j3-fortran/fortran_proposals

NINT and ANINT according to IEEE rounding

Closed this issue · 1 comments

tkoenig1 commented

It is a bit strange that, since F 2003, NINT and ANINT are defined to round away from zero. That is a bit strange, but nobody will change that.

I would propose to add two functions provisionally called IEEE_NINTand IEEE_ANINT which obey the current rounding mode.

tkoenig1 commented

Hm, looking at IEEE_INT, this is actually not needed.