The Dispatch Pipeline

An object adapter dispatches incoming requests using its dispatch pipeline. A dispatch pipeline implements the Dispatcher abstraction using a tree-like structure of dispatchers.

The leaves of this tree are servants, while intermediary nodes in this tree are middleware and an internal helper class, ServantManager.

Servants, middleware, and even the internal ServantManager class, all implement the Dispatcher abstraction.

Dispatch Pipeline Decision Tree

An object adapter gives all incoming requests to its dispatch pipeline, which performs the following steps:

• Traverse the middleware installed on this object adapter. A middleware can occasionally fulfill the request itself by returning a cached response, or more commonly, by returning an error.

• Look for the identity and facet carried by the request in the Active Servant Map (ASM). If the ASM contains an entry, dispatch the request to the corresponding servant. (End of tree).

• Look for a default servant registered for the category component of the identity carried by the request. If found, dispatch the request to this servant. (End of tree).

• Look for a default servant registered with the empty category. If found, dispatch the request to this servant. (End of tree).

• Look for a servant locator that provides a servant. We only mention servant locators for completeness, as you should not use servant locators in new applications.

• If none of the previous steps found a servant:
  - throw FacetNotExistException when the ASM contains a servant with a matching identity, but a non-matching facet
  - otherwise, throw ObjectNotExistException

As far as the dispatch pipeline is concerned, servants are just dispatchers – they accept requests and return responses. The dispatch pipeline does not rely on APIs provided by servants beyond the Dispatcher API.