[Standards] [Fwd: Re: Namespace well-formedness and RFC3920bis backwards compatibility]

Dave Cridland dave at cridland.net
Tue Sep 23 19:10:37 UTC 2008

On Fri Sep 19 20:00:52 2008, Stephan Maka wrote:
> Dave Cridland <dave at cridland.net> wrote:
> > Given that ejabberd apparently cannot do this easily, [...]
> https://support.process-one.net/browse/EJAB-680
I'm only repeating what I read here, in the case of ejabberd.

> Under which circumstances are inner elements not parsed?
Whenever we can avoid it. Why would we want to parse XML, aside from  
a need to check for namespace well-formedness?

> It is a big problem that a lot of clients base on standard XML  
> libraries
> which in the case of an unbound namespace prefix choke, report an  
> error
> and entirely refuse to continue parsing.
So either:

a) Reset the parser, reinjecting the (saved) stream open,  
string-match past the offending stanza, and then continue, or:
b) Disable the namespace processing of the offending parser and do it  
yourself, or:
c) Change the parser for one that is designed to cope robustly with  
bad XML.

> > And given that circumstance, it seems to me that requiring  
> servers to always
> > detect and reject bad prefixes is distinctly onerous.
> Which is bit orthognal to XMPP's philosophy of complexity on the
> server-side and valid XML everywhere.
Orthogonal... I do not think it means what you think it means. :-)

Nor does valid, apparently, either - I think you mean namespace  
well-formed. Unless you really intend to imply that servers have to  
do schema-checking too.

Orthogonal would mean that my suggestion was unrelated to, and had no  
impact on, the philosophy. I think you mean "counter to", or  
"contradictory to". Either of those is a valid opinion, and one I  
essentially agree with, although I'd see your XMPP philosophy and  
raise you "be liberal in what you accept, and conservative in what  
you send", which implies that whilst it's nice if servers don't send  
unbound namespaces, clients and servers alike have to deal with them  

> I agree. Unless all XMPP servers block invalid stanzas from clients  
> it
> is bad that s2s links will be shut down upon encountering an unbound
> namespace prefix. Having one bad client will affect other users on  
> the
> same server.
And you'll upgrade every server everywhere how?

Dave Cridland - mailto:dave at cridland.net - xmpp:dwd at dave.cridland.net
  - acap://acap.dave.cridland.net/byowner/user/dwd/bookmarks/
  - http://dave.cridland.net/
Infotrope Polymer - ACAP, IMAP, ESMTP, and Lemonade

More information about the Standards mailing list