On Wed, 1 Feb 2006, Lachlan Hunt wrote: > > > > Since that would be somewhat hard to define neatly, I've instead just > > added a requirement at the end of the tokeniser to cover this. > > | When an end tag token is emitted with attributes, that is an easy > | parse error. > > Wouldn't it be easier to add that requirement either before the > attribute is created or when the end tag is omitted from the the > attribute name/value states?
Well, it wouldn't be as easy as that, no. Might still be relatively easy to do, but it would not be just one line somewhere. > Is it expected that when an end tag token is emitted, it will include > all the attributes as well? I would have expected that the attributes > would be dropped during tokenisation, rather than just being ignored > during the tree construction phase. End tags are currently defined as having attributes. -- Ian Hickson U+1047E )\._.,--....,'``. fL http://ln.hixie.ch/ U+263A /, _.. \ _\ ;`._ ,. Things that are impossible just take longer. `._.-(,_..'--(,_..'`-.;.'
