- From: Larry Masinter <masinter@parc.xerox.com>
- Date: Tue, 30 Jun 1998 00:09:37 PDT
- To: "Patrik Fältström" <paf@swip.net>, "Vaha-Sipila Antti (NMP)" <antti.vaha-sipila@nmp.nokia.com>
- Cc: <uri@Bunyip.Com>, <eva@nic-se.se>
> >From my point of view, what I have used as a criteria is that IF one have a > namespace, which is well-defined, as a hostname, and bound to that some > names, AND that namespace can change name without changing the names inside > that namespace, then we have a hierarchy that works. > > For example, the URL's to files on a host in the http scheme do exist can > be formed like: > > http://foo/a > http://foo/b > http://foo/c > > If it is the case that the hostname changes from foo to bar, the > hierarchies do work still: > > http://bar/a > http://bar/b > http://bar/c > > This is definitely not the case with phone numbers as they are allocated > individually, and when the "base" changes, each one of the numbers inside > that country- or areacode have to be changed individually. It's not actually the case when a web server is changed, either. When www.att.com changed and there was www.lucent.com and www.bell-labs.com, some of the URLs changed, some stayed, and others disappeared. But it didn't mean that the hierarchy was useless to start with, just because there is an instance of renaming where it doesn't work. I don't think it is necessary for a hierarchical relationship to have a longer lifetime than the lifetime of the validity of the names that are asserted to be in the hierarchical relationship. If telephone numbers have a hierarchical relationship NOW (local relative to base), then it is useful to have that hierarchical relationship in the naming scheme. Similarly, if the URNs for the RFC series have a useful hierarchical relationship (namely, the relationship of RFC 812 to "the RFC series"), then it is useful to have the name of RFC 812 be seen as having a hierarchical (subsidiary) relationship to the name of the RFC series. The names of the RFCs are location independent. Yes, the 'same' document might have a different name at some point (STD 11), but the assertion of hierarchical relationship is not between the elements themselves but between the NAMES of the elements. My fax telephone number is: a) +1 650 812 4333 internationally b) 812 4333 in Palo Alto c) 8*923-4333 inside Xerox d) 4333 inside PARC and there is a hierarchical relationship between the forms, a > b > d and c > d, which can usefully be expressed tel://1/650/812/4333 = 4333 within tel://1/650/812/ 812/4333 within tel://1/650/ tel://xerox/923/4333 = 4333 within tel://xerox/923 823-4333 within tel://xerox where '4333' is relative to the two different 'base' locations of 'tel://1/650/812' and 'tel://xerox/923/'. All of the talk about renumbering is irrelevant. When things change, they usually change in unpredictable ways.
Received on Tuesday, 30 June 1998 03:14:08 UTC