>
> I'm repeating myself, but the whole <span class="rss:item"> is nothing
> more than a temporary kludge. The real answer is to get RSS
> produced by default by as many CMS as possible.
>
Actually, it is an example of an Architectural Form - an concept from SGML.
The difference between:
<span class="rss:item">
and
<rss:item html="span">
Is pretty minimal & mainly textual (at the file storage level).
If you give configuration info to a SAX parser - like "use 'class=' to get the tagname" then you can consume either text format & the result (sax event or DOM, whichever) looks the same. (I don't know of any SAX parser that has this built in, so this unfortunately is only xml theory, not xml practice.)
Using architectural forms lets the publisher decide what they like & makes 'transmogrifiying' relatively simple and efficient. Arbitrary transforms - like with XSL - can also do the job, but are more expensive computationally.
If you try to get each producer to use a common format - which probably is not native to their data model - then you are pushing the burden of transformation onto them. Separating this allows for a 'middle-man' to do the transform. If you don't have a system with a middle-man (like web-servers & web-browsers originally had) then you'll probably have to push this onto either the server or the client. If servers already have a plug-in approach (like a CGI/servlet/etc.) then it probably isn't that difficult for providers to do the work - but they won't unless there are enough clients.
Mike