Paul Moore writes:
And as soon as we start considering integer division, we're talking about breaking a *vast* amount of code.
Yeah, I'm ok with *not* breaking that code.
You may have misunderstood me - when I said "integer division", I meant "division of two integers",
Just to clear it up, I understood your point correctly. "I'm ok with *not* breaking that code" means "I'm talking about the mythical Python 4.0, obviously we can't change the error raised by 1 / 0".
My *only* concern with the points you and Ben were making was that you seemed to be suggesting changes to the division operator and ZeroDivisionError,
Once again, I am quite ok with *not* breaking all that code. My point about the inconsistencies is not to suggest fixing them. I'm quite sure that pragmatically we can't fix *all* of them, and most likely we'd have to go slow on fixing *any* of them. Rather that the whole float situation is so visually messy that we should leave it alone -- although it probably works fine in practice until you have need for NumPy for other reasons.