[Fiware-ngsi] one further binding question

[Tobias Jacobs]

Hi Denes,

Thanks a lot for your updates!
I accepted most of the changes, with a few exceptions I like to discuss:

-          The "unique entity id" issue has not yet been finally decided. My approach would be to decide Ids to be unique in our IoT chapter, but not to put that into the standard. This is what Boris also suggested, as far as I understood.

-          I am not happy yet about the description of Context Element: To my understanding, a context element is just a generic container for information about context entities. So it can contain any subset of attributes, subject to the restriction that if a domain name is given then it should only contain attributes from that domain.

-          I would say the attribute domain must not be present in the response to a GET on a context entity. (the question becomes obsolete if it is decided not to enhance NGSI, see next point)

-          Do we have a decision for or against enhancing NGSI with the possibility to influence how the context query response is structured, as discussed in previous e-mails?

Best regards
Tobias


From: Bisztray, Denes (NSN - HU/Budapest) [mailto:denes.bisztray at nsn.com]
Sent: Donnerstag, 29. März 2012 14:31
To: Tobias Jacobs; fiware-ngsi at lists.fi-ware.eu
Subject: RE: [Fiware-ngsi] one further binding question

Dear Tobias, all,

I updated the draft on the SVN with track changes. Please have a look at it and if OK, accept the changes in the doc and commit.

Best,
Dénes

From: ext Tobias Jacobs [mailto:Tobias.Jacobs at neclab.eu]<mailto:[mailto:Tobias.Jacobs at neclab.eu]>
Sent: Thursday, March 29, 2012 1:16 PM
To: Bisztray, Denes (NSN - HU/Budapest); fiware-ngsi at lists.fi-ware.eu<mailto:fiware-ngsi at lists.fi-ware.eu>
Subject: RE: [Fiware-ngsi] one further binding question

Hi Denes,

In Table 5.5.5 in the NGSI specs it is written "If EntityId uniqueness is only guaranteed in combination with Type, then Type SHALL be present."
I interpret this as NGSI admitting to have several Entities having the same name but different type.

Such sets on Entities with the same name would be represented by the same ../contextEntities/{entityID} resource in our binding (unless we agree to do it differently).

Best
Tobias

From: Bisztray, Denes (NSN - HU/Budapest) [mailto:denes.bisztray at nsn.com]
Sent: Donnerstag, 29. März 2012 10:09
To: Tobias Jacobs; fiware-ngsi at lists.fi-ware.eu<mailto:fiware-ngsi at lists.fi-ware.eu>
Subject: RE: [Fiware-ngsi] one further binding question

Something completely different:
I assumed until now that the Name of the EntityId  in our system has to be unique because of the nature of the RESTful binding. Does that hold?

Best,
Dénes

From: fiware-ngsi-bounces at lists.fi-ware.eu<mailto:fiware-ngsi-bounces at lists.fi-ware.eu> [mailto:fiware-ngsi-bounces at lists.fi-ware.eu]<mailto:[mailto:fiware-ngsi-bounces at lists.fi-ware.eu]> On Behalf Of ext Tobias Jacobs
Sent: Wednesday, March 28, 2012 5:51 PM
To: fiware-ngsi at lists.fi-ware.eu<mailto:fiware-ngsi at lists.fi-ware.eu>
Subject: [Fiware-ngsi] one further binding question

Dear members of the NGSI list,

The binding of NGSI-10 is now as good as finished, see the current version at
https://forge.fi-ware.eu/scmrepos/svn/iot/trunk/documents/Ngsi10-RestfulBinding-Draft.docx

Unfortunately, I have identified another issue I like to discuss with you. Please let me know your opinion.

We previously agreed that a GET on an ../{EntityID} resource results in a list of all available attribute values in only one ContextElement, without attribute domains. We further agreed that a GET on ../{EntityID}/attributeDomains results in the same information, but here the response consists of one ContextElement per attribute domain.

Now here is the problem I see: The possibility to distinguish between these two kind of queries goes beyond the specification of NGSI-10. NGSI-10 only allows a query for all attributes of an entity, and how the returned information is structured is not written in the standard.

My proposal to resolve this is to disallow the GET on ../{EntityID}/attributeDomains, which would mean that this resource becomes one which does not allow any interaction. Additionally, we could leave it to the system if the attributes should be structured by domain or not, like the NGSI-10 spec does.
Martin does not agree with that, his proposal is to accept this slight extension of NGSI-10.

What speaks for Martin is (correct me if I cite you incorrectly, Martin)

-          Allowing a GET on ../{EntityID}/attributeDomains is natural, and users would expect it

-          The functionality is useful
What speaks for my approach is

-          A gap between the functionality of the binding and the functionality of NGSI-spec might cause unforeseen problems, for example when someone has already designed a system according to the NGSI-specs and now wants to put a REST interface on top.

Thanks in advance for letting us know your opinion.
Best
Tobias

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.fiware.org/private/fiware-ngsi/attachments/20120329/fe8a3a52/attachment.html>