The API can be simplified for the users by replacing various sequences with the new IDL4 map type.

IDL4 extends the IDL type system with the types coming from DDS X-Types (maps,bitset.struct inheritance,bitmasks), the D&C specification should be extended to support these types in the xsd

Problem: The CCM DTD and SCA DTD in the assembly descriptors has the
capability to direct deployment of components on the same
processor. The capability is needed for deployment considerations to direct
which components are to be collocated on the same processor.
Example use case is there are devices in the system with the same
characteristics but when deploying the components, one wants to ensure that
certain components are deployed on the same processor not on different
processors with the same characteristics.
Recommended solution: Is to add an optional zero to many Host Collocation
element to the ComponentAssemblyDescription.

Hi,

The Deployment & Configuration XML Schema defines a Property element like:

<xsd:complexType name="Property">
<xsd:choice minOccurs="0" maxOccurs="unbounded">
<xsd:element name="name" type="xsd:string" />
<xsd:element name="value" type="Deployment:Any" />
<xsd:element ref="xmi:Extension" />
</xsd:choice>
<xsd:attribute ref="xmi:id" use="optional" />
<xsd:attributeGroup ref="xmi:ObjectAttribs" />
</xsd:complexType>
<xsd:element name="Property" type="Deployment:Property" />

This allows for the following invalid Property file to be passed silently
by the XML Schema validator:

<?xml version="1.0" encoding="UTF-8" standalone="no" ?>
<Deployment:Property
xmlns:Deployment="http://www.omg.org/Deployment"
xmlns:xmi="http://www.omg.org/XMI"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://www.omg.org/Deployment Deployment.xsd">
<name>ORBSvcConf</name>
<value>
<type>
<kind>tk_string</kind>
</type>
<value>
<string>Foo</string>
</value>
</value>
<value>
<type>
<kind>tk_short</kind>
</type>
<value>
<long>123</long>
</value>
</value>
</Deployment:Property>

The model in 6.10.8 has a containment association with a cardinality one
for value. The schema generated from that model doesn't match the semantics
in the model. This is just an example. There are a lot of elements where
the semantics imposed by the schema are different from what is described in
the model. I am curious as to why this is allowed. It can be easily fixed
by changing the schema to:

<xsd:complexType name="Property">
<xsd:sequence minOccurs="0" maxOccurs="1">
<xsd:element name="name" type="xsd:string" />
<xsd:element name="value" type="Deployment:Any" />
<xsd:element ref="xmi:Extension" minOccurs="0" />
</xsd:sequence>
<xsd:attribute ref="xmi:id" use="optional" />
<xsd:attributeGroup ref="xmi:ObjectAttribs" />
</xsd:complexType>
<xsd:element name="Property" type="Deployment:Property" />

Issue 3: Component Factory Lacking
Both the SCA and CCM has the capability of expressing factory concepts in
the assembly description. The factory concepts allow for the deployment
machinery to create components in the assembly from a component factory
that is a component of the assembly or service of the node/device.

Recommendation
Update SubcomponentInstantiationDescription to add an optional element that
can be used to reference a component in the assembly or a referenced
service that is a component factory. This new element should also contain
optional properties that can be used by the factory for component creation

Issue 2: AssemblyController Component Lacking
Lacking the capability of expressing a AssemblyController component in a
assembly component description. One use case is the SCA and the SWRadio
spec. The assembly controller is the component in assembly that is the main
controller and the Application proxy communicates with when specified.
Recommendation:
Add an optional AssemblyControllerComponent Element to the
AssemblyComponent description. The AssemblyControllerComponent Element
definition contains an attribute, which identifies the component in the
assembly that is the assembly controller component. Add
AssemblyControllerComponent Element as subsection to 6.4. Update section
6.4.9.1 figure. Update PSM XML for ComponentAssemblyDescription.

This also allows PSMs to constraint the definition to always none or
mandatory for a assembly controller.

Issue 1: New Requirement Satisfier Type
Unable to express a node/device characteristic as a set where each item is
an instance of the same characteristic class. For example, node/device may
have many runtimes, libraries, communication paths/QoS. One needs to be
able to express a characteristic such as runtime or library as one
characteristic with multiple instances, where each instance contains the
same requirement satisfier definition. For example, for runtime or library
characteristic each instance in the set would contain name and its value
and version and its value, as a set of instances of a characteristic that
contains name and version in this case. This would allow one to state the
set of libraries or runtimes in a consistent manner all belonging to the
runtime or library characteristic.

Right now for the D & C, I would have to define a RequirementSatisfier as
RunTime1 and RunTime2 instead of RunTime. I am able to express a
requirement as RunTime but this will not be matched against RunTime1 and
RunTime2.

Potential Recommendation
Add definition for RequirementSetSatisfier that contains 1 one to many
RequirementSatisfier. The constraint being the RequirementSatisfier
definition has to be the same except for the values.

Appears that an abstract definition is needed for the
RequirementSetSatisfier and RequirementSatisfier since a node or device can
have either or both of them, and they have common attributes (name,
resourceType).

This added capability would also cause changes to the update operation to
support this information.

The multiplicity of the stereotype-property Target::SharedResource::resourceUser is [0..*] in the overview picture, but [1..*] in the textual description. There seems to be no reason why there should be at least one user for a shared resource.

