W3C home > Mailing lists > Public > public-html@w3.org > July 2007

Re: "Canonical" HTML5 and XHTML5 output

From: Cecil Ward <cecil@cecilward.com>
Date: Thu, 26 Jul 2007 09:29:49 +0100
To: <public-html@w3.org>
Message-ID: <000001c7cf5f$21f9cbd0$65ed6370$@com>

Robert Burns wrote:

> Outputting as html should work (am i missing something?).

As far as XHTML5 is concerned: IMO you are never at liberty to use the output method=html from XSLT to generate XHTML because XHTML _is XML_, and an XSLT processor  told to output HTML has no reason to try to generate legal XML, it is at liberty to do horrible things that you can't control. It is allowed to do such things as omitting end tags, omitting quotes from attributes, using the wrong case, not outputting CDATA sections correctly, not outputting the right namespace declaration and a hundred other things. It's got no business outputting things like <br /> (which isn't even legal HTML). It's got about zero chance of working.

As for HTML5's HTML serialization, I'm not clear how many things there are to worry about. Nor am I clear what the best route is to take about generating the kind of XML that is UA-backwards-compatibility-friendly. Clearly, the XHTML1.0 appendix is a start, but there are possibly more issues introduced in HTML5 for all I know.

Hope that makes sense.

Another point has occurred to me. HTML5 had better legalise <br /> and <hr /> in both serializations, to legitimise all the valid XHTML 1.0 out there that has to be selectively served as text/html according to UA. Please tell me that's the case?

Cecil Ward.
Received on Thursday, 26 July 2007 08:30:10 UTC

This archive was generated by hypermail 2.3.1 : Monday, 29 September 2014 09:38:47 UTC