Precision of mpf_class

[Edward Marti]

I'm using your GMP classes with the C++ class wrappers (#include <gmpxx.h>)
with g++. I noticed that the computer tends to print unlimited precision
with mpz_class and mpq_class, but it is severely limited with mpf_class.
When I use the setprecision and function in iostream (cout <<
setprecision(100) << my_mpf;), the precision only prints 20 digits after the
decimal. Adjusting the width didn't help either. Is there a way to print
many more digits of mpf_class directly through iostream? Or do I have to
save the digits to a string, and print the string?
 - Edward Marti