It appears that a <> Big_int.zero_big_int blows up when
a is zero. Weird. Eliminating the comparison doesn't hurt
correctness, and is probably more efficient too, as negating
zero should be cheap!
Comparing from 8d6841e820fa1a4ff3a4cd7158ea0a491ee48b77 to 5f9f320a0646a628c30d62f2e51019d2f827d089:
The produced WebAssembly code seems to be completely unchanged.
Cf. fixes dfinity/motoko-base#653
It appears that
a <> Big_int.zero_big_int
blows up whena
is zero. Weird. Eliminating the comparison doesn't hurt correctness, and is probably more efficient too, as negating zero should be cheap!@timohanke feel free to review too!