Review of 'change set' terms & shapes

["Sarabura, Martin" <msarabura@ptc.com>]

Nick suggests we discuss these questions at this week's meeting, so here are the emails for reference.

Regards,

 

Dr. Martin Sarabura

R&D Fellow, PTC

 

 

 

From: Sarabura, Martin
Sent: February-08-16 1:33 PM
To: 'Nick Crossley'
Subject: RE: PTC review of 'change set' terms & shapes

 

Nick, there may be some issues with what you have proposed for change sets:

- How to treat a selection or contribution that has been removed as part of the change set

- Can a change set reference a new version of a contribution as part of the change? Why is "contribution" not part of the shape? Maybe I'm just not that clear on the meaning of contribution.

- At a more fundamental level, I suppose the question is why change sets need to be exposed as consumable external resources. Of course internally they satisfy an important requirement which is to represent the difference between two change sets in a handy format. But will an external application need this? What is the use case?

 

Using the term "overrides" certain suggests a particular meaning for the related configuration. Is it necessary to use that term specifically? You could use "RelatedTo" to suggest an agnostic relationship. I assume that typically you'd want to override a configuration in a particular context, right? In which case you'd need to reference the context too. I can see how you don't feel the spec is ready to include this property because it does open up some new issues such as context, but I can also see the benefit so maybe it's worthwhile investing the effort.

 

In summary I would like to see why change set is required as a resource for external usage. I see the value of personal configuration and overrides though I think you should flesh out overrides a little more and first class it if possible.

 

On a side note, I am confused as to why there is no "Configuration shape". Configurations are definitely resources, we talk about them as such throughout the documents. Why do they have no shape in the spec?

 

Hope this helps,

Martin

 

 

Dr. Martin Sarabura

R&D Fellow, PTC

 

 

 

 

 

From: Nick Crossley [mailto:ncrossley@us.ibm.com]
Sent: February-05-16 12:59 PM
To: Sarabura, Martin
Subject: RE: PTC review of 'change set' terms & shapes

 

The changes are in two files. First, the vocabulary at https://tools.oasis-open.org/version-control/browse/wsvn/oslc-ccm/branches/cset/specs/config-mgt/config-vocab.ttl, where we added two classes and one property:

oslc_config:ChangeSet
    a rdfs:Class ;
    rdfs:isDefinedBy oslc_config: ;
    rdfs:label "ChangeSet" ;
    rdfs:comment "A change set configuration represents a set of changes to some other configuration." .

oslc_config:PersonalConfiguration
    a rdfs:Class ;
    rdfs:isDefinedBy oslc_config: ;
    rdfs:label "Contribution" ;
    vs:term_status "testing" ;
    rdfs:comment "A configuration intended for use by a single agent, rather than shared." .

oslc_config:overrides
    a rdf:Property ;
    rdfs:isDefinedBy oslc_config: ;
    rdfs:label "overrides" ;
    vs:term_status "testing" ;
    rdfs:comment "A relationship between configurations, reserved for future use." .

And secondly, a shape for ChangeSet was added to the file https://tools.oasis-open.org/version-control/browse/wsvn/oslc-ccm/branches/cset/specs/config-mgt/config-shapes.ttl. We described the ChangeSet with some text "A reference to the resource identifying the versions of the resources (configuration items) in the configuration. This may be a delta to the selections in some other configuration, or may be an entire set of selections with the delta applied.".

IBM wants to add the oslc_config:overrides property to the vocabulary so we have a property in the right namespace, sine we want to use that property. However, we do not think the property should have any assigned meaning, nor be added to shapes, because we felt the TC as a whole has not had time to standardize its semantics in a way that would be useful, but general enough for the wide variety of configuration management systems out there, and the different ways those systems might represent change sets and their relationship to other configurations.

We did not add a new shape for a PersonalConfiguration, since we just expected providers to add that marker type to an existing stream or baseline - that is, the shape is not impacted.

Nick.