Closed marco6 closed 8 months ago
This PR adds a fast path for very big int to float conversion that always get converted to ±Inf.
Was this change opportunistic, or did the logic appear in a profile? I'm curious what workload would make heavy use of numbers larger than the universe.
Opportunistic: we had to do that for concerns not related to performance and it felt in line with the previous optimization.
In any case, looks good. Thanks for the contribution.
This PR adds a fast path for very big int to float conversion that always get converted to
±Inf
.