a particular way:
>
> ---
> xref {
> /* NOTE: This uses a proprietary XXE XPath extension. */
> content: xpath("\
> if((id(@linkend)/self::biblioentry or id(@linkend)/self::bibliomixed) \
> and \
> id(@linkend)/@xreflabel, concat('[', id(@linkend)/@xreflabel, ']'), \
> id(@linkend)/self::biblioentry or id(@linkend)/self::bibliomixed, \
> concat('[', @linkend, ']'), \
> id(@endterm), concat('\201c ', id(@endterm), '\201d '), \
> id(@linkend)/@xreflabel, \
> concat('\201c ', id(@linkend)/@xreflabel, '\201d '), \
> id(@linkend)/title, concat('\201c ', id(@linkend)/title, '\201d '), \
> id(@linkend)/self::glossentry, id(@linkend)/glossterm, \
> concat('[', @linkend, ']'))");
> color: navy;
> }
> ---
>
> Notes: This implementation formats bibliographic entry references as
> [Label]. The "\201c " and "\201d " are the CSS-escaped Unicode
> characters for the left double quote and the right double quote,
> respectively. Finally, if an xref makes it through all the way to the
> default condition, then we also use the [Label] formatting since we're
> not sure what sort of object lies at the other end.
>
> I'm attempting to be compliant both with DocBook: The Definitive Guide
> and the behavior that I experience from the DocBook-XSL distribution.
> Naturally, season to taste. And feel free to let me/the list know if
> you make general-purpose modifications to this.
Thanks for this contribution. I'm sure this is the best of what could be
done with XXE.
But my question is now: do we need to go so far in emulating the XSLT
style sheets?
For me, the fact that a document looks very different on screen and in
print is really not a problem.
In my opinion, the CSS should be only used to make the text easy to read
and the structure easy to understand.
See http://www.syntext.com/products/serna/ for a very different approach.
