Document changes to sensor value coercion #99
Conversation
| ==================== ========================== | ||
| :class:`int` :class:`numbers.Integral` | ||
| :class:`float` :class:`numbers.Real` | ||
| :class:`~.Timestamp` :class:`numbers.Real` |
There was a problem hiding this comment.
It's great that ~.Timestamp just knows to go specifically to aiokatcp.core.
| - The :class:`Address` constructor now raises :exc:`TypeError` | ||
| if the first argument is not an IP address object. |
There was a problem hiding this comment.
I know it's unlikely that the first argument to core.Address will change from being host. Still, is it worth explicitly stating which arg it is?
- e.g. "[...] if the ':attr:
host' argument is not an IP address object." (?)
There was a problem hiding this comment.
Eh, six of one, half a dozen of the other.
| - When setting the value of a sensor, it is coerced to | ||
| an instance of the sensor type (see :ref:`sensor-value-coercion`). Reading | ||
| back the sensor value will give this coerced object instead of the original. | ||
| This only applies if the value was not already of the expected type. |
There was a problem hiding this comment.
This is about as perfectly succinct as it gets. I've written and deleted my suggestions a few times; anything I think of makes it more convoluted/pedantic.
| Other changes: | ||
|
|
||
| - Use :func:`typing.get_args` instead of an undocumented API. |
There was a problem hiding this comment.
Genuine question, why is it necessary to indicate this? I see the original change was from b3d0f7c
There was a problem hiding this comment.
I considered not mentioning it does it doesn't affect how users use the package, but it is a change that users might want to know about if some future Python version removes the undocumented API and the old version stops working.
No description provided.