[infod-wg] Chapters 1 and 2 of the Base Spec
Steve Fisher
S.M.Fisher at rl.ac.uk
Thu Feb 23 10:32:25 CST 2006
On Wed, Feb 22, 2006 at 06:46:50AM -0800, Dieter Gawlick wrote:
> Steve,
>
> Sorry for the late response.
Only Susan, Chris and I turned up for the meeting so we decided to
abandon it this week. I will therefore try to continue the
requirements discussion by e-mail.
>
> I agree with your statement :"The INFOD registry will ...'
Just for clarification do you agree with my assertion that the
statement in the spec is incorrect?
> As for the goals:
>
> I agree that 0, 1, 2, 3, 4, 8 are in and 6, 7, 9 are out.
>
> I think 5 should be left in, see the car use case.
I don't think we can do 5 as it starts: "If publishers expose states"
- they have no interface to do this as far as I can see.
> Let us discuss 10 at the next conference call - my perception in Athens
> was that we like to allow publishers and consumers to offer at least
> once or at most once notification/consumption respectively. Certainly
> there will be more options as we move on
I rejected this initially becuase of the words "consumption" and
"propagation". QoS is rather hard to specify - I guess you request it
in the subscription and publishers and consumers need to be able to
announce in their properties that they can offer the desired QoS. When
I see the XML messages in the Use Case I may be convinced.
> 11 should be re-formulated, we assume identity management and do
> complement it with the INFOD registry support.
How does it appear in the interfaces?
> As to 12. We actually support the use of multiple filters - see use
> cases; certainly the sophistication could be enhanced.
I rejected this as it stands because it mentions "disseminators" I
presume that these "filters" are what are elsewhere referred to as
constraints. I would be inclined to drop "12" as I think it can be
assumed to be the case.
Steve
>
> Dieter
>
>
> Steve Fisher wrote:
>
> >I have just gone through chapter 1. I note that it says:
> >
> >"The INFOD registry will only select those publishers to act on
> >subscription who understand the vocabulary that is referenced in that
> >subscription; the same applies to consumers."
> >
> >I just want to confirm that this is incorrect as the data vocabulary
> >of the consumer is not necessarily known to the registry according to
> >what we discussed last week.
> >
> >The next problem is the set of goals. The full set of requirements are
> >set out with a note saying that #9 is not satisfied by the base
> >spec. From my understanding we only satisfy 0-4 and 8 and I would
> >propose to renumber 8 as 5 and drop 5-7and 9-12 from the base spec.
> >
> >This is rather fundamental to our understanding of the spec. If anyone
> >thinks that I am throwing out too much I am happy to argue requirement
> >by requirement.
> >
> >Have I got it wrong?
> >
> >Steve
> >
> >
> >
> >
>
More information about the infod-wg
mailing list