[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: occam REPL

> The idea of the REPL is what we call a test harness. Connect it to the
> input
> ports and outputs ports, generate data using the right protocol on the
> inputs and connect the outputs to a data gatering task with filing and
> printing. This works fine for dataflow type application tasks, but is less
> trivial for testing with real-time I/O and e.g. shared resources.

The key to distinguish occam/CPA is, I think, MULTIPLE inputs and outputs.
You needn't "connect the outputs to [one] data gat[h]ering task" but
possibly to several; similarly some of the inputs may be rare or control
inputs; ultimately the REPL has one input source (which can be
demultiplexed with tags) and one (multiplexed?) output sink, but pretty
fair generality can be achieved by such a test harness. Real-time behavior
can be simulated by generator nodes, timer ALTs and time scaling. Shared
resources are what we are trying to wean people away from --- many if not
most problems don't need them (other than channels and resource
subdivision at initialization).

Larry Dickson