True that, but I thought I'd also note that floats are represented as compactly as possible starting in 3.1.1. > Issue #1580: On most platforms, use a 'short'

True that, but I thought I'd also note that floats are represented as compactly as possible starting in 3.1.1. > Issue #1580: On most platforms, use a 'short' float repr: for a finite float x, repr(x) now outputs a string based on the shortest sequence of decimal digits that rounds to x. Previous behaviour was to output 17 significant digits and then strip trailing zeros. Another minor difference is that the new repr switches to exponential notation at 1e16 instead of the previous 1e17; this avoids misleading output in some cases.