Lecture 05

Behavioral Modeling Languages

categorized in different ways

• properties
• requirements : not the whole possibilities
• system scenarios
• analysis
• …

(Statemate = original formalization of statcharts)

Characteristics of Dynamic Languages

• Specification
  • unambiguity: there are no undefined or ambiguos (ie. that can be understood/interpreted multiple ways) parts in declaration (egyértelmű)
  • consistency: no contradicting requirements (ellentmondás mentes)
  • completeness: did not forget important parts
• Time
  • untimed
  • discrete: the range of time is taken from the natural (?) numbers.
  • continuous: the range of time is taken from the real numbers.
• Communication
  • synchronous: communication driven by a clock
  • asynchronous: the different processes are driven by a message queue
• Determinism
  • stochastic: distribution about probability selection
  • deterministic
• Causality
  • causal: from number of independent variables calculate an other/dependent one
  • non-causal
• Analysis
  • complete/incomplete: (free of flaws) a complete analysis answers yes or no, an incomplete analysis may answer yes, no, or I don't know, so it may not give you the answer
  • approximative/exact - Does the result have an error margin?

Property Specification Languages

• requirements: Either in structured text of with modeling notations.
• scenarios: specify permitted/forbidden executed paths (that is, a queue of states)

[ÁBRA - Szerintem ide am. nem kell ábra, Formális módszereken simán vettünk ilyeneket - FR]

State-based Languages

Like in a finite automaton : states, transitions, events, actions, etc. Hierarchy of states, history.

Data-flow Language

Processes with activities, communicating through channels (tokens/messages in queues), e.g.: Activity Diagrams, Petri Nets. The activity trigger is that there is data to be processed.

Event-based Rule Languages

• core concept is event (atomic, complex …)
• the system response(responds?) to events specified by rules; precondition/action specifies
• intended to be used by the domain user and not the software engineer

~controlled natural language

Agent-based languages

• real-life entities/agents with connections (e.g. FB users / profiles)
  • plus behaviour
  • plus space, mobility
  • plus environment - assumptions about the environment; eg. mobile phone in low temperature (???)

Continous-time Languages

• e.g. simulink, described with block diagram

Dynamic Metamodeling in DSE

• complete static metamodels with dynamic and execution trace metamodels. (log, that could replay the execution)
• UML specifications do this in plain English. [ábra a slide-ról]

Statecharts from Modeling Reactive Behavior

• state-based behavior modeling [definition]
  • static part is the possible states
  • dynamic part is the currently selected/active state
• interleaving: there is no way, that two systems do something at the same time, unless it is synchronized
• statecharts = states + transitions
  • concrete state: I'm 4 year-old
  • abstract state: since I'm 4, I'm a child (e.g. child-adult - senior abstract states)
  • Also, states can be placed in hierarchy [kép]