W3C home > Mailing lists > Public > ietf-dav-versioning@w3.org > October to December 2000

Re: Working collections

From: Geoffrey M. Clemm <geoffrey.clemm@rational.com>
Date: Fri, 17 Nov 2000 15:00:35 -0500 (EST)
Message-Id: <200011172000.PAA15755@tantalum.atria.com>
To: ietf-dav-versioning@w3.org
   From: Tim_Ellison@uk.ibm.com

   We agreed that URLs to versions exist outside the normal DAV namespace.
   This is concept is variously reflected in the spec as virtually hosted URLs
   (http://repo.webdav.org/his/23/ver/42) and 'stolen' parts of the root
   collection namespace (e.g. http://www.webdav.org/repo/wr-157.html).
   Clients should not, therefore, expect to be able to construct new URLs
   based on these server maintained URLs (i.e. removing / adding segments to
   reach other resources).

Yes.

   Q1:
   We know that a versioned collection is a mapping from names to version
   histories.

Actually, we no longer use the ambiguous "versioned collection"
terminology, so we actually don't know that (:-).  We do know that
a collection version is a mapping from names to version histories.
We also know that a collection version selector is a mapping from
names to other version selectors (and to resources that aren't under
version control).

   So when you check out a versioned collection what can you do
   with the resulting working collection?

Working on a checked out version selector for a collection is well
defined, so I assume you mean "when you check out a collection
version, what can you do with the resulting working collection"?

You are right that this is not defined, and does not follow from the
semantics of checked-out version selectors, so this needs to be fixed.
My first choice would be your subsequent answer, i.e. that "you can
only check out a collection version selector, and not a collection
version".  Greg doesn't like that answer, so it's probably worth exploring
what a working collection might be (although I haven't given up hope
in trying to get Greg to just use workspaces and checked out version
selectors :-).
 
   Are you constrained to creating members that bind to histories only?

If we allow it working collections, then being constrained to creating
members that bind only to histories makes sense.

   Are you prevented from creating new
   bindings at all?

I don't see any reason for this constraint.

   Q2:
   Can you delete bindings from a working collection -- presumably yes since
   if you can't change a working collection's members you can only use them to
   change properties? (There is clearly a good case for checked out collection
   version selectors).

Yes.

   When the server sees a URL to a working collection, say of the form
   http://repo.webdav.org/vr/9/wr/1/foo it can 'know' about the form of these
   URLs to determine that everything up to 'foo' is the URL to the working
   collection and 'foo' is the member of that collection, but there would be
   (typically) no relationship between that URL and the URL of the history
   resource that 'foo' is currently bound to; so it would not be possible to
   slash-through 'foo' to reach other resources.

Correct, which is why I find working collections of negligible value
(compared to checked-out collection version selectors, which you can
slash-through).

Cheers,
Geoff
Received on Friday, 17 November 2000 15:01:19 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 8 January 2008 13:57:39 GMT