Bug #63379 for Math-BigInt: Failing tests comparing inf to Inf

Sat Nov 27 07:16:16 2010 james2vegas [...] aim.com - Ticket created

Subject:

Failing tests comparing inf to Inf

t/bare_mbi.t ........ 608/3615
Show quoted text

#   Failed test at t/bigintpm.inc line 206.
#          got: 'Inf'
#     expected: 'inf'

Show quoted text

#   Failed test at t/bigintpm.inc line 206.
#          got: '-Inf'
#     expected: '-inf'

t/bare_mbi.t ........ 3257/3615 # Looks like you failed 2 tests of 3615.
t/bare_mbi.t ........ Dubious, test returned 2 (wstat 512, 0x200)

t/bigintpm.t ........ 1/3621
Show quoted text

#   Failed test at t/bigintpm.inc line 206.
#          got: 'Inf'
#     expected: 'inf'

t/bigintpm.t ........ 763/3621
Show quoted text

#   Failed test at t/bigintpm.inc line 206.
#          got: '-Inf'
#     expected: '-inf'

t/bigintpm.t ........ 2396/3621 # Looks like you failed 2 tests of 3621.
t/bigintpm.t ........ Dubious, test returned 2 (wstat 512, 0x200)
Failed 2/3621 subtests

t/sub_mbi.t ......... 520/3620
Show quoted text

#   Failed test at t/bigintpm.inc line 206.
#          got: 'Inf'
#     expected: 'inf'

Show quoted text

#   Failed test at t/bigintpm.inc line 206.
#          got: '-Inf'
#     expected: '-inf'

t/sub_mbi.t ......... 3257/3620 # Looks like you failed 2 tests of 3620.
t/sub_mbi.t ......... Dubious, test returned 2 (wstat 512, 0x200)
Failed 2/3620 subtests

Sat Nov 27 07:17:03 2010 james2vegas [...] aim.com - Correspondence added

bash-2.05a# perl -v

This is perl, v5.8.9 built for i386-freebsd

Sat Nov 27 07:17:03 2010 james2vegas [...] aim.com - Status changed from 'new' to 'open'

Sun Nov 28 02:52:34 2010 peter.john.acklam [...] gmail.com - Correspondence added

On Sat Nov 27 07:17:03 2010, JWRIGHT wrote: Show quoted text

> This is perl, v5.8.9 built for i386-freebsd

Could you show me how your Perl prints infinity? Please try with something like perl -wle 'print 1e99 ** 1e99' Older versions of Perl have several different ways of representing a numerical scalar infinity and Not-a-number. The test cases expect "inf" and "NaN" and nothing else. The test cases should be modified so it allows the cases when Perl represents these values with something other than "inf" or "NaN".

Sun Nov 28 03:07:26 2010 rafl [...] debian.org - Correspondence added

Subject:	Re: [rt.cpan.org #63379] Failing tests comparing inf to Inf
Date:	Sun, 28 Nov 2010 09:06:20 +0100
To:	bug-Math-BigInt [...] rt.cpan.org
From:	Florian Ragwitz <rafl [...] debian.org>

"Peter John Acklam via RT" <bug-Math-BigInt@rt.cpan.org> writes: Show quoted text

> Queue: Math-BigInt > Ticket <URL: https://rt.cpan.org/Ticket/Display.html?id=63379 > > > On Sat Nov 27 07:17:03 2010, JWRIGHT wrote: >

>> This is perl, v5.8.9 built for i386-freebsd

> > Could you show me how your Perl prints infinity? Please try with > something like > > perl -wle 'print 1e99 ** 1e99' > > Older versions of Perl have several different ways of representing a > numerical scalar infinity and Not-a-number. The test cases expect > "inf" and "NaN" and nothing else. The test cases should be modified so > it allows the cases when Perl represents these values with something > other than "inf" or "NaN".

Actually, that doesn't necessarily depend on the perl version. In most configurations and on most systems stringifying numbers is done by the C library of the system perl is running on. Unfortunately, there is no standard for how C libraries are supposed to stringify NaN and Inf. Common values are covered by /[+-]?inf/i and /[+-]?nan/i, but there's many more, such as '1.#INF' or '1.#IND' on win32. Perl does make no attempt at normalising this. I do happen to be working on a feature branch for perl to do that, but that is not quite finished yet and won't be available in a released version until 5.14 anyway.

Download (untitled)
application/pgp-signature 197b

Message body not shown because it is not plain text.

Sun Nov 28 03:14:10 2010 peter.john.acklam [...] gmail.com - Correspondence added

I thought this had been standardized in Perl, to take care of the problem with the differing C libraries, so that Perl now returns "inf" and "NaN" regardless of the underlying C library. I could be mistaken, though. Anyway, the test case must be changed to handle this, or temporarily removed.

Sun Nov 28 05:57:19 2010 james2vegas [...] aim.com - Correspondence added

Show quoted text

# perl -wle 'print 1e99 ** 1e99'

Inf

Mon Nov 29 07:21:14 2010 HMBRAND [...] cpan.org - Correspondence added

RT-Send-CC:

rafl [...] debian.org

On HP-UX 11.11 PA-RISC/32 with perl5.12.2: $ perl -wle 'print 1e99 ** 1e99' ++ t/bare_mbi.t ........ 756/3615 # Failed test at t/bigintpm.inc line 206. # got: '++' # expected: 'inf' t/bare_mbi.t ........ 757/3615 # Failed test at t/bigintpm.inc line 206. # got: '---' # expected: '-inf' t/bigintpm.t ........ 762/3621 # Failed test at t/bigintpm.inc line 206. # got: '++' # expected: 'inf' t/bigintpm.t ........ 763/3621 # Failed test at t/bigintpm.inc line 206. # got: '---' # expected: '-inf' t/sub_mbi.t ......... 756/3620 # Failed test at t/bigintpm.inc line 206. # got: '++' # expected: 'inf' t/sub_mbi.t ......... 757/3620 # Failed test at t/bigintpm.inc line 206. # got: '---' # expected: '-inf' HP-UX 11.11 PA-RISC/64 with perl-5.12.2: $ perl -wle 'print 1e99 ** 1e99' ++

Mon Nov 29 08:11:16 2010 peter.john.acklam [...] gmail.com - Correspondence added

On Mon Nov 29 07:21:14 2010, HMBRAND wrote: Show quoted text

> > $ perl -wle 'print 1e99 ** 1e99' > ++

Thank you! I am working on a fix. By the way, is the output numeric? If you do $ perl -wle 'print 0 + 1e99 ** 1e99' do you get the same output, or do you get Argument "++" isn't numeric in addition (+) at -e line 1. 0 It would surprise me if you get the latter, but it would be nice if you could confirm that you don't.

Tue Feb 08 05:29:22 2011 peter.john.acklam [...] gmail.com - Correspondence added

The problem was non-portable tests. They have been removed. Fixed in Math-BigInt-1.991.

Tue Feb 08 05:29:23 2011 peter.john.acklam [...] gmail.com - Status changed from 'open' to 'resolved'