Bug 1459155 - perl-Math-Pari-2.010809-5.fc27: FTBFS with Perl 5.26 on 32-bit platforms
Summary: perl-Math-Pari-2.010809-5.fc27: FTBFS with Perl 5.26 on 32-bit platforms
Keywords:
Status: CLOSED RAWHIDE
Alias: None
Product: Fedora
Classification: Fedora
Component: perl-Math-Pari
Version: rawhide
Hardware: i686
OS: Unspecified
unspecified
unspecified
Target Milestone: ---
Assignee: Paul Howarth
QA Contact: Fedora Extras Quality Assurance
URL: https://koji.fedoraproject.org/koji/t...
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2017-06-06 12:30 UTC by Jitka Plesnikova
Modified: 2017-06-24 21:12 UTC (History)
4 users (show)

Fixed In Version: perl-Math-Pari-2.010809-5.fc27 libpari23-2.3.5-11.fc27
Doc Type: If docs needed, set a value
Doc Text:
Clone Of:
Environment:
Last Closed: 2017-06-24 21:12:12 UTC
Type: Bug
Embargoed:


Attachments (Terms of Use)

Description Jitka Plesnikova 2017-06-06 12:30:45 UTC
Build of perl-Math-Pari-2.010809-5.fc27 failed with Perl 5.26 on architectures i686 and armv7hl.

PERL_DL_NONLAZY=1 "/usr/bin/perl" "-MExtUtils::Command::MM" "-MTest::Harness" "-e" "undef *Test::Harness::Switches; test_harness(0, 'blib/lib', 'blib/arch')" t/*.t
t/000_load-problem.t ...... ok
Binary number > 0b11111111111111111111111111111111 non-portable at (eval 3) line 6.
Binary number > 0b11111111111111111111111111111111 non-portable at (eval 3) line 7.
Binary number > 0b11111111111111111111111111111111 non-portable at (eval 3) line 11.
Hexadecimal number > 0xffffffff non-portable at (eval 3) line 12.
Binary number > 0b11111111111111111111111111111111 non-portable at (eval 3) line 13.
t/00_Pari.t ............... 
Failed 1/559 subtests 
<...>
t/55_program.t ............ 
Failed 1/37 subtests 
t/55_qfbsolve.t ........... skipped: no tests found in `/usr/share/libpari23/src/test/32/qfbsolve'
t/55_rfrac.t .............. skipped: no tests found in `/usr/share/libpari23/src/test/32/rfrac'
t/55_round4.t ............. skipped: no tests found in `/usr/share/libpari23/src/test/32/round4'
t/55_stark.t .............. skipped: no tests found in `/usr/share/libpari23/src/test/32/stark'
t/55_sumiter.t ............ ok
t/55_trans.t .............. ok
t/PlotRect.t .............. skipped: per MP_NOGNUPLOT
t/zz_leak.t ............... ok
Test Summary Report
-------------------
t/00_Pari.t             (Wstat: 0 Tests: 559 Failed: 1)
  Failed test:  12
t/55_program.t          (Wstat: 0 Tests: 37 Failed: 1)
  Failed test:  3

Note: It is probably related to fix BZ#1268828 (32 bit platforms uses 64 bit ints)

Comment 1 Petr Pisar 2017-06-09 09:19:26 UTC
The failing subtest in t/00_Pari.t is subtest number 12:

$ perl -Iblib/{lib,arch} t/00_Pari.t |less
1..559
ok 1
ok 2
ok 3
ok 4
ok 5
ok 6
ok 7
ok 8
ok 9
ok 10
ok 11
not ok 12
ok 13
[...]

The test does:

use Math::Pari;

$add = Math::Pari::loadPari("gadd");

test($add);                     # 8
test("$add" =~ /^CODE/);        # 9

$b= &$add (34,67);

test($b);                       # 10
test(ref $b eq "Math::Pari");   # 11
test(Math::Pari::pari2iv($b)==101); # 12


The failure is the Math::Pari::pari2iv($b) value becomes 33554840 on i686.
The $b object is bless( do{\(my $o = 3067660500)}, 'Math::Pari' );

Comment 2 Petr Pisar 2017-06-09 10:00:37 UTC
This code:

use Math::Pari;
#$b= Math::Pari::gadd (1,1);
$b= Math::Pari::gadd (1,1);
print "$b\n";
my $r = Math::Pari::pari2iv($b);
print "$r\n";

prints:

2
9

If you uncomment the first assignment form gadd() call to $b, it will print correctly:

2
2

Assigning to different variable does not help. Replacing the first assignment with:

$b=PARI 3.14;

also helps.

Comment 3 Paul Howarth 2017-06-24 21:12:12 UTC
I rebuilt libpari23 without GMP (which is the same way upstream Math::Pari would build it when bundling it) and that has resolved the problem.


Note You need to log in before you can comment on or make changes to this bug.