[Python-Dev] Unifying Long Integers and Integers: baseint (original) (raw)

Guido van Rossum guido at python.org
Thu Aug 12 05:03:11 CEST 2004

I would like to urge caution before making this change. Despite what the PEP may say, I actually think that creating a 'baseint' type is the WRONG design choice for the long term. I envision an eventual Python which has just one type, called 'int'. The fact that an efficient implementation is used when the ints are small and an arbitrary-precision version when they get too big would be hidden from the user by automatic promotion of overflow. (By "hidden" I mean the user doesn't need to care, not that they can't find out if they want to.) We are almost there already, but if people start coding to 'baseinteger' it takes us down a different path entirely. 'basestring' is a completely different issue -- there will always be a need for both unicode and 8-bit-strings as separate types.

Not so sure. I expect that, like Jython and IronPython, Python 3000 will use unicode for strings, and have a separate mutable byte array for 8-bit bytes.

In Python 3000 I expect that indeed somehow the existence of long is completely hidden from the user, but that's a long time away, and until then baseinteger might be a better solution than requiring people to write isinstance(x, (int, long)).

--Guido van Rossum (home page: http://www.python.org/~guido/)

More information about the Python-Dev mailing list