Wait a minute. I remember asking about this. Ian actually advocates against using application/xhtml+xml for XHTML pages; instead, he supports the use of text/xml. And, as he points out in the appendix and Mark Pilgrim does in his article, UAs (browsers to the rest of us) that support XHTML sent as application/xhtml+xml all support XHTML sent as text/xml too. As an added bonus, this means that TrackBack code can now be validly inserted into the file instead of hacked in inside comments.
Off to update my templates and CSS, and track down those nasty character entity references.
[Later: D’oh! First post in valid-XHTML-land and it gets screwed up by a macro expansion inside an attribute. Worked around in anticipation of a fix.]
[Even later: Following Ian’s comment, I’m back to the original application/xhtml+xml. My thoughts on text/xml allowing TrackBack inline actually still apply. And I’m still trying to figure out Pingback for Movable Type, based on Stuart Langridge’s code. Perhaps Stumbling
would have been a better title.]
I used to advocate against it. As of reading RFC3023, I have changed my mind. See http://ln.hixie.ch/?start=1036767231&count=1
And now every character entity reference (other than the minimal set defined for XML) has been replaced with its numeric equivalent. MT Macros and
grep -r ‘&[a-zA-Z]’ ${BLOGPATH} | grep -v macros | grep -v ‘"’ | grep -v ‘&’ | grep -v ‘>’ | grep -v ‘<’ | grep -v ‘'’
to the rescue again.