mpf_get_string still has limitations on 5.0.1 with MinGW.

Torbjorn Granlund tg at
Sun Dec 12 09:06:24 CET 2010

Sergio Abreu <dosergio at> writes:

  Almost everything works fine, but it seems that for constants that use
  more than 32 bits bad things still occur.
  For instance, mpf_get_str only work with values under 2^32.
  If GMP is for "unlimited arithmetic", how can this happen ???
  Could I have done something wrong during set up ?
You seems to have struck an extremely serius bug in GMP.  I thought it
could handle numbers with billions of bits, but apparently the limit is
just 32 bits, as determined by your testing.

Please send us a bug report about this problem, so that we can reproduce
and fix it.  But please don't forget to first read the manual's very
short chapter on how to report GMP bugs:

  Other point is: gmp_printf only print integer portions for mpq_t
  variables ? Isn't mpq_t intended for fractional values ?

Then gmp_printf is completely broken!  Strange that nobody else noticed
in all these years!  But wait, 'make check' works for me, and that tests
printing of rationals.  Now I am really confused.  I suppose you need to
send a bug report abut this problem too.


More information about the gmp-bugs mailing list