Reuse by Inheritance in Agent Programming Languages

From Actor Event-Loop to Agent Control-Loop

“…The do syntax allows to specify directly the name of the task type (e.g. line [18][19][20][21], along with parametersin that case, a new task object of that type is implicitly created. 4 So for each intention, a stack is used to manage the action rule block nesting.…”

“…The plan for living (lines 11-16) accounts for a sequence of subgoals: first think, then acquire forks, eat, and then release forks before starting again with the same activity. In the plan for acquiring the forks (lines [18][19], the agent interacts with the environment by performing the acquireRes actions in order. The shape of the resulting program is as simple as the one typically found in multi-threaded programming, even if here the control architecture is completely different.…”

“…Some syntactic sugar is provided to directly encode set of rules representing sequence of actions: they can be written as a chain of actions (omitting the condition part) using the semicolon separator, like a sequence of statements in the case imperative programs (e.g. lines [18][19][20][21]. 3 Actions can be external -i.e.…”

“…Otherwise, if it is an independent task, a new intention is created. In the act stage, an intention is selected to be the current intention currInt (line 5), using a round-robin schema to guarantee fairness, and then all actions that can be executed according to the rules of such an intention are collected (line 17) and executed, sequentially (line [18][19][20].…”

“…The always => A rule (e.g. lines [17][18][19][20][21][22][23] means that the action A can be always selected. Some syntactic sugar is provided to directly encode set of rules representing sequence of actions: they can be written as a chain of actions (omitting the condition part) using the semicolon separator, like a sequence of statements in the case imperative programs (e.g.…”