Tao, to answer your last question first, this looks more like an OPES issue
than WEBI.
I've added the OPES list to the distribution, replies should have the lists
pruned as appropriate.
At 15:28 7/6/2001 -0500, Tao(_dot_)Wu(_at_)nokia(_dot_)com wrote:
> -----Original Message-----
> From: ext Ian Cooper [mailto:icooper(_at_)equinix(_dot_)com]
> Sent: Friday, July 06, 2001 3:14 PM
> To: Dan Li; Tao(_dot_)Wu(_at_)nokia(_dot_)com
> Cc: webi(_at_)lists(_dot_)equinix(_dot_)com
> Subject: RE: RUP v1 document
>
>
> At 11:01 7/6/2001 -0700, Dan Li wrote:
> >At 05:03 PM 7/5/2001 -0500, Tao(_dot_)Wu(_at_)nokia(_dot_)com wrote:
> >>RUP seems to be quite focused on invalidation. Is there
> any related work
> >>on capability description of the proxy, i.e., what type of
> service it
> >>provides?
> >
> >Capability, perhaps the UDDL, web services work in W3C?
> also, the rules
> >work in OPES?
>
> That question sounds related to the IDD (Intermediary Discovery and
> Description) stuff we proposed doing within the group (well, the
> description part at least), unless I'm missing additional
> context for the
> question.
<snip>
I have not followed the posts in the mailing list very closely, and this
problem just came to me recently, so there may be similar discussions in the
past - suppose an intermediary is performing some service for a mobile host,
and the mobile host now moves to another access router, which connects to
its own cache/intermediary. Will the mobile host handoff result in
"intermediary service handoff"? Suppose the answer is yes, then it will
become necessary for the intermediaries to exchange capabilities etc.
This sounds very much like an issue I raised in the discussion recently,
and is - I suspect - one of the very big concerns of the "end to end
folks"; roaming users will need any state that would now be stored within
the network to be distributed throughout that network.
While I'll admit that I haven't thought about this specific issue too
deeply, my gut reaction would be that the effect of roaming would result in
new discovery being performed (in the WPAD world this would be similar to
new WPAD configuration being obtained as a result in changes in DHCP
configuration). That in turn would lead to the device being informed of
the services that were available and where they were located. (Another
approach is query based, "I want xxx".)
If there is no hand off then the new intermediary IF ANY would have to be
configured with the services the roaming user had previously been
using. Of course, part of the point to OPES seems to be that the user
agent does not carry this information around... Further, some of those
services may just not be available on the new network (they might be
proprietary to the other network provider), so then the user sees a totally
inconsistent view of the world - which brings me back to the issue I raised
but while I believe no-one decided to comment on.
[In short - sorry for rambling - I'm not sure how you can do things without
a hand-off... and that opens a nice big can or worms.]
So that is where my previous question came from. Is this something related
to IDD, and are there alternatives to do the same?
To reiterate, it seems more an OPES issue than WEBI.
Thanks,
-Tao