There is a drawback to this approach: the chance for human error is
multiplied tremendously across your corpus of documents. That is,
entering CSS directly in @rend works fine as long as no encoder ever
enters rend="font-style: italics" or rend="font-variant: small-cap"
(former shouldn't have an s, latter should). Personally, I'm not that
trusting. I'd rather customize the schema to allow only predefined
values for @rend ("italic", "small-caps", etc.) and then let my XSLT
translate to CSS for display.
University of Virginia Library
Martin Holmes wrote:
> Hi all,
> The global attribute @rend is wonderfully loose ("any string of
> characters", "These Guidelines make no binding recommendations for the
> values of the @rend attribute..."). However, I'm increasingly finding
> that CSS is everything I need and more for the @rend attribute, and in a
> current project, I'm considering converting all my old formulations
> (rend="italic" etc., derived from old usage, examples and guidelines)
> over to pure CSS (rend="font-style: italic").
> Can anyone see any objection to this on principle? Have you come across
> any rendering features which are impossible to describe in CSS?