Next message: Tim Ellison/OTT/OTI: "Re: stable URL's"
To: ietf-dav-versioning@w3.org
Message-ID: <OF85CFE481.75CE67E1-ON852568EE.006F7347@ott.oti.com>
From: "Tim Ellison/OTT/OTI" <Tim_Ellison@oti.com>
Date: Mon, 29 May 2000 16:37:34 -0400
Subject: RE: Workspaces as versionable resources
Pls see <tim> below.
"Clemm, Geoff" <gclemm@Rational.Com>
20-05-00 06:24 PM
To: "'Tim Ellison/OTT/OTI'" <Tim_Ellison@oti.com>, ietf-dav-versioning@w3.org
cc:
Subject: RE: Workspaces as versionable resources
There are three questions here:
- have we lost the ability to create a configuration?
- have we lost the ability to create a deep revision of a collection?
- are CHECKOUT/CHECKIN the appropriate way to create baselines?
The answer to the first question is "no".
Once we removed dynamic revision selection from workspaces,
a workspace provided all the functionality of a configuration,
so there was no reason to have two different resource types
(i.e. you can create a configuration, but it's now called
a "workspace").
<tim>
Adding a collection to a configuration was a shallow operation (i.e., the
configuration could contain 'holes' in the namespace).
</tim>
The answer to the second question is also "no".
The way you create a deep revision of a collection
is to place that collection in a workspace, and then
create a baseline for that workspace. So a deep revision
of a collection is called a "baseline".
<tim>
Here the implication is that creating a baseline is a deep operation
(i.e., all reachable resources from root). In this definition, a baseline
will offer a 'deep revision' of a collection, but does not provide for a
configuration.
I'm not sure why I would want configuration anyway, though I can image
others would.
</tim>
As for question three, Chris has also objected to using
CHECKOUT/CHECKIN to create baselines, so let's just switch
to a MKBASELINE method. This will also make things more
uniform, since otherwise there were two ways to create a
workspace (MKWORKSPACE and CHECKOUT applied to a baseline).
I'll make this change, and try to get out a 04.6 draft by
Monday.
Cheers,
Geoff
-----Original Message-----
From: Tim Ellison/OTT/OTI [mailto:Tim_Ellison@oti.com]
Sent: Friday, May 19, 2000 9:18 AM
To: ietf-dav-versioning@w3.org
Subject: Workspaces as versionable resources
I would like to declare my dissatisfaction with workspaces as versionable
resources.
From what I can make out, the only reason that this was considered was to
provide a "snapshot" mechanism to capture the state of the workspace (and
in particular its target selections).
However, by making workspaces versionable, we also get into the quagmire
of
selecting revisions/working resources of a workspace resource.
The obvious question is, when specifying (in a header) a URL to a
versioned
workspace, which workspace is used to select revisions of the workspace
and
versioned collections along the path?
...but it also permits structures that don't seem to have any useful
meaning, such as branching in workspace history.
We appear to have lost two useful concepts.
(1) Configurations, (which could be used to capture the state of a
workspace target selection),
(2) Deep versioning of collections
Tim