Modeling of Component Archictures inside modes & creating a Deployment for these architectures
Current status
Mode automata are used to express different operational modes of system based on some input. Currently, mode automata can be used to model sub-parts of a Component to define alternations in the behaviour of a Component. Deployments are then defined on the top-level Component Architecture and allow an allocation of Components up to the granularity of atomic Components or Components containing a Mode automaton. Hence, even with altering configurations of that Component, the deployment is always well-defined.
Requested Feature
An more general approach to the concept of modes would allow defining
modes also on an architecture level that allows to express much varying
system states. In line with this feature, it would be required to
express different deployments for each of these modes.
However the Deployment MM allows only references to Component
Architectures which makes it impossible to reference the
ModeComponentSpecifications that are Components. Here, a solution should
be found that allows defining “complete” logical system descriptions
within Modes while being addressable from deployment and scheduling MM
(amongst others).
Initial Analysis & Prototype “Solution”
Attached is a patch which redefines the ModeComponentSpecifications to
contain ComponentArchitectures instead of Components directly. With an
included minor modification, each ComponentArchitecture containted in a
Mode Automata is selectible for a deployment.
This approach however might break some assumptions and needs discussion.
Note: The patch is just a prove of concept and contains some minor “hacks” to make it work and also the functionality needs more refinement (e.g. What happens wit lower level mode automatons?)
(from redmine: issue id 2436, created on 2015-10-15, closed on 2016-07-18)
- Uploads: