deg2rad unexpected behavior

[AllenDowney]

The wrapped version of np.deg2rad does not do what I expect if the argument is a dimensionless Quantity.

Expected: I think a dimensionless argument should be assumed to be in degrees, since that's the normal behavior of this function given a simple number (not a Pint Quantity).

Actual: It treats the argument as if it is in radians:

Example:

angle = 45 * ureg.dimensionless
np.deg2rad(angle)

The result is 45 radians, not 0.78 radians, as I expected.

Is the current behavior as intended?

[hgrecco]

We have not discussed this as using deg2red and similar is unnecessary with Pint. I open to suggestions in terms on expected behaviour and implementantion.

[AllenDowney]

For expected behavior, I would think that deg2rad should:

• If the argument is in degrees, convert to radians (current behavior)
• If it's already in radians, leave it alone (current behavior)
• If it's not a Quantity, convert to radians (current behavior)
• If it's dimensionless, convert to radians (not current behavior)

I think this last one is correct because degrees are dimensionless, so if the argument for this function is dimensionless, we should assume it's degrees (that is, we should assume the caller got it right because we don't have ground to think they're wrong).

For implementation, it looks like you already have a special case for radians:

def __ito_if_needed(self, to_units):
        if self.unitless and to_units == 'radian':
            return

        self.ito(to_units)

Maybe there should also be a special case for degrees?

[hgrecco]

Revisit after #905

[jthielen]

This issue was left unchanged by #905, but it has been made easier to fix. With that in mind, is @AllenDowney's proposed expected behavior what Pint should go with?