- From: Julian Reschke <julian.reschke@gmx.de>
- Date: Wed, 9 Oct 2002 10:40:00 +0200
- To: "Seth Osher" <seth@ipicorp.com>, "Webdav WG" <w3c-dist-auth@w3c.org>
> From: w3c-dist-auth-request@w3.org [mailto:w3c-dist-auth-request@w3.org]On > Behalf Of Seth Osher > Sent: Wednesday, October 09, 2002 5:54 AM > To: Webdav WG > Subject: RE: Interop issue: Proposal for fixing lock token provision > > > For example a long running folder copy, move or delete. > > If the operation takes long enough that locks might expire > during the operation, then the semantic might not even be > really enforceable. Particularly, it servers is incapable of checking the > locks in advance, or to do so might be expensive from a time > perspective (say an operation on a optical-archive, where the full set > of resources is only known on the media). ...in which case I'd say that the locks should have had a longer timeout in the first place. Why bother locking, if you're unsure that the timeout is long enough to protect the subsequent request? > In these cases you want the copy or move to perform "as much work > as possible" so you just want to provide a list of all your > current tokens. Really? Again -- if it's irrelevant whether the lock is still there, why was it made in the first place? > A second example is where a server provides multiple views of the same > data. A client may not know for certain what resources it has locked > and are going to participate in a copy, move or delete operation. > Rather than fully enumerating the set of resources to be operated on, > presenting "all-my-tokens" would be useful, if not critical. I don't understand this scenario. Could you please define what a "view" technically is on your system? Is it an additional binding to the same resource? Every lock that the client made has been made on a particular request URI -- and that's the URI that needs to be submitted alongside with the lock token (in the If header). > A third case is where a user wants the operation to succeed, even > if etags for the locked resources have changed (perhaps a shared-write > has updated the resource, changing its etag). In this case, we want to > present the lock, and don't want to worry about the etag. It might even Why do you submit the etag if you don't want it to be enforced? > have be deleted by the shared-write lock so the token is no longer valid. Why would the resource be deleted if you still have a shared-lock on it? > Perhaps it is possible to construct an If: header that handles these > cases, but a shorthand that avoids recursive introspection on the server > would be useful, particularly in low-bandwidth environments. As shown in yesterday's mail, it is possible. It's just syntactically different. -- <green/>bytes GmbH -- http://www.greenbytes.de -- tel:+492512807760
Received on Wednesday, 9 October 2002 04:40:29 UTC