- From: Ian Hickson <ian@hixie.ch>
- Date: Fri, 25 Jun 2004 13:04:43 +0000 (UTC)
On Thu, 24 Jun 2004, fantasai wrote: >>> >>># If a select element has a data attribute, it must be a URI that points to a >>># well-formed XML file whose root element is a select element in the >>># http://www.w3.org/1999/xhtml namespace. The MIME type must be an XML MIME >>># type [RFC3023], preferably application/xml. It should not be >>># application/xhtml+xml since the root element is not html. >>># >>># UAs must process this file if it has an XML MIME type [RFC3023], if it is a >>># well-formed XML file, and if the root element is the right root element in >>># the right namespace. If any of these conditions are not met, UAs must act as >>># if the attribute was not specified, although they may report the error to the >>># user. UAs are expected to correctly handle namespaces, so the file may use >>># prefixes, etc. >>> >>>These two paragraphs should be merged and >>> >>> UAs must process this file if it has an XML MIME type [RFC3023], if it is a >>> well-formed XML file, and if the root element is the right root element in >>> the right namespace. If any of these conditions are not met, UAs must act >>> >>>replaced with >>> >>> UAs must process this file if and only if these conditions are met. If they >>> are not met, UAs must act >>> >>>There's no reason to state the requirements twice. >> >> The problem is the requirements aren't _quite_ the same. In particular, >> authors SHOULD not use application/xhtml+xml, but UAs MUST process the >> file if it is (since it is an RFC3023-compliant type). >> >> How exactly would you phrase this? > > What I wrote. s/conditions/requirements/, maybe. If the file is sent as > application/xhtml+xml, it still meets the conditions set forth in that > paragraph, though it doesn't follow all the recommendations. Yeah, but that doesn't really seem as clear to me. > If you really want, you could change the paragraph to say that > It should not be >(but could be)< application/xhtml+xml And that, although not technically doing so, looks like it allows that MIME type, whereas I really want to disallow it as much as RFC3023 will let me (which is basically as much as a SHOULD NOT). The benefit of merging the two paragraphs IMHO doesn't outweigh the disadvantages that come from the added ambiguity, IMHO, even if that ambiguity is only in possible technically incorrect interpretations rather than as actual ambiguities. -- Ian Hickson U+1047E )\._.,--....,'``. fL http://ln.hixie.ch/ U+263A /, _.. \ _\ ;`._ ,. Things that are impossible just take longer. `._.-(,_..'--(,_..'`-.;.'
Received on Friday, 25 June 2004 06:04:43 UTC