- From: Tim Berners-Lee <timbl@w3.org>
- Date: Sun, 28 May 2000 21:23:17 -0400
- To: "Liam Quin" <liam@holoweb.net>, <xml-uri@w3.org>
-----Original Message----- From: Liam Quin <liam@holoweb.net> To: xml-uri@w3.org <xml-uri@w3.org> Date: Saturday, May 27, 2000 3:08 AM Subject: Re: Namespaces and namespace names: a new synthesis? >Larry Masinter wrote: >[...] >> You have to define what it means for two namespaces to be "the same". >> There's no obvious, unique, equivalence relationship. The relation is, I think (but you have to swallow that there is a philosophy of the meaning of a message here ;-). Suppose a message M2 uses namespace n1 with URI u1, and is a valid representation to the receiver of the intent of the author, and a message M2, formed by taking M1 and changing it by only changing all occurrences of u1 to u2. If in all such cases M2 is also a valid representation to the receiver of the intent of the author then u1 and u2 identify equivalent namspaces. Liam said: >All of these problems stem from using a URI as a name for something >instead of idenftifying the location for something. No no no no no. An HTTP URI is a name, and there is a lookup function for it. There are not rules against reuse imposed globally, but there are as many levels of indirection as you like available. Please stop refering to it as a location....! >It *would* make sense to compare two documents that defined namespaces, >although no-one has invented such things yet. The syntax of a namespace can be defined by a schema langauge. There is of course more definitive information about a namespace, but so long as you realize that there will always be teh posisbility of encoding mroe information with better technology, then a schema is a good place to start. >It would also make sense to have a "metaschema" document that contained >publicly-agreed-upon names of namespaces. What, a central registry for all namespaces? [The centralization alarm goes off. ..] Think of each Filemaker pro database generating a new namespace and a new schema for it. >Finally, it would make sense to abandon the idea of having a URI in there >at all, and use a formally defined name, preferably one that does not >start with "http:" in any examples :-) So, rather than simply focuss this debate on the incomapatability between string comparisons and relative URIs, the "names vs addresses" can of wroms has been opened. >Lee > >-- >Liam Quin - Barefoot in Toronto - liam@holoweb.net - http://www.holoweb.net/ Tim whose desktop is now is a mess and is wondering whether he's missed any messages..and pauses to reboot.
Received on Sunday, 28 May 2000 21:21:56 UTC