- From: Ian Hickson <ian@hixie.ch>
- Date: Fri, 21 Jul 2006 03:55:36 +0000 (UTC)
On Thu, 20 Jul 2006, Stewart Brodie wrote: > > I'm seeing inconsistent behaviour in my test pages due to an odd transition > in the "before head" state. Consider these documents: one completely empty > (i.e. the first token delivered is the end-of-file token); another with just > "<html>" as its content, the third with content "<html></html>"). These > generate very different trees: > > Empty content: <html> <head> </head> <body> </body> </html> > <html> : <html> <head> </head> <body> </body> </html> > <html></html>: <html> </html> > > Surely all three should build to the same thing? Or is this another example > of well-formedness overriding the implicit element generation? The third > tree's brevity is caused because the </html> tag in "before head" is just > causing a switch to the trailing end phase. It could just follow the > "anything else" case (and that would require "in body" to handle </HTML> too > (to act as if an eng tag for BODY had been seen and then reprocess the > current token - just like an end-of-file token). You are correct. Fixed. -- Ian Hickson U+1047E )\._.,--....,'``. fL http://ln.hixie.ch/ U+263A /, _.. \ _\ ;`._ ,. Things that are impossible just take longer. `._.-(,_..'--(,_..'`-.;.'
Received on Thursday, 20 July 2006 20:55:36 UTC