Good point. The spec is clear on the
normal expectations and any deviations need to depend on both sides understanding
some extension element in order to not lose interoperability.
Rich
"Andre Kramer"
<>
06/10/05 10:09 AM
To
"Subbu Allamaraju"
<>, <>
cc
Subject
RE: [wsrp-interfaces] [wsrp]
UserContext caching
A practical consideration is the fact that a UserContext
can include
extensions not understood by generic producer containers which could
make delta updates problematic.
Regards,
Andre
-----Original Message-----
From: Subbu Allamaraju [mailto:]
Sent: 10 June 2005 15:04
To:
Subject: Re: [wsrp-interfaces] [wsrp] UserContext caching
Thanks for the reference.
It would be interesting to see how Producer implementations interpret
this. As discussed yesterday, profiles tend to larger than public
params, and usually (atleast to my experience), only one or few items
change on some requests. In this case, it makes it efficient for the
consumer to send only updated values. It would also be desirable for
Producers to interpret that they are free to use previously supplied
values for items not supplied in the current request, as opposed to
replacing the entire UserContext with the new one.
Subbu
Rich Thompson wrote:
>
> Relative to the discussion on the Interfaces SC call today, the
> relevant statement in the spec (5.1.15) is "Note that the Consumer
MAY
> send UserContext information on any invocations as a replacement for
> information the Portlet MAY be storing in a session". I understand
> this to say that when the Consumer has a change to the cached
> UserContext it sends the entire UserContext.
>
> Rich
