Re: [sysinfo] draft ready for review

On 23/12/2009 18:38, JOSE MANUEL CANTERA FONSECA wrote:
> Hi,
> Although I understand your position I would like to have some
> mechanism to ensure there are no inconsistencies between the ontology
> and the API.

I agree. I've started looking at DCO & Bondi to try and make things as 
consistent as possible. Some things are easy, like agreeing that Battery 
level is percents and not [0,1]. Some things are harder, like what's the 
unit of CPU frequency in DCO, where's the definition of BatteryLevel in 
Bondi. Some things seem definitely incompatible, like DAP's notion of 
addressing individual hardware units by Id. I'll send something more 
exhaustive some time next week.

> What about having some kind of automatic generation of API properties
> from a branch in the ontology? Ensuring, for example, that the name,
> data type and description (precise semantics) of the properties in
> the API should be the same as the name of the properties in the
> ontology.

I'm not sure how that would work. Would you be able to elaborate?

> However there is an issue as the ontology has not created a property
> for everything, as there are some properties which can apply to
> multiple entities, for example 'vendor', 'model' or 'version' can
> apply to multiple entities, CPU, Battery, Camera, whatever.
> Also if this mechanism is sufficiently documented will allow third
> parties to extend the API taking as input other properties present in
> the ontology.
> By the way, the ontology does not include all the CPU properties you
> are considering. They will be added for the next version. I think
> your API specification can indeed provide a very good input for the
> ontology.

Thanks! I'm happy to provide feedback.


> I just want to be sure there is consistency between the specs.
> Best R.
> -----Mensaje original----- De: Max Froumentin
> [] Enviado el: lunes, 21 de diciembre de 2009
>; Asunto: Re: [sysinfo]
> draft ready for review
> On 21/12/2009 13:10, JOSE MANUEL CANTERA FONSECA wrote:
>> Hi,
>> the spec looks good but my concern is that you are duplicating the
>> work already done in the Delivery Context Ontology specification.
>> I think it is needed to have cross-reference between this API and
>> the DCO specification whenever possible, in order to have a
>> normative definition of each property in only one specification
> I think that this specification rather selects a few properties,
> which you should be able to find in the much more exhaustive DCO
> specification, and incorporates them into a self-contained document,
> IMHO following the simplicity principle mentioned in the charter of
> this WG.
> As noted in an issue of the draft, we need to map its properties to
> DCO's, if possible, and perhaps add that the other properties from
> DCO can be used in get/set/watch, if corresponding callback objects
> are defined. Perhaps you can help with that. Just now I looked for
> where in DCO is the CPU's current frequency but couldn't find it, so
> I obviously need assistance.
> Max.

Received on Friday, 25 December 2009 11:20:01 UTC