xmlhack: Concern at new URI schemes for namespaces

Concern has arisen in the last month over the trend toward creation of new URI schemes for use as namespace identifiers in new XML applications.

Writing on the W3C's URI list, Tim Berners-Lee expressed surprise at WebDAV's use of the DAV: scheme:

The DAV spec really invents a totally new whole URI scheme just for a namespace? If there has to be a root of anything anywhere, then with whom will it be safe? The root of the URI space now become a free for all...

Earlier in the month, Dan Connolly had commented on the SyncML spec's use of the syncml: URI scheme.

I wonder why they used a new URI scheme for their XML namespace; what's wrong with http://syncml.org/2000/metinf ?

While inventing new URI schemes for namespace purposes certainly seems gratuitous, it is worth some investigation as to why this is happening--it may be that http: URIs for namespaces still cause discomfort and confusion, causing some to opt for a custom solution.

Re: Concern at new URI schemes for namespaces (Robin Berjon - 00:00, 28 Jan 2001)

There's one solution to this confusion: spread RDDL and sread it fast ! Once it is out there and starts being used widely enough, people will start wanting to use http instead of just about any garbage scheme that comes to their mind.

Re: Concern at new URI schemes for namespaces (James Marca - 04:57, 27 Jan 2001)

I am sure this topic is beaten to death in the relevant mailing lists, but I am not on those lists so I'll add my 2 bits here. As a newcomer and evangelist of XML in my department, I find it a bit odd to say the least that I am arbitrarily selecting our department web address and tacking on an ending to specify the namespace. The department sysadmin and my PhD advisor are soon going to notice this fact, and will probably ask who gave me permission to use that URL, and so on.

I will explain to them that the namespace doesn't point to anything, it just has to be unique. They will probably buy that argument, but they also might wonder if I am up to something or have somehow cracked the web server.

So perhaps the urge to invent new protocols is two-fold. First, the newcomer to namespaces has to become accustomed to the fact that the http://blah.blah.edu/things points to nothing (I notice that even the parser of this text presumes that http:// points to something and assigns it a link!).

Second the address that is used is typically something that is assigned by the local sys/web admin, and using it without their permission might inspire a turf battle. If had invented gpsdata://www.its.uci.edu, well, no one would ever mistake that for http:, and I'd have no explaining to do. Further, there is nothing to prevent me from inventing gpsdata://, so that path could easily be the path of least resistance. And if I was more egotistical, why not invent james:// as the root of all the namespaces that my code requires?

Hope this adds more signal than noise to the argument.

James