On Feb 2, 2015, at 10:03, MRAB email@example.com wrote:
On 2015-02-02 16:45, Chris Barker wrote:
On Mon, Feb 2, 2015 at 7:27 AM, Thomas Güttler <firstname.lastname@example.org mailto:email@example.com> wrote:
Well, inf is supported in floats because it is supported in the native machine double. I suspect that adding inf and -inf (and NaN) to integers would end up being a pretty heavy-weight addition.
Python 3's int isn't limited to a native machine integer, so it's not impossible to extend it to include infinity...
I doubt it would be a serious performance problem if done right. The big question is how it interacts with other types. (Those confusing corner cases already brought up...)
On the other hand, my quicky off the cuff implementation of InfDateTime is, in fact, a universal infinity -- i.e. it compares as greater than any other object, not just a datetime object.
Including float('inf') and float('nan')? That doesn't seem like a selling point...
I did that more because it was easy than anything else, but perhaps it could be generally useful to have a Inf and -Inf object, kind of like None.
Or it would just create a huge set of confusing corner cases :-)
And who knows what it could break!
I would much rather see a infinite datetime object that a big discussion about the implications of a generic infinite object -- it is focused, has proven used cases, and wouldn't impact anything beyond datetime.
Python-ideas mailing list Pythonfirstname.lastname@example.org https://mail.python.org/mailman/listinfo/python-ideas Code of Conduct: http://python.org/psf/codeofconduct/