Don't allow float(quantity) and int(quantity)

[twmr]

No description provided.

[apdavison]

Please can you explain the motivation behind this pull request?

[twmr]

Sure, the problem is that float(quantity) returns a float, which is a bit surprising.

With my commit you'll get an exception if you do print("%f" % quantity). I think this is better than implicitly stripping the unit in float(quantity).

[apdavison]

I guess I don't see it as surprising or implicit; float(quantity) is explicitly asking to discard the units, and retain only the value in the current units, i.e. very similar to quantity.magnitude.

Does anyone else have an opinion on this?

[bjodah]

I can see arguments both for and against.
I would propose making the stricter behaviour proposed here opt-in through an environment variable (e.g. QUANTITIES_STRICT=1), even though that would essentially be akin to a module level global variable. Otherwise there need to be a context object being moved around.

But even then this patch should be updated to allow conversions to float when the dimensionality is quantitites.dimensionless, right now it does not (on this branch):

>>> float((3*pq.m/pq.km).rescale(pq.dimensionless))
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/bjorn/vc/python-quantities/quantities/quantity.py", line 417, in __float__
    raise TypeError("Quantities can't be converted to floats")
TypeError: Quantities can't be converted to floats

[K3UL]

I also do not find any surprise in having float(anything) returning a float. I actually NEED this to happen whenever I am doing some billing. Like we have a rate of x euros per kWh and so I need to do € * kWh = €.

I think @bjodah's suggestion makes some sense, but I still don't see why I could not do float() and just get the value part, stripped of its unit.

[bjodah]

@K3UL my point is that there's already .magnitude.