On Wed, 6 Dec 2006, Cyril Concolato wrote: > > In Section 4.5, the specification says: > "If this causes a loop [...] then the user agent must only apply each binding > in the chain once" This would look like a erroneous > content. Why not considering the binding in error when there is a loop ? I > think this would avoid weird cases.
Which binding would be the one in error? It seems easier for implementations to attach all the bindings until a loop is detected than to try to detect a loop first. At the end of the day, though, it doesn't really matter what is done here, so long as it is interoperable. -- Ian Hickson U+1047E )\._.,--....,'``. fL http://ln.hixie.ch/ U+263A /, _.. \ _\ ;`._ ,. Things that are impossible just take longer. `._.-(,_..'--(,_..'`-.;.'
