procgraph 1.1 documentation

FAQ

Contents

FAQ

Why no feedback loops?

The main reason is that, for arbitrary blocks, it is hard to reason about a consistent semantics.

For example, consider the simple arrangement with one block with a closed loop:

         ________
u ----> |        |
        | block  | -----> y
 -----> |________|  |
|___________________,

At time 0, the signal u is defined, but y is not. That means that the code in block must have a special case for this situation. It’s unclear at the moment how to design an API that takes into account this special case.

I felt that any choice would add a great deal of complexity, and would make ProcGraph into a full-fledged framework rather than a quick and non-obtrusive tool for rapid prototyping.

Suggestions are welcome...

Contents