Re: Comments on the "new" 2518

The request from an unmodified client cannot be interpreted
by a server as requiring atomic move semantics,
since 2518 requires the standard request to be
interpreted as "best effort".  Therefore a client is going to
have to modify its request to indicate it wants atomic
move semantics, and should use the request we already have
defined (REBIND).

Cheers,
Geoff

Lisa wrote on 03/06/2006 07:59:39 PM:
> 
> I read 'T' for 'F' when I initially read John's email, so I guess 
> that would require clients to change their code to get atomicity. 
> John would a solution that required clients to change their code to 
> require atomicity be acceptable?
> 
> Lisa
> 
> On Mar 6, 2006, at 2:59 PM, Julian Reschke wrote:
> 
> > Lisa Dusseault wrote:
> >> I am on the fence for how to go with this issue,  but I do note 
> >> that your solution doesn't solve John's problem.  There are many 
> >> existing, deployed clients who already implement MOVE and John 
> >> wants WFS to behave appropriately for them without forcing 
> >> widespread client (in some cases, operating system) upgrade.
> >
> > Sorry?
> >
> > John asked for an extension that allows clients to state that MOVE 
> > should be atomic (the default staying non-atomic for backwards 
> > compatibility).
> >
> > Geoff pointed out that that extension already exists and is called 
> > REBIND.
> >
> > Clients that would want to take advantage of it would need to be 
> > upgraded in both cases. And, as a matter of fact, they would still 
> > need to be able to handle servers with non-atomic MOVE 
> > implementations.
> >
> > What am I missing?
> >
> > Best regards, Julian
> 
>