RE: Agenda of 19 September 2006 TAG teleconference

Ed Rice writes:

> I've reviewed the document in detail, and I'm comfortable that all 
> my concerns have been addressed. 

Terrific.  Particularly coming at a time when I know you're busy, the 
quick and positive review is much appreciated.  I still plan to suggest we 
let it sit for another week or so in case Stuart or others want to weigh 
in, but process wise I think we have the formal decision we need to 
publish at any time.

> I still have a problem with the finding that says a URI that ends in
> .xml cannot be assumed to be an .xml file,

I know you have that concern, and the finding does take the other 
position.  It was among the reasons I called it out explicitly.

> but I understand that's another finding and that's not what we're 
> reviewing in this case. (you reference it so I have to put my 2c in)... 

Well, at the risk of snatching defeat from the jaws of victory, I think it 
is this finding that takes a stand.  My impression is that other TAG 
members including me feel fairly strongly that, in general, a suffix like 
.xml does not ensure that a file is of any particular media type.   The 
finding does point out the possibility that a browser would issue a 
warning in case of a mismatch between suffix and media type, because 
consistent use of a suffix like this is common practice, but it is not an 
error IMO.  While intentionally publishing a non-XML file with .xml may be 
bad practice, I don't think there's a normative spec we can point to that 
says it's an error (and I personally don't think it should be). To make it 
an error I think we would have to write some complicated specifications 
that would, for example, list the media types that could legally be 
returned for a given suffix like .html.  We'd have to update that 
everytime something new came along like application/xhtml+html, a media 
type which did not come into being until long after .html came into common 
use.  Furthermore, this would at least in principle be a breaking change, 
since as of today it would be legal for a web site to source text/plain 

As you know, we do have a formal TAG finding [1] that says that media 
types take precedence, so even if we did document all the legal media 
types that "match" *.html, a mismatch could only result in an error.  It 
would still have to be prohibited to honor the suffix in preference to the 
media type.

So, I think we've done the right thing, and I understand that you are and 
have been uncomfortable with that decision.  As much as I'd like to agree 
that "that's another finding", my honest opinion is that it's in this 
finding.  I hope you will continue to support publication, but if you feel 
more discussion is needed then we should delay after all. This is an 
important point and quite central to the finding.  I think we need to get 
it right.

Thank you again for the quick and careful review.



Noah Mendelsohn 
IBM Corporation
One Rogers Street
Cambridge, MA 02142

Received on Monday, 18 September 2006 23:09:42 UTC