Hi,
The stylesheets currently don't have a method of generating text for an xref
to a mediaobject. While it might seem obvious to use the caption text, that
is not always appropriate. I've seen captions with multiple paras and an
equation or two, and all that text is not appropriate for an xref link. The
same problem applies to an xref to any element that can contain a mix of
block elements. Even a para can contain other block elements.
One solution is to put the mediaobject inside a figure element, move the
caption text to the figure title, and xref to the figure. But figure titles
are numbered, and you may not want that.
Another solution is to put an id attribute on the para inside the caption,
and add an endterm attribute to the xref that points to that id:
<para id="ghist">Graphical history of the <filename
class="directory">hello</filename> repository</para>
<xref endterm="ghist" linkend="fig.tour-basic.history"/>
See this reference for more information on the use of endterm:
http://www.sagehill.net/docbookxsl/CrossRefs.html#OptsGenXrefYou could also develop a customization that uses the caption for the
generated text for an xref to a mediaobject.
Bob Stayton
Sagehill Enterprises
bobs@sagehill.net