[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: Re: [emergency] Unique Identifiers Discussion: Seehttp://taguri.org
Bob / All - I'd support adding the "cap:" URN scheme... with an eye toward the day when we'll have other, perhaps EDXL-based formats also in play. How hard/lengthy is the IANA registration process? Would it be feasible to get that into this revision? - Art At 1:53 PM -0500 3/10/05, Bob Wyman wrote: >To provide some context and background on the discussion during >today's call concerning unique identifiers, I would like to suggest >that CAP folk become familiar with the "tag" URI scheme. >Documentation can be found at: > > <http://taguri.org/>http://taguri.org/ > >Tag URI is currently documented as an Internet Draft; however, it >has been accepted for publishing as an RFC. When the RFC is issued, >the ID will become obsolete although the RFC will contain >substantially the same text as the current draft of the ID. The ID >is temporarily available as: ><http://taguri.org/07/draft-kindberg-tag-uri-07.html>http://taguri.org/07/draft-kindberg-tag-uri-07.html > >A sample "tag-uri" looks like: tag:hawke.org,2001-06-05:Taiko >This URI, and others which are assigned using the tag uri scheme, is >unique across all time. > >An approach similar to the Tag URI has been used by the Newspaper >industry in the definition of the NewsML URN's (URN = Uniform >Resource Name) that are used in identifying news content. For more >information on the NewsML URN scheme, see: ><http://ietf.org/rfc/rfc3085.txt>http://ietf.org/rfc/rfc3085.txt >A sample NewsML URN is: urn:newsml:iptc.org:20001006:NewsMLv1.0:1 > >Like Tag URI's, NewsML URN's are unique across all time. Both of >these schemes use the basic idea that a unique identifier is a >sequence of three parts: >1. A string which is "owned" at some specific time by the >creator of the identifier and is guaranteed not to be assigned or >owned by more than one administrative entity at any one time. (But, >whose ownership might change over time) >2. A timestamp that identifies a point in time or period of >time during which the creator of the identifier actually "owned" the >string. >3. Some value which is unique among all those used by this >particular creator in combination with 1 and 2 above. > >If CAP reference ids are constructed in the following manner: >(partial ABNF below): > >CAP-reference = sender "," sent "," identifier > >And, if sender is either a domain name or something like an email >address that incorporates a domain name, then Cap-references could >be guaranteed to be globally unique - across all time. > >Given the discussion above, I would suggest we might want to >register a CAP URN scheme with IANA. Such an URN might then have the >following form: > >CAP-URN = "cap:" sender "," sent "," identifier > > bob wyman > >
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]