Glossary
- attached core
A core that does not construct middleware shared runtime objects itself. It attaches to already initialized shared memory and binds to the existing
MemoryLayoutinstance.- Application Cluster
- Cluster
- middleware cluster
A middleware cluster is a logical grouping of middleware endpoints running in one task context. Proxies and skeletons assigned to the same cluster share one incoming queue and are processed together by a cluster processing entrypoint.
- cluster connection
A cluster connection is the generated middleware object that routes messages from one cluster to another. It dispatches incoming messages to proxies and skeletons and provides the write path to the destination queue.
- Core
A CPU execution context inside one ECU instance. In multi-core setups, each core can host its own middleware cluster(s) and lifecycle loop.
- ECU
- Electronic Control Unit
An ECU is a vehicle control unit that runs one or more software components and may communicate with other ECUs and other local cores via middleware.
- global deployment
The model input that defines where middleware entities are deployed, for example queue names, ECU/core assignment, cluster IDs, and service-provider placement.
- integration and initialization
The practical setup steps required to make middleware communication work in an ECU:
Wire generated code into the build.
Create or attach shared memory.
Initialize allocators and queues.
Initialize cluster connections.
Run cluster processing cyclically.
- Message
The middleware transport unit exchanged between clusters through queues. A message typically carries header metadata (such as source cluster and service/member identifiers) plus payload bytes.
- Modeling the interfaces
The authoring step where communication APIs are described in IDL/model files. In this user guide it focuses on service interfaces.
- owner core
The single core responsible for constructing
middleware::shm::MemoryLayoutexactly once in shared memory during startup before other participating cores attach.- payload
The application data carried by a message, excluding protocol metadata such as headers, routing information, or control fields.
- Platform cluster entrypoints
Generated functions for the platform cluster lifecycle, typically including initialize, process, and timeout-update entrypoints. They are called by the application lifecycle to drive middleware routing and timeout handling.
- Provider
- Consumer
A provider (server, skeleton side) offers a service interface. A consumer (client, proxy side) uses that interface by sending requests, receiving responses, and subscribing to updates.
- queue
- Queues
Middleware FIFO buffers used for inter-cluster message transport. Each cluster typically has an incoming queue that is polled and dispatched during cluster processing.
- Service
- Services
A service is a middleware API contract containing methods, attributes, and events/broadcasts. Services are implemented by providers and used by consumers.
- service interface
- Service interfaces
Interface definitions used for service-oriented communication. They describe typed methods, attributes, and events independent of the low-level transport mechanism.
- SHM
The shared memory area used by middleware runtime objects such as queues, allocator instances, and synchronization primitives. One core usually initializes this region and other cores attach to it.
- Software Component
- SwC
A deployable software unit that owns application logic. SwCs use middleware through generated proxies/skeletons and wrapper code.
- startup barrier
A platform synchronization point used to signal that owner-core shared memory construction is complete and attached cores may safely bind to the shared
MemoryLayoutinstance.- task
- RTOS task
A schedulable execution context provided by the operating system. Middleware uses the current task identity to associate application clusters with their execution context. This may be an RTOS task handle, a POSIX thread ID, or a fixed value on a bare-metal single-task system.