W3C home > Mailing lists > Public > w3c-dist-auth@w3.org > October to December 2003

Re: WebDAV MOVE

From: Geoffrey M Clemm <geoffrey.clemm@us.ibm.com>
Date: Tue, 25 Nov 2003 10:32:03 -0500
To: Helge Hess <helge.hess@opengroupware.org>
Cc: "'Webdav WG'" <w3c-dist-auth@w3c.org>
Message-ID: <OF10C75B0B.B73E04D1-ON85256DE9.00543A57-85256DE9.005550FA@us.ibm.com>
There are 3 obvious alternatives to move one set of resources
to another location:
- MOVE (cheap, least likely to work)
- COPY/DELETE (more expensive, more likely to work)
- PUT*/DELETE (most expensive, most likely to work)

So a client that cares about efficiency will try these three
methods in the above order, until one succeeds.  So what use
would it be for a server to say "out of the namespace I process"?
The client will try the next method in either case.

So is there anything that a server can do to expedite this
sequence?  The only thing I can think of is to have the
server have a special error code for MOVE that says "don't try to fake
MOVE with COPY or PUT*, because I know those will also fail"
and a special code for COPY that says "don't try to fake copy
with PUT*, because I know that will also fail".

But I think the likelihood that the MOVE or COPY implementation
will know this is small, and the benefit of skipping those
attempts is also small, so the overall benefit is very small.

Cheers,
Geoff


Helge Hess <helge.hess@opengroupware.org> wrote on 11/25/2003 09:51:17 AM:

> On 25.11.2003, at 15:02, Geoffrey M Clemm wrote:
> > I'm not sure what the benefit would be of this additional error
> > code for MOVE.  If the client wants to try COPY/DELETE as an
> > alternative to MOVE, then it should do so.  I don't see that
> > the clients decision to do so would be affected by a special
> > error code here.
> 
> [maybe we got the method wrong, of course COPY suffers the same issue! 
> the alternative to MOVE is GET/PUT/DELETE, not COPY/DELETE]
> 
> Well, MOVE is usually a cheap operation on the server and on the 
> network while PUT is expensive on the network. So I would assume 
> (without having thought through everything, I have to admit), that a 
> code which signals the client why it couldn't perform the operation 
> would be quite useful (something like out-of-the-namespace-I-process).
> With the Forbidden code we do not know whether the request was rejected 
> because of access restrictions or just because it is located on a 
> different server, right?
> 
> Note that I'm assuming that a server is usually "dumb" and will not try 
> to act as a proxy with regards to the other namespace (if it does we do 
> have the 502 code to detect errors, if I remember right).
> 
> Doesn't make sense?
> 
> regards,
>    Helge
> -- 
> I'm describing in a paragraph what took 14 hours to figure out. You may 
> want to sniff glue for a while, then reread this when you get out of 
> rehab. [aLa]
> 
Received on Tuesday, 25 November 2003 10:32:19 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 2 June 2009 18:44:05 GMT