This can be interesting for the "benchmark" to compare other libs with the native bigints. As the results are very close when your lib is using natives.
And only Division is two times slower, this change should fix it.
Coverage decreased (-0.2%) to 95.127% when pulling 38206db40e7904003d92a2e105bf9acbcc9956c7 on Yaffle:master into 4a5b73a2d6a9a7a8aa78e4a095cc55e4d498f691 on peterolson:master.
This can be interesting for the "benchmark" to compare other libs with the native bigints. As the results are very close when your lib is using natives. And only
Division
is two times slower, this change should fix it.