On Thu, 25 Oct 2007, David Orchard wrote: > > Not sure why it's the opposite but I may be missing something. The > advantage of allowing extensibility with an extension processing model > of "ignore" is that a future version of HTML could do something with the > extra arguments in a compatible way with this version. Sure. But in the case we're talking about, we can't change what happens with the extra arguments, they _must_ be processed as described in the spec (concatenation) because it's common and widely interoperably implemented that way. > If multiple arguments are not allowed, then there can never be a > compatible evolution of HTML that allows multiple arguments. Why not? We disallowed <section> in HTML4, and allow it in HTML5, where's the problem? Similarly, all methods disallow more than their defined number of arguments, but we can add more arguments in future versions... > Exactly akin to HTML's longstanding practice of allow extra markup and > ignoring unknown markup. In general, I like the model of allowing extra > things and requiring they be (roughly speaking) ignored. You can disallow something but require that it be ignored; allowing it and requiring that it be ignored seems nonsensical to me. -- Ian Hickson U+1047E )\._.,--....,'``. fL http://ln.hixie.ch/ U+263A /, _.. \ _\ ;`._ ,. Things that are impossible just take longer. `._.-(,_..'--(,_..'`-.;.'Received on Thursday, 25 October 2007 23:06:22 GMT
This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:40:04 GMT