homepage

This RFE proposes two new methods "to_signed" and "to_unsigned" on 'int' objects and on the numbers.Integral ABC.
Semantics (and number.Integral implementation):
def to_unsigned(self, bits):
 "Convert this integer to its unsigned two's complement equivalent for the given bit length"
 if self.bit_length() >= bits:
 raise ValueError("{} is too large for {}-bit two's complement
precision".format(self, bits))
 if self >= 0:
 return self
 return 2**bits + self # self is known to be negative at this point
def to_signed(self, bits):
 "Convert an integer in two's complement format to its signed equivalent for the given bit length"
 if self < 0:
 raise ValueError("{} is already signed".format(self))
 if self.bit_length() > bits:
 raise ValueError("{} is too large for {}-bit two's complement
precision".format(self, bits))
 upper_bound = 2**bits
 if self < (upper_bound / 2):
 return self
 return upper_bound - self
To add these methods to numbers.Integral, a concrete numbers.Integral.bit_length() operation will also be needed. This can be implemented simply as:
def bit_length(self):
 return int(self).bit_length()
(Initial concept from this python-ideas thread: http://mail.python.org/pipermail/python-ideas/2011-December/012989.html)

Instead of requiring bit_length(), couldn't you simply compare self to 2**(bits-1) (for to_unsigned) and 2**bits (for to_signed)?

The 'self.bit_length() >= bits' condition for to_unsigned doesn't look right to me.
E.g., if bits == 32, I'd expect the acceptable range of values to be range(-2**31, 2**31)---i.e., including -2**31, but excluding 2**31. But the ValueError for 'self.bit_length() >= 32' means that -2**31 is excluded.

On the feature request itself, I have to say that I'm unconvinced. Doing x % (2**32) seems good enough to me, in situations where you don't need the bounds checking. (And it's not clear to me that needing the bounds checks is the more common situation.)

> On the feature request itself, I have to say that I'm unconvinced.
> Doing x % (2**32) seems good enough to me, in situations where you
> don't need the bounds checking.
Ah, I didn't know that modulo worked for that. Nice trick.

I'm for closing this issue. This is not the common operation.

If you open the can of worms, other features will be requested like uint32+uint32 which would silently overflow.
IMHO it would be better to have a package (on PyPI) providing integers of fixed size implementing two's complement arithmetic: int8, uint16, etc. I failed to find an existing module.