- From: Mark Baker <distobj@acm.org>
- Date: Wed, 23 Aug 2006 12:08:37 -0400
- To: "Robin Berjon" <robin.berjon@expway.fr>
- Cc: public-appformats@w3.org
On 8/23/06, Robin Berjon <robin.berjon@expway.fr> wrote: > > On Aug 23, 2006, at 15:30, Lachlan Hunt wrote: > > An document with <xbl xmlns="http://.../xbl"> as its root element > > will already provide that functionality much more reliably than the > > MIME type. It's the namespace that matters in XML, not the MIME type. > > Which reopens the discussion about whether peeking inside is good or > not, etc. External identification can be useful. Right. But more than "useful", I'd say. It's the only way I know of to separate the concerns of document data and metadata. Conflating the two, as assuming the namespace of the root element is special does, guarantees a) security problems (the usual ones inherrent in "sniffing"), b) performance problems (if the payload is encrypted, say), c) incompatibility with deployed tools and software, and d) prevents XML schema designers from using the root element as they see fit (as XSLT and RDF/XML do). But I admit that this practice is becoming widespread, so I'm not going to argue that we outlaw it. Instead I'll just argue that we need to provide the tools necessary for those that wish to work within the existing constraints of Web architecture. > I think that every single damn WG that's defined an XML syntax of > some form (and in some cases WGs that haven't defined any) goes > through this dance. Yup! 8-) Mark.
Received on Wednesday, 23 August 2006 16:10:03 UTC