Problems with atan(-2.56B0) on on emachine but not another



If I try to calculate atan(-2.56B0) on one of my machines maxima goes off into an infinite loop.
I can, howvwer, calculate atan(2.56b0) and atan(-2.56e0)

The problem machine is an SGI Irix6.5 Origin 300 with clisp-2.28 and Maxima 5.9.0pre-cvs.

It doesn't happen on a PC with either:
 - cygwin/clisp-2.28/maxima CVS
 - windows/gcl-2.50/maxima-5.9.0

I will try and investigate this some more.