- From: <ccjason@us.ibm.com>
- Date: Tue, 17 Aug 1999 11:39:50 -0400
- To: "Slein, Judith A" <JSlein@crt.xerox.com>
- cc: "'WebDAV'" <w3c-dist-auth@w3.org>
Judith, > Two people have now taken a careful look at the lock / MOVE scenarios. At least three people. You overlooked my note. <JS> Kevin's intuitions are consistent: It is always the lock at the destination that survives after a MOVE, whether that lock is inherited from a collection or is directly on the resource at the destination. <JS> Agreed. They seem consistant and your interpretation of what he's said sounds right. As my note pointed out, I think he had some trouble on whether lock tokens are required though in scenarios 9/10. It seemed to be based on his assumption that changing the membership of a collection isn't an operation on the collection. Just on the resource already at the URI. In subsequent notes in response to Jim, Kevin seemed to recognize the role of the parent though. <JS> Jim Amsden's intuitions about the scenarios were different: No singleton lock ever survives a MOVE, but a resource that is moved into a locked collection does inherit the collection's lock. </JS> Just for the sake of consistancy, I prefer the singleton and depth locks to be treated the same. Just as Kevin apparently does. <JS> Cases 9 and 10 poke at the complexities added by multiple URI mappings to the same resource. Kevin's intuitions say that there is really no added difficulty about resolving these cases. It's the resource that gets locked, as rfc 2518 says, and it's still true that any lock, inherited or singleton, at the destination, survives the move. </JS> I disagree somewhat with "it's the resource that gets locked" as sumarizing Kevin's interpretation. If the resouce moves and doesn't carry the lock with it at times, then I don't think it's correct to say it's the resource that is locked. At least not without further explanation. That's what my note pointed out. But yes, if a resource is locked at one mapping, it is locked at all. But only the original URI mapping is "protected"... using the AdvColl definition of "protected". At least that was my proposal... based on AdvColl discussions. J.
Received on Tuesday, 17 August 1999 11:42:02 UTC