CGPM is written very defensively in that methods throughout perform significant error checking on their inputs (sometimes as assert but typically by raising a ValueError.
Advantages:
Disadvantages:
This issue is partly (but not entirely, since we check the relationships between arguments at runtime) related to python being a dynamically typed language. In practice, such problems are typically solved using some form of static analysis and program verification (e.g. predicate transform or abstract interpretation) except I wonder whether how well these tools would work for a complex set of predicates such as especially equality between data structures that are specified only at runtime.
Either way, we need to determine whether CGPM should keep these checks. If yes, it might be worth thinking of some strategies to reduce the complexity of these useful checks.
CGPM is written very defensively in that methods throughout perform significant error checking on their inputs (sometimes as
assertbut typically by raising aValueError.Advantages:
Disadvantages:
This issue is partly (but not entirely, since we check the relationships between arguments at runtime) related to python being a dynamically typed language. In practice, such problems are typically solved using some form of static analysis and program verification (e.g. predicate transform or abstract interpretation) except I wonder whether how well these tools would work for a complex set of predicates such as especially equality between data structures that are specified only at runtime.
Either way, we need to determine whether CGPM should keep these checks. If yes, it might be worth thinking of some strategies to reduce the complexity of these useful checks.