CIM 1.92 open issues - placeholder ticket

[domlowe]

putting all open CIM 1.92 issues in this ticket for now.

Rupert Recently I have been trying to get the basic concepts required to describe the coupling of software components (which is mostly part of the software description) into CIM1.9x and would like to propose a few updates to CIM1.9 that did not make quite make it into this release. Note that most of these concepts, or relationships, are not in CIM1.5.

1: I would like the "partition" attribute to be changed to "content any". Partition is an attribute of simulatedProperty and can be seen in the Configuration diagram. It is there to capture the way in which the data associated with the property is partitioned amongst processes for the software components. There is no standard way to capture this information (that I know of) so this will allow people to add in their own way of capturing it.

2: I would like the "scheduleDetails" attribute to be changed to "content any", scheduleDetails is an attribute of Schedule, see the CompositeObervation Complex Parameter Types diagram. It is there to capture the relative running order of the software components in the implementation. There are a number of ways to capture this information - workflow languages being one. Again this will be a place where people can add there own way of capturing this information.

3: I would like ProcessSet to reference 1 or more ConfiguredSoftwareDescription's. In this structure a deployment can consistent of one or more executables. Each executable can consist of one or more sets of software components which *share* threads. See the CompositeObervation Complex Parameter Types diagram for the structure. The reference to which software components share threads is missing in the current version.

4: I would like to add an attribute called "propertyStructure" (or something like that), again of "content any" to simulatedProperty. This would capture the way in which a property is implemented in software (e.g. a structure of arrays of certain sizes). Rather than model this directly in the CIM I thought we could allow people to use their own way of of capturing it. This is one place where we could try to define a standard but it would be quite complex to work in the general case.

5: I would like to be able to capture the concept of software ("SoftwareDescription" presumably) being compiled by a "Compiler" (see the Common diagram).

Kristin

1. Reference stereotype – how to handle. Extension of CI_OnlineResource? Use MO_OnlineResource?

· There is a list in my report of all the attributes and associations that had the reference stereotype attached to them in v1.5 (it was also circulated by email a month or two ago).

· Need to go through list and categorise them as either CI_OnlineResource (specialised) or gml by ref (inline..) or normal associations (as in the rest of the model). Bryan to provide input.

2. Grids package:

· Need Grid abstract – document summarising the grid package?

· needs to be divided up for software and science as per Bryan’s email 21/3.

· Is not compliant with ISO 19123 and 19129. I’ve made it broadly compliant with the foundation standards, but not the coverage standards, since this would require really major re-engineering.

Tobias My main concern actually is whether this pulls in additional attributes from GML - from what I've read, a GML AbstractFeatureType? has (can have?) spatial attributes. So does declaring e.g. Conformance as a FeatureType? mean that a CIM conformant XML document may in fact define a spatial dimension/geometry for a Conformance instance? If yes, we should say something in the CIM model about restricting this (I guess this is a core validation issue).

See email thread 'timing of CIM 2.0' for further discussion

[domlowe]

noting Tobias's comments

[allyn]

Allyn Various US projects had requests for the CIM:

1. COG has a desire for a project that is much more general and high-level than a formal NumericalExperiment with all of the structure that that requires. Just a description and some contact details would be sufficient.

2. CCSM has a desire for having "custom attributes" (when CCSM/ESMF use the word "attribute" what they generally mean, in Metafor terms, is "ComponentProperty"). A custom attribute would be one that is not bound to a CV and is not a standard property (such as those provided by the Questionnaire). In CIM v1.7, this has been solved by adding a type="custom" attribute to ComponentProperties.