Re: Moving a label

From: Geoffrey M. Clemm (geoffrey.clemm@rational.com)
Date: Tue, Jul 04 2000

  • Next message: Geoffrey M. Clemm: "Re: Moving a label"

    Date: Tue, 4 Jul 2000 08:17:30 -0400 (EDT)
    Message-Id: <200007041217.IAA03826@tantalum.atria.com>
    From: "Geoffrey M. Clemm" <geoffrey.clemm@rational.com>
    To: ietf-dav-versioning@w3.org
    Subject: Re: Moving a label
    
    
       From: Jim Whitehead <ejw@ics.uci.edu>
    
       > The semantics for LABEL have changed somewhat.  A client used to be able
       > to add a label to a revision and cause it to be moved from any other
       > revision -- this was a useful 'atomic' operation.  Now, it seems, the
       > client must remove then add the label to move it -- this may cause race
       > condition problems as well as inconsistencies if the client fails to
       > complete the add.
    
    There is no race condition.  The "replace" operation moves the label
    from whatever revision of the versioned resource it currently is on
    (if any) to the specified revision.
    
       Yep, I noticed this as well.  The "replace" operation within the LABEL
       method appears to be missing an operand.  It's replace X with Y, but right
       now the marshalling only includes "X" (well, it could be "Y" I suppose, but
       it's one or the other, not both).
    
    The operation is not to "replace one label with another", but rather to
    "replace the revision selected by the label by another revision".  So
    only one operand is required, i.e. the revision that the label will 
    select following the "label replace" operation.
    
    Cheers,
    Geoff