Details
-
Brainstorming
-
Status: Reopened
-
Minor
-
Resolution: Unresolved
-
None
-
None
-
None
-
None
Description
I keep getting lost in the WebProxy code on what graphs are named what. It would be easiest if graphs were named by the URL they were fetched from. The expectation would then be that the graph named something is what I get when I do a curl -H "Accept: application/rdf+xml" url
So when running my localhost instance the graph from <http://bblfish.net/people/henry/card> should have that same name and not
<http://bblfish.net/people/henry/card.cache> or some other name.
local graphs can just have the url of the local server, eg on bblfish.net:8443
the admin graph would be named
<https://bblfish.net:8443/user/admin>
and not
<https://bblfish.net:8443/user/admin.graph>
There won't be any clashes between remote and local graph of course, since we are all in one namespace: urls.
The only thing I remember being needed is if one wanted a local graph to have additional information about a remote user. But that would be easy. The local information would be placed in a local graph such as
<https://bblfish.net:8443/user?id=$urlencode
{remoteid}>That makes clear where the responsibility for that URL lies - with bblfish.net, not with the remote server as it does in the {remoteid}
.cache notation. Of course it could also be at
<https://bblfish.net:84432/user/hash1231> or any other local url.
So how would one know what the local addition to a remote graph is? I think that is application specific anyway. It requires a function to map from remote graph to a local transformation for each app that requires such a thing.
Things can be made to be more complicated, but I think it helps if to start with these things are as transparent as possible.
I would like to implement this for the WebProxy and the WebID login. I just keep getting lost as I said with what graph is what.