rint() bug

[Luoqi Chen]

Does anyone know about this particular bug (and have a plan to fix it)?
http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=4831589

I haven't checked any 32-bit system, on 64-bit systems, gcc is generating
SSE instructions for floating point operations, so the i387 version of rint()
is no longer used and the C version is. Linux (RHEL4 at least) has the
same problem.

The fix would be very straightforward: cast lrint() back to double.

-lq