Re: Re-registration of text/html

Henri Sivonen, Thu, 11 Mar 2010 03:40:28 -0800 (PST):
> "Sam Ruby" <> wrote:
>> Leif Halvard Silli wrote:
>>> The XHTML2 working group is meant to soon (re)announce that XHTML
>>> 1.1 documents can be served as 'text/html'.
>> That will make for a rather 'interesting' Last Call period then.  I 
>> would advise anybody who wishes to make such a recommendation to seek
>> wide input prior to that point.
> So far, the XHTML2 WG has put most of its MIME type advice in a 
> document that's not on the REC track and, thus, not subject to 
> mandatory Last Call review.
> Personally, I think WG Notes should be ignored as binding precedent 
> for REC track publications, because otherwise, Notes would provide a 
> loophole for subverting the REC Process. That is to say, I think the 
> XHTML Media Types WG Note should be disregarded when deliberating 
> what HTML5 should say about text/html (even if the XHTML2 WG 
> published a second edition of a REC that normatively referenced the 
> Note).

The 2002 revision of XHTML 1.0 contains not only Appendix C - but also 
refers to the year 2002 version of the XHTML Media note. [1] So the 
loophole is eventually 8 years old.

The 2002 version, which comes in a HTML4 'text/html' copy and an 
'application/xhtml+xml' XHTML 1.1 copy, summarizes everything in a 
table at the bottom which says that only "HTML compatible" XHTML 1.0 
documents MAY be served as 'text/html'. Whereas all other variants of 
XHTML, including all other variants of XHTML 1.0 documents, SHOULD NOT 
be served as 'text/html'. Hence, there never were any MUST NOT 
prohibition against serving "not so HTML compatible" XHTML as 
'text/html'. [2]

How much the 2009 version (which uses a XHTML 1.1. DOCTYPE, and is 
served as text/html) really widens the 2002 version of the "loophole", 
could be debated. However, considering all the XHTML syntax features 
that HTML5 is permitting, then IMHO both the 2002 version as well as 
the 2009 version put far too many restrictions of how HTML compatible 
XHTML should be authored. E.g. HTML5 permits "/>" for <LINK> elements - 
which is totally against Appendix C. (Fortunately, though, it is only a 
note ...)

Note, also, that XHTML 1.1. is based on Modularization of XHTML 
(XHTMLMOD) [3] and is: [4] 

 ]]essentially a reformulation of XHTML 1.0 Strict" [...] [legacy] 
facilities are available through modules defined in Modularization of 
XHTML, and document authors are free to define document types based 
upon XHTML 1.1 that use these facilities[[ 

Below I note how I personally would have amended the XHTML 1.1 doctype 
to include the legacy module. [5]

There is also at least one specification based on XHTMLMOD which 
explicitly states that documents conforming to that profile *may* be 
served as text/html, namely the XHTML Mobile Profile specification from 
2001. [6] (It is perhaps significant that is was written before the 
XHTML Media note - but then again, that note is, as you said, not a REC 

The W3C QA Team in 2005 advised the readers of A List Apart about how 
to use XHTMLMOD if you need to define your own document type - I am 
pretty sure that the W3C QA Team knew that it was speaking to a 
'text/html' audience! [7]

The point is that there is no MUST NOT against serving XHMTL 1.1. or 
XHTMLMOD as 'text/html'. E.g. XHTML 1.1. doesn't say anything about how 
it should be served. And the Legacy Module of XHTMLMOD is primarily 
useful in 'text/html' browsers. So there is absolutely no reason to be 
whether in shock or awe, if XHTML 1.1. is made even more compatible 
with 'text/html'. And there is also very little justification of 
'disregarding' XHTML 1.1. or XHTMLMOD.





[5] Here is how I would have added XHTMLMOD’s Legacy module:

    <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.1//EN" 
    [<!ENTITY % xhtml-legacy.mod
       PUBLIC "-//W3C//ELEMENTS XHTML Legacy Markup 1.0//EN"
"" >
    <?parser-hack ><!-- ?>]><!--><?!-->



leif halvard silli

Received on Friday, 12 March 2010 02:08:58 UTC