RE: Jim Whitehead's Hong Kong talk - Adding versioning to the web

   From: Alan Kent [mailto:ajk@mds.rmit.edu.au]

   My reading of the DeltaV spec in Section 6.7 indicates that
   VERSION-CONTROL is creating a

       "new version-controlled resource for an existing version history".

Yes.  But since the VCR needs to identify a particular version
from that version history, a particular version is required
when you create the new VCR.

   The examples the spec has then uses URIs like /his/12/ver/V3
   identifying a version from a history. Jim's example in his slide
   is a version controlled resource (slide 52).

Jim's example doesn't display the actual syntax, but describes
the result, so it leaves out this kind of detail.

   Should the above text in the first paragraph of section 6.7 of the spec
   be expanded to make it clear that the URI identifies a version from the
   version history?

The introductory text gives you the motivation for the semantics,
and adding in this kind of detail risks blurring the overall picture.

   Or can any sort of resource be identified (version
   controlled, version, version history, etc) as long as the version
   history can somehow be identified?

No, only a version.

   Or is the spec OK and the example in the slides is wrong.

I'd describe the example in the slides as "simplified" rather than
"wrong".

   Maybe the slides are a bit hand-wavy to get concepts across more
   easily... The spec seems to state quite clearly that a version URI
   should be supplied

Yes.

   (it makes no comment about using a version-controlled resource URI
   - unless its mentioned somewhere else I have not found!).

Nope, no hidden semantics.  One of the benefits of the rigorous
precondition/postcondition approach, is that you don't have to worry
about semantics being scattered througout the protocol text ...
if it's not in the preconditions or postconditions, it ain't so.

Cheers,
Geoff

Received on Wednesday, 27 June 2001 14:16:04 UTC