5.9.1 release now integrated into ponie

Arthur's not here this week, so it seems that I'm holding the fort.
(Or at least, he delegated me the messy job - integrating current bleadperl)

I've updated the "perl" in the ponie CVS tree to release 5.9.1.
This involved some messy unpicking of conflicts, particularly in hv.c

Annoyingly at this point some tests now fail, but I figured that it was
better to commit to CVS and then attempt to fix them. On x86 Linux I see:

Failed Test                    Stat Wstat Total Fail  Failed  List of Failed
-------------------------------------------------------------------------------
../ext/POSIX/t/sigaction.t                   25    6  24.00%  10 12-13 20-21 25
../lib/ExtUtils/t/Embed.t                     9    9 100.00%  1-9
../lib/Math/BigInt/t/bigints.t  255 65280    51  102 200.00%  1-51
88 tests and 519 subtests skipped.
Failed 3/873 test scripts, 99.66% okay. 66/75425 subtests failed, 99.91% okay.


On OS X (under harness) I see:

Failed Test                    Stat Wstat Total Fail  Failed  List of Failed
-------------------------------------------------------------------------------
../lib/ExtUtils/t/Embed.t                     9    9 100.00%  1-9
../lib/Math/BigInt/t/bigints.t  255 65280    51  102 200.00%  1-51
../lib/warnings.t                           503    1   0.20%  224
op/exec.t                                    21    1   4.76%  21

however, under TEST, or when running by hand with PERL_DESTRUCT_LEVEL=2
t/op/ref fails test 64 on OS X. This is


runperl(prog => 'sub UNIVERSAL::DESTROY { warn } bless \$a, A', stderr => 1);
if ($? != 0) { print "not " };
print "ok ",++$test," - warn called inside UNIVERSAL::DESTROY\n";


ie it's testing the exit code of
  ./perl -le 'sub UNIVERSAL::DESTROY { warn } bless \$a, A';

(which happens to SEGV with 5.8.0)

Arthur and I met this same problem with test 64 before, and I can't remember
how we solved it, or whether it just "went away". I can't reproduce it on
x86 Linux, and valgrind doesn't show anything up.


For information, Parrot itself isn't passing one of its tests at the moment,
which will cause ponie's make test to fail before it tests perl. That failure
isn't caused by anything ponie.

Nicholas Clark