W3C home > Mailing lists > Public > ietf-dav-versioning@w3.org > July to September 2001

RE: WebDAV Invalidation (Was Re: Allow: header and supported methods)

From: Lisa Dusseault <lisa@xythos.com>
Date: Sun, 2 Sep 2001 09:47:59 -0700
To: "Eckhard Kantz" <deltav@wegalink.de>, <ietf-dav-versioning@w3.org>
Message-ID: <HPELJFCBPHIPBEJDHKGKMEBBCMAA.lisa@xythos.com>
The WebI WG in the IETF is also working on cache invalidation.  I've been
following this effort in order to ensure that it works for DAV permissions
changes as well as content changes.  As far as I can tell with their
requirements, it will work, because invalidation of caches due to
permissions changes is simple:

 - the only documents with permissions that can be cached at all are those
that are publicly readable
 - the only permissions change that affects the cache is a reduction in read
permissions, from publicly readable to not publicly readable
 - the only message that needs to be sent to the cache is "invalidate your
cache for this item"  Possibly the only detail to add is whether a new
download is recommended or not.

You specifically mention push, which could mean just sending a simple
notification, or it could mean pushing the body.  I don't like pushing the
new body to clients.  The only thing that should be pushed is a message, as
simple as possible, saying "invalidate your cache for this item; a new
download should be done".   This leaves the client able to make the request
when most appropriate, which is more likely to be "just-in-time" than
pre-fetch.

It's possible for web sites to push a new body to sophisticated caching
servers that they have a special relationship with; that makes web sites
that update in the night able to vastly reduce traffic by using low-load
times to get the new content out one hop.  However, that's very different
than pushing to all clients.  I'd point out that these caching servers are
trying to figure out how to charge the web sites to improve their
availability and reduce load in this manner.  In other words, push is
something web sites are likely to have to pay for, because most end-points
don't want to receive a costly push for free.

lisa

> -----Original Message-----
> From: ietf-dav-versioning-request@w3.org
> [mailto:ietf-dav-versioning-request@w3.org]On Behalf Of Eckhard Kantz
> Sent: Wednesday, August 22, 2001 12:41 PM
> To: ietf-dav-versioning@w3.org
> Subject: Re: WebDAV Invalidation (Was Re: Allow: header and supported
> methods)
>
>
> The protocol described in the ESI document allows to invalidate
> documents that have been downloaded
> to a local machine by applying a push technology. This could
> solve already several conflict
> situations or even partly prevent problems.
>
> On the other hand there seems to be an increasing need for more
> fine-grained notification services
> that extend the traditional access control systems. Picture 1 in
> the following longer article tries
> to classify them:
>
> "Beyond 'Yes or No' - Extending Access Control in Groupware with
> Negotiation and Awareness"
> (http://www.informatik.uni-bonn.de/~os/Publications/COOP98a.ps)
>
> Maybe those needs are also worth discussing if they could be
> supported in the spec in order to allow
> applications to build up on them. The invalidation protocol seems
> to be a good basis also for this.
>
> Eckhard
>
>
> -----Ursprüngliche Nachricht-----
> Von: Eric Sedlar
> An: ietf-dav-versioning@w3.org
> Gesendet: Dienstag, 21. August 2001 19:00
> Betreff: WebDAV Invalidation (Was Re: Allow: header and supported methods)
>
>
> Check out
> http://www.esi.org/invalidation_protocol_1-0.html for some work
> that looks pretty similar
> to what we are talking about.
>
> --Eric
Received on Sunday, 2 September 2001 12:49:04 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 8 January 2008 13:57:42 GMT