Bug #111149 for Math-BigInt-GMP: Devel::CheckLib breaks build of Math-BigInt-GMP-1.49

Wed Jan 13 08:46:43 2016 sisyphus [...] cpan.org - Ticket created

Subject:

Devel::CheckLib breaks build of Math-BigInt-GMP-1.49

Hi, The Math-BigInt-GMP-1.49 Makefile.PL contains: check_lib_or_exit(lib => 'gmp', header => 'gmp.h', @ARGV); On MS Windows, if I want to install Math-BigInt-GMP-1.49, I first need to comment out that line - after which the module builds, tests and installs flawlessly. Otherwise I get an error about the header and/or library not being found and the Makefile.PL aborts. (I forget the exact error message.) On *my* system gmp.h is automatically locatable because $ENV{CPATH} specifies its location, and -lgmp is automatically resolved because $Config{libpth} specifies the location of libgmp.a. I would have expected that Devel::CheckLib would be alert to those options, but apparently that's not so. I don't like Devel::CheckLib, I don't use it, and I don't see why authors want to make use of it - so, understandably, I haven't done any digging :-) However, I *am* prepared to do some investigating if you have trouble working out why/how Devel::CheckLib is blocking my attempts to build Math-BigInt-GMP-1.49. Cheers, Rob

Thu Feb 04 04:36:08 2016 peter.john.acklam [...] gmail.com - Correspondence added

Hi Rob, I view this a two separate issues. One is that if Devel::CheckLib can't find your gmp.h, then there is a bug in Devel::CheckLib. The other is whether Math-BigInt-GMP should be using Devel::CheckLib. I can't do much about the first issue, but you might want to create a ticket here: https://rt.cpan.org/Public/Dist/Display.html?Name=Devel-CheckLib As for using Devel::CheckLib with Math-BigInt-GMP, this was added before I started maintaining the Math::Big(Int|Rat|Float) mess. I have no strong preferences on whether Devel::CheckLib should be used or not. One advantage with Devel::CheckLib is that - if I remember correctly - CPAN::Reporter produces a report with status NA rather than UNKNOWN if "gmp.h" can't be found.

Thu Feb 04 04:36:09 2016 The RT System itself - Status changed from 'new' to 'open'

Thu Feb 04 07:58:53 2016 sisyphus1 [...] optusnet.com.au - Correspondence added

Subject:	Re: [rt.cpan.org #111149] Devel::CheckLib breaks build of Math-BigInt-GMP-1.49
Date:	Thu, 4 Feb 2016 23:57:26 +1100
To:	<bug-Math-BigInt-GMP [...] rt.cpan.org>
From:	<sisyphus1 [...] optusnet.com.au>

Show quoted text

-----Original Message----- From: Peter John Acklam via RT Sent: Thursday, February 04, 2016 8:36 PM To: sisyphus@cpan.org Subject: [rt.cpan.org #111149] Devel::CheckLib breaks build of Math-BigInt-GMP-1.49

> One advantage with Devel::CheckLib is that - if I remember correctly - > CPAN::Reporter produces a report with status NA rather than UNKNOWN if > "gmp.h" can't be found.

Yes, I think that's the way it's supposed to work, though I don't personally care about the distinction and therefore don't see any "advantage" in this. Mind you, there are some cpan-testers (using App::cpanminus::reporter instead of CPAN::Reporter) that will incorrectly report a FAIL if "gmp.h" is not found. The good news is that gmp.h is always (AFAIK) found on these machines - at least, with my modules that access the gmp library, I've never had a FAIL because "gmp.h" was not found. (And I don't do any pre-checking re the availability of the gmp header and library.) But I did recently get an incorrect FAIL with Math::Float128 because "quadmath.h" could not be found: http://www.cpantesters.org/cpan/report/7a4a4ad6-c608-11e5-9eb0-875dfcf88752 It would be nice if Math::BigInt::GMP were to build straight out of the box, but I know how to work around the problem so it's not a big issue for me. This bug report was filed primarily as an FYI and I don't intend filing a bug report against Devel::CheckLib. Cheers, Rob

Mon Dec 05 08:16:40 2016 peter.john.acklam [...] gmail.com - Correspondence added

Removed the use of Devel::CheckLib. Failing when everything is OK is just too silly.

Mon Dec 05 08:16:41 2016 peter.john.acklam [...] gmail.com - Status changed from 'open' to 'resolved'

Mon Dec 05 08:16:41 2016 peter.john.acklam [...] gmail.com - Fixed in 1.6002 added