- From: Roy T. Fielding <fielding@kiwi.ics.uci.edu>
- Date: Tue, 07 Jul 1998 18:42:17 -0700
- To: Patrik Fältström <paf@swip.net>
- cc: Larry Masinter <masinter@parc.xerox.com>, uri@Bunyip.Com
>I.e. the problem I see is that we use different criteria when talking about >hierarichal names, which in turn leads into the space of relative names. > >The criteria I have used is that not only two names are in the same >namespace (at a perticular moment) but also that the namespace can change >without any changes in the namespace specific name. If both of these >criteria are true, then we can use the generic URI syntax, relative URI's >etc etc. > >For the second criteria to be true, it is a must that the naming of each >object in the namespace be made in the namespace itself and independent of >other namespaces (or subnamespaces). It is not necessary for all names in a namespace to be hierarchical to justify a hierarchical namespace. After all, a nonhierarchical name just never uses the hierarchical bits. For example, all of the phone numbers in California are hierarchical. That is an awful lot of phones. Two years ago my office prefix (the three numbers after area code) changed. This year, my office area code changed. In neither case was the individual phone number assigned to my office "reassigned" by any central authority -- they just changed the switch. Personally, I don't particularly care whether phone numbers are defined as explicitly hierarchical or not -- my point was that this is a better solution in general than using separate naming schemes for local vs non-local phone numbers. An alternative would be something equivalent to the "+" country code indicator. The reason I don't care is because a phone number is unlikely to occur as a base URI in Internet-based exchanges, even though it is possible with XML, so whether or not it follows the generic syntax is just a question of user convenience. ....Roy
Received on Tuesday, 7 July 1998 21:53:30 UTC