xml-dev - Re: Question About Namespaces and DTDs

[Tim Bray <tbray@textuality.com>]

At 07:38 PM 25/07/00 -0400, Simon St.Laurent wrote:

>> What is more important is that the parties involved (ie the W3 gods, 
>>Working Group members, Namespaces Rec editors) understand that they did the 
>>wrong thing, and will try to not do it again.
>>
>>I think these people are aware of the problems with namespaces, but I don't 
>>know if they share my opinion of these problems as mistakes they've 
>>inflicted on us.
>
>I don't think they do, though sending them a bill for wasted time might
>help make that clear.  There are still people who'll claim that Namespaces
>is _almost_ perfect, though I think they're declining in number lately.

Well, I'm one of those people.  Namespaces do a very good job of what
they were designed to do - give elements and attributes globally unique
names that allow software to recognize and differentiate.   They have
been implemented by pretty well every interesting piece of XML software
in the world (hint: it's not hard) and in practice, in real-world
implementations, just work.

Yes, namespaces and DTDs don't get along very well.  At the end of the
day, the conclusion was that uniquifying names in a simple and clean way
was an important enough piece of the puzzle that that was an acceptable
cost.  Although, to be honest, nobody foresaw how far into excessive
overengineering schemas were going to veer... 

Anyhow, the notion that namespaces were foisted on a resisting populace
by a small group of insiders is totally silly.  There was tons of input and 
agonizing hand-wringing and meetings and emails and successive drafts etc 
etc ad nauseum.  And in some parts of the spec, the community consensus was 
questionable.  But the basic notion of using URIs to extend names and make 
them unique had overwhelming buy-in from almost everyone (except those who 
wanted to stay with architectural forms).  Lots and lots of alternatives 
were considered.

There's no requirement that you like the namespace system, and
it's certainly possible that there was a better way to have done it that
the community wasn't smart enough to cook up.  But please drop this notion
that there's consensus on the other side: that namespaces are broken in
obvious ways and that these errors are avoidable in future.  -Tim