[standards-jig] Namespaces

Dave Smith dizzyd at jabber.org
Wed Aug 7 14:57:07 UTC 2002


On Wednesday, August 7, 2002, at 08:42 , Mike Lin wrote:

>> Yes, I do agree with Rob's assessment that prefixes should apply on a
>> per document basis (as opposed to a per stream basis).
>
> If both ends speak whatever prefixes they want, I'm really confused as
> to what we are actually gaining by doing this. The same two namespaces
> are always used at the stream/packet level, and AFAIK the server will
> refuse to route anything it doesn't recognize at the stream/packet
> level. So what is _actually_ wrong with fixing a set of prefixes as well
> as a set of URIs?
>
I don't follow. Can you elaborate with an example?

  As an aside, I will say that prefixes are arbitrary and not 
important -- it's the URI that is meaningful. Applications which work 
with namespaces (correctly) should not be concerned or even aware of the 
prefixing.

> I'm not sure I buy arguments about needing to do this in order to be
> "XML correct". We're already using XML in what I would call an extremely
> incorrect manner, as a wire protocol rather than a document format. So
> long as that is the case, I am sort of squeamishly starting to think
> that there is really nothing wrong with having an agreed-upon set of
> namespace declarations at the c2s stream level, so long as we are using
> this "streaming document" paradigm.
>
Oy. The streaming vs. document argument is not relevant here. What _is_ 
relevant is that I should be able to compose an XML fragment with the 
appropriate namespaces and not have to worry about my serializer picking 
the "correct" prefix for a URI -- I should not be required to always use 
the prefix "stream" for the http://etherx.jabber.org/streams namespace.

Diz




More information about the Standards mailing list