In perl.git, the branch smoke-me/jhi/ldblm has been created
<http://perl5.git.perl.org/perl.git/commitdiff/5f9a4aa2f060bffb804deebd1603f0131c25cf3a?hp=0000000000000000000000000000000000000000>
at 5f9a4aa2f060bffb804deebd1603f0131c25cf3a (commit)
- Log -----------------------------------------------------------------
commit 5f9a4aa2f060bffb804deebd1603f0131c25cf3a
Author: Jarkko Hietaniemi <[email protected]>
Date: Tue Nov 11 18:36:57 2014 -0500
Special case exp(1) for ppc64-linux.
M t/op/sprintf2.t
commit 2a1abfde63476edc9b2e271da5ae19dedccaf4e2
Author: Jarkko Hietaniemi <[email protected]>
Date: Tue Nov 11 18:27:51 2014 -0500
Revert "sisyphus thinks the test value is simply wrong here."
This reverts commit 5399a05dbb94d8e0c2a0ea2b588ff80ad23f2093.
(sisyphus thinks the bug is actually in ppc64 linux exp())
M t/op/sprintf2.t
commit ff621fd923c3800eec5cdbe64ce9b3b1fa88255b
Author: Jarkko Hietaniemi <[email protected]>
Date: Tue Nov 11 18:14:40 2014 -0500
If long double math functions do not work, abort.
This moves an prototype version of the test from the freebsd hints
to Configure, thus stopping also other platforms that have lacking
longdouble implementations. The earlier test also just disabled
uselongdouble, but now we abort. This is probably more honest,
nd safer.
Configure has already aborted at this spot for certain broken long
double math functions, now we just extend the set of tested functions.
The test is not comprehensive (not all long double interfaces are tested),
but it covers some of the most common ones.
The earlier test was actually wrong, so no FreeBSD could ever pass.
Sorry, FreeBSD.
M Configure
M hints/freebsd.sh
-----------------------------------------------------------------------
--
Perl5 Master Repository
