#define DOCSTRING "\n\
$Id: float2rat.doc 2337 2006-04-03 20:42:09Z bremner $\n\
\n\
Converts floating point coefficent $f$ to rational by the \n\
simple expedient of outputing 10^k*f/10^k for appropriate \n\
$k$. Does no reduction of numbers.  In particular this may cause overflow in \n\
old versions of lrs input (and I'm not about cdd).\n\
"