Strategy

A strategy can be written to improve test coverage according to some
chosen measure.  A strategy is (optionally) used by the tester, pmt,
to pick the next enabled action in the model to execute.  If no
strategy is provided, the tester picks an enabled action at random.

A strategy is written in its own module.  To use a strategy, name that
module on the pmt command line with the -g or --strategy option.

PyModel already includes some strategies you can use (besides the
default random strategy): ActionNameCoverage and StateCoverage (in the
pymodel directory).

A strategy module provides a function selectaction (or select_action
or SelectAction) that takes a list of tuples: enabled actions,
arguments, and their next states.  It returns one chosen enabled
action with arguments.  Each strategy chooses this return value
improve a particular coverage measure.

Unlike models or steppers, you do not need to write a different
strategy for each implementation.  Some strategy modules can be used
with any model; the provided ActionNameCoverage and StateCoverage
strategies are examples.  

It is also possible to write a strategy that is adapted for a
particular model.  For example, you might want a strategy that
enforces an "operational profile" where each action appears in traces
with a particular frequency.


Revised July 2011