- From: Larry Masinter <masinter@parc.xerox.com>
- Date: Tue, 25 Feb 1997 05:23:59 PST
- To: Steve Carter <SRCarter@novell.com>
- CC: w3c-dist-auth@w3.org
Steve Carter wrote: > > No, I don't feel that expediency is the issue. Clustering (granularization) is a common > method of handing shared things. In many operating systems the easiest way > to do this is to use byte range locking. It is a quick way of creating a shared > semaphore. The web is an unusual operating system, though. (The Internet is an unusual computer.) In Webdav, we've been pursuing using links and site maps to do clustering, and it seems that so far we can handle these issues by having separate resources with a relationship between them, and having the lock operation work on the separate URLs. It might be that in one implementation "http://server.dom/resource/bytes=1-12" is a byte range of "http://server.dom/resource", but we don't need to standardize on that, only the relationship between them. Clearly we have to address the container/contained relationships, that locking a chapter has some effect on the lock state of the entire book. If we can handle the relationship at the larger granularity, we probably should use the same mechanism for dealing with locks even with a finer grain. Larry
Received on Tuesday, 25 February 1997 09:24:05 UTC