Monitor or Verify Lacking in Deployment Process: In order to make the deployment process more perfect,this specification adds deployment planning before real deployment process begins,the monitoring should be added after the deployment as well.Only implementing deployment does not means the end for an application,it will be wonderful if monitoring the effects of the performed deployment on the target system which may be chosed from several valid deployment plans.This will also be helpful to assess which choice of deployment plans is a better one and to support dynamically upgrade of existing component versions. Recommendation: Monitoring should be added to the deployment process(chapter1.3) and Monitorer should be conrrespondingly added to the Actor (chapter 7). Monitoring may involve many things here we can choose some tightly correlated to deployment.Monitoring the existing application may be easy.During the perfomation of current deployment plan,the deployer can use deploy tools to get CPU or memory's running state of each node in the domain and the network traffic may also be considered.Add some corresponding xml files or elements to describe them will be ok.But for monitoring of dynamically upgrating,we consider a example that a component will be replaced by a set of its versions encapsulated in a wrapper.During the monitoring process,the wrapper is responsible for "hiding" from the rest of the system the fact that a given component exists in multiple version and the role of the wrapper is to relay to all component versions each invocation that it receives from the rest of the system then to propagate the generated results to the rest of the system.We can authoritate different operations to different versions and only the result of the authoritative version will be propagated outside while results of other versions will be logged for comparison of their perfomance?correctness and reliability.After some given time,assesses a better version for use and removes others.There are other algorithms to find out.

In section 6.8.2.2 (ArtifactDeploymentDescription attributes),
the text for the "node" attribute explains,

node: String
The name of the node where the artifact is to be installed.
If blank, the node is implied by the InstanceDeploynment-
Description parent.

The empty string should not have a special meaning. Rather,
the attribute should be optional.

Proposed resolution:

Change the attribute to have 0..1 multiplicity, and change
the text to read as follows:

node: String [0..1]
The name of the node where the artifact is to be installed.
If missing, the node is implied by the InstanceDeployment-
Description parent.

For components created at runtime instead of configuration time, they cannot utilize advantages of Assembly-configurator infrstructure. Those components must using ad hoc Naming or ior to locate other components or assemblies.

Resolution:

When a component is created, assembly object call object configurator of type Components::Configurator's mothod to connect component to other components or assemblies:

void configure (in CCMObject comp)

Some extension must add to deployment descriptors, assembly object using these information to connect newly created component to other components or assemblies.

Replace the following text in formal/02-06-05 on page 7-14

<!ELEMENT connections
( connectinterface

with

<!ELEMENT connections
( connectinterface

Add the following text in formal/02-06-05 on page 7-14

<!ELEMENT connectcomponentinstantiation
(homeplacementref
, (connectinterface

The capability is lacking to indicate if the capacity model for a capacity
is managed by the node/device or not. The Deployment and Configuration
specification should not constraint System Architecture/Developers from
architecting/implementing the system but be flexible. The OMG SWRadio spec
allows for this flexibility. One Example use case of this is multiple
domains trying to reserve capacity from the same device.

Recommendation
At a locallyManaged boolean attribute to the RequirementSatisfier.
Add a new interface called CapacityManager or CapacityAllocator. This
allows a node/device to realize this type of interface. A node/device would
only have to support this interface if it had capacities that are locally
managed. Recommendation for the interface definition is based upon the
SWRadio DeviceComponent allocateCapacity and deallocateCapacity operations

Only the DomainApplicationManager and NodeApplicationManager have a startLaunch, on both classes this leads to a call to a DomainApplication/NodeApplication but this has no name in figure 9.1. We propose to seperate construction with the real functionality. Add startLaunch also to DomainApplication/NodeApplication which means adding it also to the application interface

On page 51 (section 7.6.1) there is inconsistency between TargetManager interface depicted in figure in subsection 7.6.1.1 and operations' description in subsection 7.6.1.2. In the figure there is 'commitResources' operation whereas in the description there is 'createResourceCommitment' operation. Which one is correct? Just to mention that on page 121 (section 9.10) in figure 9.1 - Executor in Action there is invocation of commitResources operation.

The multiplicity of stereotype-property Target::CommunicationPath::connectedNode is [1..*] in the overview picture, in textual description it is [*]. The multiplicity of [1..*] makes more sense, since the source for the derived property is
self.interconnect.connectedNode, where CommunicationPath::interconnect has multiplicity [1..*] and Interconnect::connectedNode has multiplicity [1..*], so there should be at least one connectedNode.

The name and description of stereotype-property Components::Port::UID seems odd. Other stereotypes have a UUID property, providing a unique identifier. The description "The primary type of the port." does not seem to fit.

The property Components::ComponentImplementation::capacity is of type Sequence(Capacity). The type Capacity is not existent, neither in the UML meta-model nor in the DEPL specification. It is most likely (also judging from the stand-alone PIM meta-model) that the type Capability was meant here. So the property should be exchanged by the property:
Components::ComponentImplementation::capability(Capability)

In the overview diagram the property Components::Component::ownedPort has the multiplicity [1..*] in textual description, however, the multiplicity is [*]. There seems to be no reason why a component should have at least one port.

Supports For Many Applications: The current specification is more and more consummate for deploying an application.In the case of actual distributed environments such as embedded system,several applications running at the same time may be more hopeful.Although deploying several applications involves more things to consider,firstly we can add an ApplicationController component.The application controller is the component that coordinates and communicates with each application. Recommendation: The application controller component can be considered an assembly-based component in which each assembly component represents an application.Add an optional ApplicationControllerComponent Elment to the assemblyComponent description which contains an attribute that identifies the component is the application controller component.The corresponding PSM XML should be updated for assemblyComponent description.In implementation,an ApplicationFactory should be added which is reponsible for creating applications.

The property Components::ExternalReference::location is of type URL. This type is neither defined in the UML meta-model, nor in the DEPL profile. The type should be String and maybe a constraint or regular expression should be defined to restrict the value to a URL.

The satisfierPropertykind should not be an enumeration type. This does not
allow for growth or expressing other capacity models. This issue was
discuss at the St. Louis meeting
Recommendation
Is to change from a enumeration type to string but keep the enumeration
list as well-define string values along with their meaning.