Re: URN UUID question

[Peter Saint-Andre <[e-mail]>]

It sounds to me as if you might want to use the 'tag' URI scheme:

https://tools.ietf.org/html/rfc4151

Peter

On 3/11/14, 1:31 PM, Joel Kalvesmaki wrote:
> Thanks for the suggestion. I had considered that option, but the string
> must conform to XML schema's <anyURI>,[1] to suit the purposes I
> mentioned (RDF, etc.). And no <anyURI> may begin with a numeral. It
> would also be important to have a scheme that any human or computer
> would recognize immediately as a URN.
>
> I had also considered the following (where NN = a suitable
> project-specific abbreviation or dt = date-time, as a project-agnostic
> urn scheme):
>
> urn:uuid-NN:f60330fd-1900-44ac-a825-__de70074e142e::2014-02-07Z
> urn:uuid-dt:f60330fd-1900-44ac-a825-__de70074e142e::2014-02-07Z
>
> But would either one require registration with IANA? If so, what what's
> the process and is it worthwhile? I have heard of at least one project
> that has simply coined its own urn:NNN: scheme on its own, but I do not
> fully understand what consequences face anyone setting out on that
> direction.
>
> One other possibility occurs to me, that of treating the time-date stamp
> as a fragment identifier:
> urn:uuid:f60330fd-1900-44ac-a825-__de70074e142e#2014-02-07Z
>
> But I suspect that would violate the RFC 4122 definition, no?
>
> Best wishes,
>
> jk
>
> [1] http://www.w3.org/TR/xmlschema-2/#anyURI
>
>
> On Tue, Mar 11, 2014 at 1:24 PM, Peter Saint-Andre <[e-mail]
> <mailto:[e-mail]>> wrote:
>
>     On 3/4/14, 8:32 AM, Joel Kalvesmaki wrote:
>
>         I am developing an XML data model that requires users to name
>         versions
>         of a document. Each version's name should be unique, but
>         patterned to
>         allow anyone (human or computer) to associate it with the names
>         of other
>         versions of that document and to place it in chronological
>         sequence. The
>         name of each version must be a single string, specifically a
>         IRI/URI (to
>         facilitate, among other things, straightforward declarations in
>         RDF). It
>         should not be split into two elements. Naming must be as
>         decentralized
>         as possible.
>
>         My favored scheme for naming these entities would concatenate a UUID
>         (any style), a middle delimiter, and an ISO date/dateTime, e.g.,
>
>         urn:uuid:f60330fd-1900-44ac-__a825-de70074e142e::2014-02-07Z
>         urn:uuid:f60330fd-1900-44ac-__a825-de70074e142e::2014-02-__28T00:20:58.3Z
>
>         Would it be misleading to begin such a string with "urn:uuid:"
>         and if
>         so, what are the alternative best practices?
>
>         Perhaps there already exists a urn scheme that does what I intend?
>
>         Are there any other issues I should consider before adopting a
>         naming
>         scheme like this?
>
>
>     As far as I can see, you don't really need or necessarily want the
>     "urn:uuid:" string at the front. Why not things like this?
>
>     f60330fd-1900-44ac-a825-__de70074e142e::2014-02-07Z
>     f60330fd-1900-44ac-a825-__de70074e142e::2014-02-28T00:__20:58.3Z
>
>     Peter
>