|
Posted by Henri Sivonen on May 6, 2006, 4:19 am
Please log in for more thread options
> Henri Sivonen wrote:
> > What's my wrong assumption?
>
> That browser indeed treats quoted part of DTD as a unit (a la opaque
> strings in namespace declarations). Sorry if I'm wrong and it was not
> your assumption.
Gecko and WebKit indeed extract the public id as a string, fold it to
lower case and match the resulting string as an opaque string against a
list of known lowercased quirky public ids and almost standards mode
public ids. Like I said, I have not seen the source of IE, so I'm
refraining from claiming to know how exactly it does what it does.
> I see... Ignorance is the bless ;-)
ITYM bliss. ;-)
> > Homegrown DTDs for XML are legitimate for XML (but still arguably a bad
> > idea on the Web). It is not so clear whether homegrown DTDs are
> > appropriate for text/html.
>
> Proprietary DTD's are fully OK for XML, thus for XML+XSL transformers.
DTDs on the Web are a bad idea, because processing them is optional and
DTDs cause infoset augmentation, so the infoset reported to the
application may be different depending on whether the DTD was processed
or not.
> That is pretty close to how Windows Vista file management will work
Eh?
> > > > http://ln.hixie.ch/?start=1144794177&count=1
> >
> > > But I would guess that the author simply mis-interpreted the
> > > rumors about hackers attacks using Content-Type tricks.
> >
> > He wasn't going by rumors. He has actually worked for Netscape and Opera
> > and also followed the bug database of Safari.
>
> Then his statement gets really strange - especially when anyone can
> prove it wrong.
He said "largely ignore". Your example is one of the cases not covered
by "largely".
--
Henri Sivonen
hsivonen@iki.fi
http://hsivonen.iki.fi/
| 
XML Sitemap