[PATCH] Some uses of M_PI in libstdc++ testsuite

[Paolo Carlini]

David Billinghurst wrote:
> Following http://gcc.gnu.org/ml/libstdc++/2009-05/msg00091.html there
> are a couple of uses of M_PI in the libstdc++ testsuite.  Defining M_PI
> seems safe, as these are testsuite cases.
Nope, let's not fiddle with M_PI, which is not standard: some people may
be fooled again into thinking this is a portable macro. Let's have
instead an unconditional:

  #define _Pi  3.1415926535897932384626433832795029L

and of course s/M_PI/_Pi in the testcase itself. Preapproved.

Thanks,
Paolo.