CIM Provider HOWTO Document

John Dennis jdennis at redhat.com
Mon Apr 8 13:59:49 UTC 2013


On 04/08/2013 04:51 AM, Roman Rakus wrote:
> On 04/05/2013 08:47 PM, John Dennis wrote:
>> On 04/05/2013 07:35 AM, Roman Rakus wrote:
>>> I'm confused in section where you are describing structures and array of
>>> structures. I think this is where Embedded Instances take place.
>>> However, the KonkretCMPI does not support embedded instances.
>>
>> Hello Roman:
>>
>> I'm not familiar with embedded instances. Can you elaborate or provide a
>> pointer to documentation?
>>
>> Thanks.
>>
> I can't find much in documentation. Just in CIM Metamodel (CIM
> Infrastructure Specification) [1] are few words about Embedded Objects.
>
> Embedded object is qualifier. For example you can have parameter of
> method with this qualifier. Such parameter will be embedded object
> (instance or class). The type of parameter is string. When we talk about
> instances - this string is encoded (embedded) instance of some class.
> There are functions in CMPI to create instance from this string
> representation.
>
> I have never used this. But I think it could help to encapsulate
> definition of parameter and passing it, but with need of such definition
> in model.

Ah, O.K. yes this now rings a bell with me and I did momentarily 
consider this as an option.  My understanding is it amounts to 
formatting the data in MOF syntax and passing it as a string. One then 
parses the MOF formatted string.

My take on this is:

* It's very awkward to use, it's not a friendly API scheme.

* I didn't find any support for parsing MOF syntax in any of the 
libraries a provider links with.

Thus at the moment it doesn't seem like a viable option and even if we 
grow support for parsing MOF syntax I'm not sure one would want to 
design an API around passing MOF snippets.

> [1]
> http://www.dmtf.org/sites/default/files/standards/documents/DSP0004_3.0.0.pdf
>


-- 
John Dennis <jdennis at redhat.com>

Looking to carve out IT costs?
www.redhat.com/carveoutcosts/


More information about the openlmi-devel mailing list