After the segments of code that are computationally-independent
are identified, they are separated from the main task program and
placed in parallel functions. A parallel function is coded as a normal
C function that follows several rules required for correct operation
with MTF. Besides to data independence, there are rules for:
- Parallel functions
- Calling other functions
- Separate storage for separate modules
- Passing data
- Input and output
- Exception/signal handling
- Function termination