You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The current quickcheck tests only produce behaviours that should typecheck. It would be cool to have a generator that produces behaviours that should fail to typecheck.
One possible approach would be to introduce some known bad mutations into the behaviours produced from the current generator. Some ideas:
Multiple writes to the same storage location
Duplicated behaviour names
Duplicated names for storage entries in the constructor definition
Duplicated names for calldata args
Many bad specs are impossible to generate at the level of the Refined AST (e.g. we cannot produce an expression adding a bool to an int), so perhaps in the future we may wish to investigate a generator that produces act as a string directly which would give us a lot more freedom in producing incorrect acts.
The text was updated successfully, but these errors were encountered:
The current quickcheck tests only produce behaviours that should typecheck. It would be cool to have a generator that produces behaviours that should fail to typecheck.
One possible approach would be to introduce some known bad mutations into the behaviours produced from the current generator. Some ideas:
Many bad specs are impossible to generate at the level of the Refined AST (e.g. we cannot produce an expression adding a bool to an int), so perhaps in the future we may wish to investigate a generator that produces act as a string directly which would give us a lot more freedom in producing incorrect acts.
The text was updated successfully, but these errors were encountered: