Context. Consider the following piece of code.
from classes import typeclass
@typeclass
def render(data_value) -> str:
"""Pretty-print a value."""
@render.instance(int)
def _render_int(data_value: int) -> str:
return f'🔢 {data_value}'
render(True) == '🔢 True'
(this should be runnable as-is on classes 0.4.0.)
I would have expected this code to fail with a NotImplementedError because the bool case wasn't specified; but instead, that case is handled by the int implementation because
In [3]: issubclass(bool, int)
Out[3]: True
Decision. In an int implementation, recognize if the provided value is actually a bool and refuse to process that value.
Consequences. I have been writing in Python for quite a few years now and I might even have encountered this relationship between int and bool types before, but that is not quite a type of thing that I keep in my short-term memory to be able to instantly recognize.
I believe this change can help avert bugs which might be hard to trace otherwise.
Context. Consider the following piece of code.
(this should be runnable as-is on
classes 0.4.0.)I would have expected this code to fail with a
NotImplementedErrorbecause theboolcase wasn't specified; but instead, that case is handled by theintimplementation becauseDecision. In an
intimplementation, recognize if the provided value is actually abooland refuse to process that value.Consequences. I have been writing in Python for quite a few years now and I might even have encountered this relationship between
intandbooltypes before, but that is not quite a type of thing that I keep in my short-term memory to be able to instantly recognize.I believe this change can help avert bugs which might be hard to trace otherwise.