Re: RIM and RS Issue: Identifier Usage (UDDI and URN)

[Michael Joya <mike.joya@xmlglobal.com>]

Len,

  A UUID is easy to check for validity - it just has to be 128 bits long and follow a certain computation pattern. Urn's have to be checked for authentication of the namespace. Unless there's some governing body administrating those namespaces, people will inevitably clash over urn prefixes. (urn:etoys:....)

  In your argument you claim that urns are semantically meaningful and that UUID's are not. I would argue the reverse. With a URN, a company is inventing its own semantics. With a UUID, these semantics are set in stone. Let's compare the two, in terms of constituent parts:

UUID: <mac address> + <timestamp> + <entropy>
URN: "urn" + <arbitrary namespace> + <arbitrary semantics>

  Any valid UUID can be broken down into well-understood meaningful constituent parts. Any URN cannot; especially if people are inventing their own meanings at every turn.
  
  As to whether or not a URN is more friendly than a url, I'd say only in theory. The simplicity of the urn spec won't stop people from embedding their own id semantics (possibly even UUID!) into the third part of the URN.

ie: urn:my.company:this-is_the#383749#1905-Numbering#format_I_USE_23991.dtd

  The above is a valid URN and is only human friendly to the person that invented it.

  As far as "remembering" ids goes. People seldom ever remember ids anyway. I do the same thing everyone else does when I need to remember a resource identifier - cut'n'paste it somewhere or mail it to myself. I'm sure that good registry clients will also have histories and bookmarks just like modern web browsers do.



-- 
// Michael Joya
// XML Global Research and Development
// 1818 Cornwall Ave. Suite 9
// Vancouver, Canada
// 604-717-1100x230