Re: unlock question

At 4:37 PM -0500 3/2/01, Jason Crawford wrote:
>  > 1. Does UNLOCK require IF header specifying applicable write lock token?
>I've tentitively added this as...UNLOCK_NEEDS_IF_HEADER
>I believe that the current behavior is appropriate in that the only thing
>the server needs to verify is that the client is aware of that lock and the
>requestor is the principal that owns it.  If the client is explicitly
>requesting that the lock be destroyed, the client obviously knows about the
>lock and it's redundant to ask them to provide it in a second header.

I agree that the current behavior is fine when the lock that would go 
in the IF header is the one that is in the LOCK-TOKEN header.

>   I
>think we need to treat the IF header not as a way for the server to confirm
>that the client knows about a lock, but (as documented) a way for a client
>to put a precondition on an request.   If you look at the documentation for
>the IF header, this is the only thing it talks about.  And if the server
>needs to verify that a client knows about a lock, it should be adequate for
>the client to mention the lock in a any documented headers that we
>enumerate to verify that the client knows about the lock.

I agree.  The issue is when there is more than one lock that applies 
to the resource.  Then I believe that the write lock needs to be 
specified in the IF header when an unlock is done on one of the other 
kinds of locks.

In general I find the spec weak around the requirements on 
non-exclusive lock handling.

>  > 2. UNLOCK should return 412 if valid lock token is not supplied as
>I've tentatively added this as...UNLOCK_WITHOUT_GOOD_TOKEN

Thanks!  Is there a more current version of the issues document than 
the one posted at ?  That 
one shows a last update of 7/2000 and there was another issue I had 
raised in September that I wasn't sure had made it on.

Daniel Brotsky, Adobe Systems
tel 408-536-4150, pager 888-461-0237
2-way pager email: <>

Received on Friday, 2 March 2001 18:52:50 UTC