- From: Ruben Verborgh <ruben.verborgh@ugent.be>
- Date: Fri, 9 Oct 2015 10:58:27 +0200
- To: Tom Johnson <tom@dp.la>
- Cc: Markus Lanthaler <markus.lanthaler@gmx.net>, Hydra <public-hydra@w3.org>
Hi Tom, > My sense is that the semantics decided on were really the only reasonable choice. Actually, I'd say the opposite. Now it's impossible (without adding extra things) to refer to graphs. That's most unreasonable. > Intuitively, we may want to say that a graph name denotes the graph; but that > stipulation is not required for many (any?) use cases How would we not want to be able to refer to a graph? It's the Resource Description Framework, but we can't describe one of its basic concepts :'( > - All allowable universes of resources must contain all the graphs in a dataset. That doesn't seem a problem to me. > - Certain IRIs must map to those graphs, simply by virtue of their position in the > dataset's quads; no other IRIs exhibit similar behavior. It's not "simply by virtue of their position". If something is a car, it is a car, not because I say ":x a :Car". If something is a graph, it is a graph, not because I say ":a :b :c :x". > Concretely, it rules out many interpretations that are in common use: e.g. the > pattern of grouping statements relevant to a resource <a> in a graph named > <a>. It's up to people writing such grouping statements whether or not they want to confuse a thing and its description. > In short, it would be Range-14 all over again: "wait, that can't be Moby Dick, it's > clearly just the _graph about Moby Dick_." It's worse than httpRange-14. At least in that case, you can refer to things. Now, we can't refer to graphs (without special mechanism). Best, Ruben
Received on Friday, 9 October 2015 08:58:56 UTC