Bug with square roots
tg at gmplib.org
Wed May 9 11:23:09 CEST 2012
Emil Vladov <emil at emovla.com> writes:
The program that follows takes sqrt of a long integer, then takes
root(2). The resulting numbers are different (and both are wrong).
'make check' tests pass.
If a trivial program fails like you say, and 'make check' passes, then
it will hardly be of the same libgmp compile.
One thing seems clear: You're linking the trivial test to a badly
miscompiled library. If it is installed on your system, you should
Sometimes system policies bind to surprising library files, not the ones
one asks for. This might happen to your trivial test, or it might
happen to 'make check'.
More information about the gmp-bugs