> or there is
>
> > consider each element type individually

I'm tempted to say that I like this idea better, but knowing nothing about
how XSL processors are implemented, I'm worried that this might increase
parsing time a lot.  If that's the case, considering that XML/HTML documents
are really meant to be read by a computer, and being human-readable is just
an added benefit, I'd rather have quicker parsing and harder reading then
slower parsing and pretty output.  (Heck, if it's *really* messy, and I
*needed* to read the source, I'd just declare the document as XML and view
it in IE5 or another XML viewer, and let it format it for me.)

I remember there being a discussion on the Cocoon
mailing list about this.

I don't remember the details, but there was some
mode or other which did "pretty printing", and I
seem to recall that they ended up resorting to
element-by-element rules. I think the concensus
was, though, that it was a pretty workable solution.

You might want to have a trawl through the Cocoon
list archives (no, I don't know where they are),
or have a look at the open-source files ...

-------------------------
James Robertson
Step Two Designs Pty Ltd
SGML, XML & HTML Consultancy
Illumination: an out-of-the-box Intranet solution

http://www.steptwo.com.au/
jamesr@xxxxxxxxxxxxxx