float error

log(1.20515621346031e463) outputs INF in normal precision, NAN in extended precision (tried with up to 1e6 extra digits).

Thanks for the report, I'll have a look.

I've tracked this down now. It's normal that you can INF in double precision mode: the maximum value that can be represented here is around 1e300. In Extended precision, you get 1066.something which is the correct value. However, there was a bug in Arbitrary precision mode that caused NAN to be returned. I'll fix this for the upcoming 6.03 update.