develooper Front page | perl.perl5.porters | Postings from September 2018

[perl #131388] g++ vs subnormals in 5.26.0

Thread Previous | Thread Next
From:
James E Keenan via RT
Date:
September 23, 2018 12:37
Subject:
[perl #131388] g++ vs subnormals in 5.26.0
Message ID:
rt-4.0.24-20820-1537706272-591.131388-15-0@perl.org
On Sat, 22 Sep 2018 12:14:40 GMT, sisyphus@cpan.org wrote:
> On Tue, 11 Sep 2018 19:11:02 -0700, sisyphus@cpan.org wrote:
> 
> > I suppose, to be thorough, we should also be performing similar tests
> > for subnormal NV values on -Duselongdouble and -Dusequadmath builds.
> 
> On closer inspection I can see that's already happening.
> In fact, I don't see any need to make any alteration to the version of
> t/op/sprintf2.t that is currently in the smoke-
> me/jkeenan/sisyphus/131388-subnormals branch ... except, perhaps, for
> one unrelated observation:
> 
> At line 897 of smoke-me/jkeenan/sisyphus/131388-
> subnormals/t/op/sprintf2.t, we begin 7 tests on usequadmath builds in
> response to ticket #128843.
> I believe those tests could also be applied to uselongdouble builds
> where the long double conforms to the IEEE 754 standard.
> I'm not bothered one way or the other about this, and I don't have
> access to a system that provides such long doubles so I can't test it.
> But I'm attaching the patch (against smoke-me/jkeenan/sisyphus/131388-
> subnormals/t/op/sprintf2.t) for review, anyway.
> 
> Cheers,
> Rob

Last night I applied your patch to the smoke-me/jkeenan/sisyphus/131388-subnormals branch and pushed it to the origin for smoke-testing.

Most smoke-test results are good, but I got one failure on my my own FreeBSD-11 smoke-testing rig.  See: http://perl5.test-smoke.org/report/70990.  Where the configuration was:

#####
-des -Dusedevel -Duseithreads -Doptimize=-O2 -pipe -fstack-protector -fno-strict-aliasing -Dcc=g++ -Dusemorebits -DDEBUGGING
#####

... I got this error:

#####
Test failures:
~~ ../dist/threads/t/exit.t .................................... FAILED 14
  [perlio] -Duseithreads -Doptimize="-O2 -pipe -fstack-protector -fno-strict-aliasing" -Dcc="g++" -Dusemorebits DEBUGGING
#####

Test 14 in dist/threads/t/exit.t is coded like this:

#####
$out = run_perl(prog => 'use threads 2.21 qw(exit thread_only);' .
            'threads->create(sub {' .
            '   threads->set_thread_exit_only(0);' .
            '   exit(99);' .
            '});' .
            'sleep(1);' .
            'exit(86);',
    nolib => ($ENV{PERL_CORE}) ? 0 : 1,
    switches => ($ENV{PERL_CORE}) ? [] : [ '-Mblib' ], 
    stderr => 1);
{
    local $TODO = 'VMS exit semantics not like POSIX exit semantics' if $^O eq 'VMS';
    is($?>>8, 99, "set_thread_exit_only(0)");
}
#####

I then went to my git checkout on the same machine, checked out this branch, built perl with the same configuration and ran the test file through the harness.  It PASSed.

So the test failure in the smoke-test may be a transient failure due to resource limitations.  ISTR seeing such failures in that file previously.

Thank you very much.

-- 
James E Keenan (jkeenan@cpan.org)

---
via perlbug:  queue: perl5 status: open
https://rt.perl.org/Ticket/Display.html?id=131388

Thread Previous | Thread Next


nntp.perl.org: Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at ask@perl.org | Group listing | About