- From: Larry Masinter <LMM@acm.org>
- Date: Wed, 1 Jan 2003 13:00:18 -0800
- To: <sandro@w3.org>
- Cc: <www-archive@w3.org>
Happy new year! > I'm having a moment of really wanting tdb URIs. I'm trying > to organize my notes and realizing about half the time I'm > writing a note about a document (web page, web page fragment, > technical paper, book) and about half the time I'm writing > about the subject of that document. Ugh. > > But I think I'd be happier with tdb not having dates. I just want > urn:tdb:http://www.w3.org/Consortium > and when I want to be bound to a date, I can do > urn:tdb:urn:duri:2002:http://www.w3.org/Consortium > There are two dates: there is the date of the resource and separately there is the date of the interpretation of the description. The date in 'tdb' is more important for the former than the latter; e.g., with a resource like "data:,the%20president%20of%20the%20united%20states" you still might mean different things on different dates. Normally in a literary work in natural language, both the reference itself as well as the referenced work are dated, e.g., a footnote in an article written in 1967 might talk about a "private communication" which itself had a date. With URIs, we want the references to be able to escape their context while retaining their meaning, so giving a date in the reference seems natural. > I understand the feeling of too much float in using tdb > without a date, but I also want a resolution mechanism, and > without a date resolution is easy. I think the resolution mechanism with the date is just as easy as the resolution mechanism without the date: just ignore the date if you want. Yes, the results might be incorrect if you ignore the date, but no more incorrect than they are if the date is not given. I could imagine suggesting that the date be omitted from tdb when it is completely redundant or unnecessary or when the continuum of interpretation is wanted, urn:tdb::http://www.w3.org/Consortium. > I'm also inclined to just step outside of URIs and use some > notation like *http://www.w3.org/Consortium for > urn:tdb:http://www.w3.org/Consortium, since I'm not actually going to > hand it to anything with the "*" or "urn:tdb" still there. (the "*" > is meant in vaguely the same sense as the *-prefix in C.) It's always possible to use other syntactic notations, and for any particular purpose, those are likely to be more concise or appropriate. You choose the 'URI' notation to fit into a framework that accepts URIs or that otherwise needs URI extensibility. Since some applications really wanted URIs for abstractions, it seemed like 'tdb' was the minimum syntax that would get what was needed. 'urn:tdb' adds 4 more characters, but at least puts you on notice that this is no ordinary URI scheme. Larry
Received on Wednesday, 1 January 2003 16:00:07 UTC