- From: Gabe Beged-Dov <begeddov@jfinity.com>
- Date: Wed, 08 Dec 1999 18:18:44 -0800
- To: Sergey Melnik <melnik@DB.Stanford.EDU>
- CC: RDF Interest Group <www-rdf-interest@w3.org>
Sergey Melnik wrote: > Furthermore, why is it undesirable to query "open" models? Cannot we > read from files open for writing? My mental model would be a journaling file system where all versions of the file are addressable. When you query, you are querying a particular version of the model/file. In a journaledor versioned file the ID is based on a timeline while in our discussions it is based on the content. In either case, you can always use an alias that gives you the "default" view of the file. This is also the case with regular files which may often have different versions while in "open" mode with the read going to a particular version. I think that updating of a model should be an explicit activity with distinct boundaries. Whether a model can be closed and reopened is a question of what granularity you want versions of a model to be visible at. We are talking about a raw model layer anyway though, and you could always layer these kind of policy layers that have distinct modes and different interfaces for different roles so I guess I wouldn't worry about it. > If say Y becomes an explicit name later on, the URI for Z remains > intact, but X changes. If X gets a name, both Y and Z remain intact. You and DanB have both mentioned noname resources turning into explicit resources but I don't see how this can happen at the model layer. This seems a much higher layer concept. Am I missing something? As to your algorithm, I can't comment on it as I don't grok it yet but since youre in my web of algorithm trust, I'm not too worried :-) Gabe
Received on Wednesday, 8 December 1999 21:34:16 UTC