Bug #55997 for Math-BigInt-GMP: Using Math::BigInt::GMP in DateTime::Format::Epoch on Windows causes install failure

Sat Mar 27 09:08:14 2010 RPKELLY [...] cpan.org - Ticket created

Subject:

Using Math::BigInt::GMP in DateTime::Format::Epoch on Windows causes install failure

DateTime::Format::Epoch 0.11 fails to install when using Math::BigInt::GMP as it's precision math library, and this appears to be due to a bug in Math::BigInt::GMP. Math::BigInt::GMP appears to return an incorrect value for the last two tests in t/402_format.t. Using other precision math libraries causes the test suite to pass. This is on Strawberry Perl v5.10.1 (*) built for MWin32-x86-multi-thread. See also https://rt.cpan.org/Ticket/Display.html?id=55996 .

Mon Feb 07 10:01:26 2011 peter.john.acklam [...] gmail.com - Correspondence added

You're not showing me the output from the test script, so I am not sure what output you get vs. the output you should have got. I have looked at the two last tests in t/402_format.t. In those two tests, pure Perl is used to subtract 31241376000000000 from another number. That is risky, since Perl can only handle such large integers if Perl is built with support for 64 bit integers or long doubles. For instance, on Perl without 64 bit support you get 0, although the two numbers differ by 1: $ perl-5.13.9-cygwin -wle 'print 31241376000000001 - 31241376000000000' 0 I consider this to be a poorly written test. It should use smaller numbers or only execute these tests if Perl can support such large integers. If you can trace this issue to a bug in Math::BigInt::GMP, I'd be happy to re-open this ticket.

Mon Feb 07 10:01:27 2011 The RT System itself - Status changed from 'new' to 'open'

Mon Feb 07 10:01:27 2011 peter.john.acklam [...] gmail.com - Status changed from 'open' to 'rejected'