RE: Don't copy live properties

For those new to this conversation you may want to refer to
http://lists.w3.org/Archives/Public/w3c-dist-auth/1998OctDec/0302.html for a
general discussion of why live properties are a bad idea. You might also
want to check out
http://lists.w3.org/Archives/Public/w3c-dist-auth/1998OctDec/0303.html for a
general discussion of why we ended up with live properties.

As I mention in section 7 of
http://lists.w3.org/Archives/Public/w3c-dist-auth/1998OctDec/0302.html for
us to be completely rid of live properties we would have to come up with
alternative mechanisms, along the lines discussed in the memo, to the
current live properties in WebDAV.

However, in the meantime, we can certainly make some progress by banning all
new live properties for the reasons discussed in
http://lists.w3.org/Archives/Public/w3c-dist-auth/1998OctDec/0302.html. Thus
the problem of copying live properties becomes significantly smaller,
restricted only to the ones defined in WebDAV.

Also, I understand that Jim Amsden has implemented a WebDAV server which
talks with a number of heterogeneous backend servers so there is experience
to be had. I suspect that JimA's server will be only the first of its ilk.

As for an actual scenario, I can make one up if you like but I prefer real
world experience. Given how little experience there has been with WebDAV in
the real world it would seem a bit early to muck with the standard's
language. 

JimD, I'm not proposing that your suggestion is wrong. Rather I feel that we
just don't know what we don't know. A little experience will, I believe,
help to clarify matters.

				Yaron

> -----Original Message-----
> From: Jim Davis [mailto:jdavis@parc.xerox.com]
> Sent: Thursday, December 31, 1998 1:18 PM
> To: w3c-dist-auth@w3.org
> Subject: RE: Don't copy live properties
> 
> 
> At 01:08 PM 12/31/98 PST, Yaron Goland wrote:
> >I think it is too soon for us to make this judgement. I am 
> willing to bet
> >that once we ban copying live properties we will find out 
> that this too
> >breaks many scenarios. Its a definite damned if you do, 
> damned if you don't
> >scenario. Thankfully there is a way out. Ban live properties.
> 
> While I deeply respect your intuition and judgement, I would 
> rather see
> someone propose an actual scenario that breaks.  Since we 
> know the current
> definition is no good, how could we be worse off?
> 
> Bear in mind, this problem could only arise when a resource 
> is copied to a
> location where some property are not live.  This has never 
> happened yet,
> because no WebDAV server in existance supports copying to a different
> server.  So there can't be a compatibility issue with an existing
> implementation.
> 
> How can you throw away all live properties without breaking locking?
> 
>

Received on Thursday, 31 December 1998 16:57:50 UTC