According to the documentation, Buffer Sensor takes float[ ] or Single[ ] as observations. Any way for it to take int[ ] instead?

Looking at the source code, floats seem hardcoded into the BufferSensor. But it also looks like it could be easy to edit and make my own IntBufferSensor just by replacing “float” with “int” in a few places - would this work?

I am doing this because my data is discrete and not continuous.

Update: Changing BufferSensor to use ints wouldn’t work as it goes even deeper, with ObservationWriter only dealing with floats as well. Do I have an incorrect assumption? Is using floats fine, even with discrete data values, since it is state-based and floats vs. ints doesn’t matter?

Update 2: Looking at the VectorSensor source, it looks like even then with its AddObservation(int) function, it immediately just implicitly converts it to a float. So I guess my current question is still whether I have an incorrect assumption. Is the learning state-based such that using integers vs. floating points is the same?