- From: Ian Hickson <ian@hixie.ch>
- Date: Wed, 17 Mar 2010 23:55:49 +0000 (UTC)
On Sat, 20 Feb 2010, Nikita Popov wrote: > > I noticed, that mistyping the ending title tag causes a white page and a > title containing the whole source of the page (from the starting title > tag on) in firefox3.6 with enabled HTML5 Parsing Algorithm. > > As I haven't read the the HTML5 Parsing Algorithm section, I don't know > whether this is a misimplementation or is defined this way in the > algorithm. > > Obviously this cannot be the expected behavior. The second <title> tag > should be interpreted as a </title> tag. On Mon, 22 Feb 2010, Henri Sivonen wrote: > > The only type of recovery that doesn't introduce even more problems > would be first parsing until EOF and then rewinding to the <title> > element start and reparsing until the next <. Implementing the recovery > strategy from the previous sentence would make the parser vulnerable to > executability escalation attacks by forcing a premature EOF. (Granted, > the attacks are rather contrived.) > > Consider: > <title><title><script>alert("Attack code could go here!");</script> .... </title> ... EOF > vs. > <title><title><script>alert("Attack code could go here!");</script> ... EOF > > In practice, IE doesn't reliably recover from this class of authoring > error, so it's fair to believe that the Web doesn't depend on browsers > recovering from this situation. So far, I haven't seen bug reports about > this HTML5 behavior affecting pages on the Web. I have seen one report > about a non-IE-oriented intranet page (Mozilla's own!). This is indeed the expected behaviour. -- Ian Hickson U+1047E )\._.,--....,'``. fL http://ln.hixie.ch/ U+263A /, _.. \ _\ ;`._ ,. Things that are impossible just take longer. `._.-(,_..'--(,_..'`-.;.'
Received on Wednesday, 17 March 2010 16:55:49 UTC