Re: Allow: header and supported methods

Well, the client's listening port could also be the standard http port if it is not occupied by a
running web server. This should reduce firewall problems

Alternative ways to get the invalidation responses through could be:
- consider the use of UDP packages
- let clients know about each other to a certain degree and use them to spread out messages (maybe
btree like)
I am not quite sure about the consequence of this approach but it could reduce the server's number
of sockets considerably.

Maybe polling could be offered as a fallback in case a firewall blocks all other communication.
However, opening a listening port on client side will probably provide to less overall network
traffic and to least delay.

Eckhard


-----Ursprüngliche Nachricht-----
Von: "Eric Sedlar" <Eric.Sedlar@oracle.com>
An: <ietf-dav-versioning@w3.org>
Gesendet: Dienstag, 21. August 2001 01:41
Betreff: RE: Allow: header and supported methods


> No, you could do something like FTP does and have an HTTP
> method that tells the server a port to connect to in order
> to deliver invalidation responses, something like:
>
> REG_NOTIFY /foo.txt HTTP/1.1
> Host: www.oracle.com
>
> <respond-to>
>   <href>http://192.251.211.44:4234/notify</href>
> </respond-to>
>
> and the server would then post a list of HREFs to invalidate
> to that address.
>
> You just require that the client also be able to function
> as a server.
>
> The disadvantage of this approach is that is that it will
> have problems with many firewalls.  The advantage is that
> the server doesn't have to keep open a socket to the client
> just in case an invalidation would have to be done.  The
> server could end up with a very large number of sockets in
> this case.
>
> --Eric
>
> > -----Original Message-----
> > From: ietf-dav-versioning-request@w3.org
> > [mailto:ietf-dav-versioning-request@w3.org]On Behalf Of Jim Amsden
> > Sent: Sunday, August 19, 2001 5:20 PM
> > To: ietf-dav-versioning@w3.org
> > Subject: Re: Allow: header and supported methods
> >
> >
> > We'd have to introduce push technology into HTTP.
> >
> >
> >
> >
> >
> > Peter Raymond <Peter.Raymond@merant.com>
> > 08/19/2001 08:15 AM
> >
> >
> >         To:     Jim Amsden <jamsden@us.ibm.com>
> >         cc:     Eric.Sedlar@oracle.com
> >         Subject:        Re: Allow: header and supported methods
> >
> >
> >
> > Hi Jim,
> >
> > Jim Amsden wrote:
> >
> > > Anyone interested in a new WebDAV working group to add event
> > notification?
> > > We could call it DAVE.
> >
> > Sounds like fun and certainly something I would be interested in.
> >
> > But, how would we do that over HTTP? Since the protocol is
> > request-response based the client
> > would
> > have to be polling at regular intervals asking if any events of interest
> > have occurred.
> > Would this perform well?
> >
> > Regards,
> > Peter Raymond - MERANT.
> >
> >
> >
> >
> >

Received on Tuesday, 21 August 2001 03:38:51 UTC