moonbitlang / core / 7059

88%

push

github

fix(float): Show for Float prints the float's shortest digits

Float::to_string was self.to_double().to_string(): the shortest digits of
the double the float widens to, so Float(0.1) printed 0.10000000149011612
and Float(3.2) printed 3.200000047683716.

Port f2d of Ryu (the ryu crate 1.0.23) for the digits and lay them out by
the rules Double uses (ECMAScript's Number::toString). NaN, the infinities
and the zeros print as before.

Checked against the ryu crate on 3181 floats (powers of two and ten with
neighbours, subnormals, exact ties, random bits): the digits, that the text
reads back as the same float, and that Double prints the same decimal the
same way. Snapshots that pinned the old output are updated.

77 of 78 new or added lines in 2 files covered. (98.72%)

16837 of 18813 relevant lines covered (89.5%)

360014.52 hits per line