Next message: Eric Sedlar: "Localized baselines"
Message-ID: <011201bf78ff$01c75ee0$ab171990@us.oracle.com>
From: "Eric Sedlar" <esedlar@us.oracle.com>
To: "Clemm, Geoff" <gclemm@rational.com>, <ietf-dav-versioning@w3.org>
Date: Wed, 16 Feb 2000 20:25:22 -0800
Subject: Re: Members of a collection
Wait, is there no way to ask a collection for only those members that would
be currently selected? If I use ClearCase, I never have this problem. I
think that will be what users most frequently want, and I don't think we
want to wait for DASL to handle this important query.
I also disagree with Tim's claim that this is the same as a DELETE happening
while you weren't paying attention. Let's take the case where I know that
I'm the only person operating in a particular section of the URL hierarchy
(I have my own virtual single-user system). Clearly, I know that no delete
has occurred, yet I'm seeing garbage in my directories because I have a bug
in my currently selected RSR's.
--Eric
----- Original Message -----
From: Clemm, Geoff <gclemm@rational.com>
To: <ietf-dav-versioning@w3.org>
Sent: Wednesday, February 16, 2000 7:44 PM
Subject: RE: Members of a collection
> I agree that this is not a problem, but it might be worth
> having a couple of special status codes, i.e.:
> 4xx (No Such Revision)
> 4xx (No Such Working Resource)
> and then reserve 404 for when there is no versioned resource
> (or any other resource) at that URL. What do folks think?
>
> Cheers,
> Geoff
>
>
> > -----Original Message-----
> > From: Tim_Ellison@oti.com [mailto:Tim_Ellison@oti.com]
> > Sent: Wednesday, February 16, 2000 5:05 PM
> > To: ietf-dav-versioning@w3.org
> > Subject: Members of a collection
> >
> >
> >
> > To determine the members of a collection on a versioning
> > server, a client
> > issues a PROPFIND. In a non-versioning world if you are told
> > that /foo/ has
> > a member /foo/bar then you have a pretty good chance that you can GET
> > /foo/bar. However, in a versioning world your workspace may
> > not select any
> > revision of /foo/bar, so you 'see' that /foo/ has a /foo/bar
> > but you get a
> > 404 when you try to GET /foo/bar.
> >
> > This is going to be particularly interesting for 'browser'
> > type applications
> > that reveal one layer of the namespace at a time. However, I
> > claim that
> > this is no different than a non-versioning server showing its
> > members, then
> > a member being DELETEd before the client GETs it. One
> > difference is that
> > the versioning anomaly is more likely to happen.
> >
> > Just an observation.
> > Tim
> >
>
>