[Developers] Cactus RDF: URIs
dylan.stark at gmail.com
Thu Sep 7 12:07:00 CDT 2006
On 9/7/06, Thomas Radke <tradke at aei.mpg.de> wrote:
> Erik Schnetter wrote:
> > On Sep 7, 2006, at 11:03:11, Dylan Stark wrote:
> >> Hi,
> >> I would like to find out what a good naming convention would be for
> >> giving URIs to the different entities we will be collecting the
> >> metadata on, such as thorns and parameters. I know that in the RDF we
> >> can talk about the concept of "a thorn with a name", but it could be
> >> beneficial to have a URI for that concept. An example might be
> >> http://www.cactuscode.org/thorn/WaveToy.
> >> The statement "ParameterFile activates <http://.../WaveToy>" would be
> >> equivalent to "ParameterFile activates the Thorn with name 'WaveToy'.
> > In the past, people have copied thorns from one repository to another,
> > creating two thorns with the name name. While this is bad practice, it
> > has happened and will probably happen again. An additional qualifier
> > is needed.
> > Cactus already has the concept of arrangements. What about
> > http://www.cactuscode.org/rdf/thorn/CactusWave/WaveToy
> > instead? The "rdf" part would make sure that it does not conflict with
> > any web pages that people may create, and "CactusWave" is the
> > arrangement name. So far, arrangement names have been unique.
> > Of course, thorns change with time, so even that identifier does not
> > have a static meaning.
> The entity (eg. 'WaveToy') that we actually want to describe is not some
> anonymous entity but a very concrete one, namely the one that's being
> used in a specific simulation. As Erik says, many Cactus users may use a
> thorn called "WaveToy" but each one could have a different
> implementation; there are no thorn signatures to uniquely identify
> thorns. So the URI referencing that entity must somehow indicate that it
> belongs to a specific simulation.
> Therefore we decided for the Cactus metadata management service
> developed in AstroGrid that the URI prefix (ie. the unique namespace for
> all metadata about that simulation) contains a unique simulation ID.
> After that prefix, all other metadata is listed, basically following the
> hierarchical structure you suggested (eg. "ThornList/<thorn>",
> "<thorn>/<parameter>", etc.).
So that answers my versioning question :). I'll take a look at what is
stored in your info service to get a better idea. Thanks.
> Cheers, Thomas.
> Developers mailing list
> Developers at cactuscode.org
More information about the Developers