[whatwg] Sanctity of MIME types
Sam Ruby
rubys at intertwingly.net
Mon Dec 4 12:59:39 PST 2006
Here's a random half dozen examples, picked to show a bit of diversity:
http://beta.versiontracker.com/mac/osx/home-edu/updates.rss
http://city.piao.com.cn/rss.asp?85
http://feuerwehr-melle-de.server13031.isdg.de/index.php?id=199
http://hesten.innit.no/hru/rss.php?START=0&STOP=3
http://httablo.hu/pages/rss.php
http://skopjeclubbing.com.mk/rss_djart.asp
Independent of what the specs say *MUST* happen, I'd like people to
bring up one or more browsers with a URL from this list, and see if the
browser asked them if they wanted to subscribe. Subscribe is not a
normal feature associated with text/html, which is the Content-Type that
you will find for each.
The point is not to label these guys bozos (as I said in previous
messages, bozos outnumber you). But to get you to consider what
browsers can, and will, do.
In these days of GreaseMonkey and its brethren, the client is king.
- - -
Where does this leave HTML5? I am of the opinion that HTML5 should
describe a set of rules that a compliant HTML5 parser should follow.
The MIME and DOCTYPEs specified in the document should be
recommendations. Something outside of the parser may chose to dispatch
based on this information, but that's outside of the control of the
parser. IMHO, the parser itself shouldn't complain when it finds a
HTML4 DOCTYPE, or an XHTML2 DOCTYPE for that matter.
Of course, a lot more HTML4 documents would be valid HTML5 than XHTML 2
documents.
- Sam Ruby
More information about the whatwg
mailing list