- From: Ian Hickson <ian@hixie.ch>
- Date: Mon, 25 Sep 2006 01:46:56 +0000 (UTC)
On Sun, 24 Sep 2006, Anne van Kesteren wrote: > > On Thu, 21 Sep 2006 19:40:18 -0000, Ian Hickson <ian at hixie.ch> wrote: > > > When you have phrases like 'Change the insertion mode to "in > > > select".' it would be nice if "in select" was a pointer to the > > > relevant section. Currently "insertion mode" is a pointer and that's > > > hardly useful. > > > > The problem is the "in select" mode includes the start of the "in > > body" mode. e.g. whatever insertion mode you're in, if you find a > > <!DOCTYPE>, you have to handle it as described at the top of > > 8.2.4.3.7. But I suppose we could add the links anyway... What do you > > think? Do you really want the links, even though they'd be kind of > > technically incorrect? > > The "in select" mode seems to assert something else than what you say, > actually. It handles all cases itself already... Including "anything > else". Your confusion here is exactly what I'm worried about. The point is that in the main phase, whatever the insertion mode, you have to start each token's processing from the start of the main phase. Which handles the DOCTYPEs and certain character nodes, before looking at the insertion mode flag and handling the other things (including "anything else"). Insertion modes are not phases. They are all processed as part of the main phase process each time. -- Ian Hickson U+1047E )\._.,--....,'``. fL http://ln.hixie.ch/ U+263A /, _.. \ _\ ;`._ ,. Things that are impossible just take longer. `._.-(,_..'--(,_..'`-.;.'
Received on Sunday, 24 September 2006 18:46:56 UTC