Next message: jamsden@us.ibm.com: "Re: DAV:revisions property for a workspace resource"
Message-ID: <38FC417D.5B647D0F@marconicomms.com>
Date: Tue, 18 Apr 2000 13:05:33 +0200
From: Edgar Schwarz <Edgar.Schwarz@marconicomms.com>
To: DAV Versioning <ietf-dav-versioning@w3.org>
Subject: Re: DAV:revisions property for a workspace resource
jamsden@us.ibm.com wrote:
>
> <jra>
> Two things. Versioned collections are pretty hard to implement,
Bad news, because I'd like to implement them :-)
> Second, a versioned collection only
> versions the members of the collection, the names of the bindings and what
> they are bound to. It does not version the contents of those members. This
> is not the same thing as a configuration. From the model point of view, a
> configuration (and activity, workspace, and label) is a revision selector
> while a collection revision is not.
I meant a deep revision. This qualifies as a revision selector, right ?
> <gmc>
> Note that you don't send your updates back with a CHECKIN request,
> but rather with a PUT request. The CHECKIN request just tells the
> server to remember the current state of the resource as a new revision.
> There have been protocol's designed to address this concern, but since
> this is not a versioning issue (i.e. it's just about optimizing the
> PUT operation), it is not one that is addressed by the versioning
> protocol.
> </gmc>
I think that delta stuff comes very natural in a versioning context. Is
there a more intuitive case for delta information than in a
CHECKIN <resource> <baseversion> <delta data> ?
And how much of protocol do you need to design ? You just need to define the
delta format and a delta header for checkin. Rather easy to implement
I think. But perhaps I'm missing something.
> <jra>
> There was talk in the original WebDAV spec a couple of years ago to send
> deltas. Unfortunately, this is resource type dependent, and difficult to
> put in the protocol without requiring changes to the protocol standard to
> support new resource types. Clearly this is unacceptable. It is also not
> clear that this optimization is required given the level of traffic and
> frequency of updates in an authoring environment.
> </jra>
Why do you need a new resource type ? There are very simple byte based
delta descriptions which would be shorter in any case than PUTing
the whole new version and doing a checkin afterwards.
Just send the data as application/octet-stream and you are done.
Cheers, Edgar
--
Edgar.Schwarz@marconicomms.com, Postf. 1920,D-71509 Backnang,07191/133382
Marconi Communications, Access Networks Development, Software Engineering
Privat kann jeder soviel C programmieren oder Videos ansehen wie er mag.
Niklaus Wirth. Make it as simple as possible, but not simpler A.Einstein