This is the mail archive of the
docbook-apps@lists.oasis-open.org
mailing list .
Re: Re: Fw: Re: XHTML Stylesheets
- From: Steffen Maier <Steffen dot Maier at studserv dot uni-stuttgart dot de>
- To: Lars Trieloff <lars at trieloff dot net>
- Cc: Norman Walsh <ndw at nwalsh dot com>, docbook-apps at lists dot oasis-open dot org
- Date: Sun, 16 Dec 2001 17:09:15 +0100
- Subject: Re: DOCBOOK-APPS: Re: Fw: Re: XHTML Stylesheets
Hello,
On Sun, 16 Dec 2001, Lars Trieloff wrote:
> For creating valid XHTML 1.0 Strict/XHTML 1.1 we will have to
> - change the doctype
And not to forget the correct namespace on the html root element. That
problem was addressed in a recent thread on this list. IIRC Norm said it
wouldn't be so easy to accomplish this. So are we supposed to process the
xhtml-output once more to add the namespace? This should be no problem as
the output is well-formed xml and xslt can be used, right?
I tried different other patches to get the xhtml namespace continually
throughout the whole xhtml output files, all without success (which could
just be due to my limited xslt knowledge):
-adding xmlns="..." to my stylesheet driver => every other stylesheet
fragment imported/included is still missing that declaration (and that are
quite a lot both in website-xsl and dbxsl).
-<xsl:namespace-alias stylesheet-prefix = "#default" result-prefix =
"#default"/> together with xmlns="..." in my driver (in the hope this
would override the missing declarations in other fragments).
I had quite some tough hours this weekend figuring out why Mozilla (0.96
with MathML) doesn't render XHTML with embedded MathML correctly in all
cases. So far I would say it requires a correct namespace for html and
mathml-elements, the correct content-type (text/xhtml or text/xml?) or the
correct file-suffix (.xhtml or .xml) if loading through file:-protocol. If
MathML markup uses symbolic character entities one additionally needs a
system identifier of "mathml.dtd" which is IMHO a dirty fudge to resolve
them internally, but this is not necessary with dbxsl because by default
only numeric charents are output.
> + one thing that was already mentioned, the proper nesting of elements:
> there are no div elemtents allowed inside p elements (validator.w3.org says:
> "Error: element "div" not allowed here; possible cause is an inline element
> containing a block-level element"),
> so we have to replace nearly all p elements by div elements, one exception
> would be the simpara-->p transformation, which seems to be ok.
> this are the first things that have come into mind, so this list has to be
> filled.
Wow, this seems to be a really radical switch to complete CSS-reliance to
me. I wonder where to draw the border to not generate _any_ xhtml-elements
but <div>s and <span>s and some other necessary action-(e.g. <a>) and
meta-elements? What I mean is: We don't say goodbye to e.g. <b>,<i>,<hx>
and others because it could all be done with CSS, do we?
Bye,
Steffen.
--
http://w3studi.informatik.uni-stuttgart.de/~maiersn/
mailto:Steffen.Maier@studserv.uni-stuttgart.de