Yes :)! So I mention PEP 3107 (https://www.python.org/dev/peps/pep-3107/) in the thesis. This allows type annotations in the function signature. Cannoli leverages both type annotations in function signatures and assignments to output optimized code. Other projects like Pythran also use type annotations.
PEP 3107 does say:
> By itself, Python does not attach any particular meaning or significance to annotations.
However, I think this will change especially as more projects begin to outperform CPython. In the "Results > Object Optimization" section of the thesis paper, I cover using these very type annotations to optimize the code.
The biggest problem with Python annotations right now is that they don't really mean anything. Nothing is really enforced so it is totally valid to have 'x : int = "string"'. The compiler would have to just ignore this annotation since it was provided the wrong data. This could also be difficult to identify if a variable was being used and its type mislabeled. So it's not perfect but I think it's a step in the right direction.
PEP 3107 does say:
> By itself, Python does not attach any particular meaning or significance to annotations.
However, I think this will change especially as more projects begin to outperform CPython. In the "Results > Object Optimization" section of the thesis paper, I cover using these very type annotations to optimize the code.
The biggest problem with Python annotations right now is that they don't really mean anything. Nothing is really enforced so it is totally valid to have 'x : int = "string"'. The compiler would have to just ignore this annotation since it was provided the wrong data. This could also be difficult to identify if a variable was being used and its type mislabeled. So it's not perfect but I think it's a step in the right direction.