[v3] libstdc++/39629 (final bits)

[Paolo Carlini]

Hi,

the below resolves the PR to zero -m32/-m64 fails. In a couple of cases
a formula was incorrect + we had equality comparison operators for the
distributions, not required by the draft Standard and really not making
sense because involving equality comparison of floats. At some point we
may want a robust, proper way to test the serialization of the
distributions, closely following the draft standard (26.4.1.5/5).

Tested x86_64-linux, -m32/-m64, committed to mainline.

Paolo.

////////////////////
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: CL_39629_part2
URL: <http://gcc.gnu.org/pipermail/libstdc++/attachments/20090409/4dbac731/attachment.ksh>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: patch_39629_part2
URL: <http://gcc.gnu.org/pipermail/libstdc++/attachments/20090409/4dbac731/attachment-0001.ksh>