Re: XBL Namespace uses the data: scheme

On Jun 30, 2006, at 10:56, Charles McCathieNevile wrote:
> Your first paragraph belies the second one - the single most  
> important feature of a namespace is that it is a unique identifier.

Correct, but mnemonic qualities of a string intended for human use  
can hardly be sufficiently stressed. As it currently stands, the only  
thing that beats the nsuri policy as example of what not to do here  
is UUID URNs.

> "I have received complaints" is a statement that applies to almost  
> any situation or permutation

Again, agreed on the form, but in this case there are many of us who  
have heard many times the very same complaint. It's bad enough that  
people looking for information on the W3C have to deal with  
randomised URIs for resources clearly intended for human use (humans  
certainly do use them), with namespaces they have to learn the same  
by heart.

Take a typical document I was working on earlier this week. I invite  
whoever believes that the current policy is good to time themselves  
linking the year to the tech for the namespaces that document is using:

   1999    xml
   2002    xlink
   2000    xhtml
   1999    xforms
   1998    svg
   2001    rdf
   1999    xml-events

I wouldn't even consider that document to be making heavy use of  
namespaces, there certainly are quite a few others from W3C that  
could be added to that list (in fact, the actual document contains a  
few more but they all follow a "Web of Humans" type of convention and  
so are unproblematic).

> The current data: URI is ludicrously non-mnemonic, and choosing a  
> more useful one could be achieved even with the constraint of  
> having to include two years and a colour name.

Right, I'm not super fond of the data: thingie but it's meant to be  
temporary.

> That said, it is useful. w3.org/2006/xbl is, IMHO memorable enough

I'm being dead serious when I say that having totally random years in  
namespace URIs is the only thing I ever found genuinely difficult  
with namespaces. If there had been a mnemonic assignment system  
instead of a machine-oriented one I'm fairly certain that a fair  
percentage of the complaints about namespaces would have vanished.

> - the difficulty I find in hand-writing namespaces is in fact  
> remembering whether the thing ends with a slash or not

Yeah that's a pain, albeit I find a much lesser one, notably given  
that there are far fewer options to explore when things don't work.

-- 
Robin Berjon
    Senior Research Scientist
    Expway, http://expway.com/

Received on Friday, 30 June 2006 11:00:06 UTC