David Nesting wrote:
On Tue, May 03, 2005 at 07:42:55PM +0200, Arve Bersvendsen wrote:
1) Change the attribute value for the rel from "alternate" to "feed", or some similar wording. A feed is not always an alternate of the HTML document in which it occurs. Suggested replacement text:
I'm all for avoiding pigeonholing like that. But what about the case when a feed IS an alternate version? HTML says to use "alternate", this says to use "feed". Which takes precedence?
Neither. You can put both.
Plus, "feed" is kind of application-specific. What about "related"?
A link, by its very existence, is indicating that the two documents are related. rel="related" is redundant.
Consider also that HTML allows alternate stylesheets to be identified with the relationships "alternate" and "stylesheet". HTML overrides the semantics of "alternate" in this case (since the stylesheet isn't an alternate version of the page, it's just an alternate stylesheet).
This is recognized as a mistake on the part of the HTML 4.01 spec authors. The 'rel' attribute takes a space-separated *list* of rel values: this exception aside, one value does not modify the other.
Who's to say we can't overload it a little for this case?
You are not writing the HTML 4.01 spec, you're writing an autodiscovery spec that takes advantage of the syntax *and semantics* given in HTML 4. Your specification should be consistent with HTML 4, not contradictory to it.
~fantasai
