Bug #46499 for Math-GMPz: v. 0.26 test suite bugs on Mac OS X 10.5

Fri May 29 16:51:18 2009 bobjkuo [...] gmail.com - Ticket created

Subject:	v. 0.26 test suite bugs on Mac OS X 10.5
Date:	Fri, 29 May 2009 15:50:48 -0500
To:	bug-Math-GMPz [...] rt.cpan.org
From:	Bob Kuo <bobjkuo [...] gmail.com>

Short version: The test suite for Math::GMPz 0.26 fails on new.t test 3, not_zero.t test 3, and use64bigint.t tests 1 and 3 Long version: I used MacPorts to install gmp and am using gmp @4.2.4_1+universal (active) To build, I had to point the configure script to a different location because they are installed in a weird location on Mac. Here is the relevant system info `uname -a` Darwin bob.chapter 9.7.0 Darwin Kernel Version 9.7.0: Tue Mar 31 22:52:17 PDT 2009; root:xnu-1228.12.14~1/RELEASE_I386 i386 `perl -V` Summary of my perl5 (revision 5 version 8 subversion 8) configuration: Platform: osname=darwin, osvers=9.0, archname=darwin-thread-multi-2level uname='darwin b02.apple.com 9.0 darwin kernel version 9.3.0: tue aug 12 17:18:07 pdt 2008; root:xnu-1228.5.90~13release_i386 i386 ' config_args='-ds -e -Dprefix=/usr -Dccflags=-g -pipe -Dldflags= -Dman3ext=3pm -Duseithreads -Duseshrplib' hint=recommended, useposix=true, d_sigaction=define usethreads=define use5005threads=undef useithreads=define usemultiplicity=define useperlio=define d_sfio=undef uselargefiles=define usesocks=undef use64bitint=define use64bitall=define uselongdouble=undef usemymalloc=n, bincompat5005=undef Compiler: cc='cc', ccflags ='-arch ppc -arch i386 -g -pipe -fno-common -DPERL_DARWIN -no-cpp-precomp -fno-strict-aliasing -Wdeclaration-after-statement -I/usr/local/include', optimize='-O3', cppflags='-no-cpp-precomp -arch ppc -g -pipe -fno-common -DPERL_DARWIN -no-cpp-precomp -fno-strict-aliasing -Wdeclaration-after-statement -I/usr/local/include' ccversion='', gccversion='4.0.1 (Apple Inc. build 5465)', gccosandvers='' intsize=4, longsize=4, ptrsize=4, doublesize=8, byteorder=1234 d_longlong=define, longlongsize=8, d_longdbl=define, longdblsize=16 ivtype='long', ivsize=4, nvtype='double', nvsize=8, Off_t='off_t', lseeksize=8 alignbytes=8, prototype=define Linker and Libraries: ld='cc -mmacosx-version-min=10.5.7', ldflags ='-arch ppc -arch i386 -L/usr/local/lib' libpth=/usr/local/lib /usr/lib libs=-ldbm -ldl -lm -lutil -lc perllibs=-ldl -lm -lutil -lc libc=/usr/lib/libc.dylib, so=dylib, useshrplib=true, libperl=libperl.dylib gnulibc_version='' Dynamic Linking: dlsrc=dl_dlopen.xs, dlext=bundle, d_dlsymun=undef, ccdlflags=' ' cccdlflags=' ', lddlflags='-arch ppc -arch i386 -bundle -undefined dynamic_lookup -L/usr/local/lib' Characteristics of this binary (from libperl): Compile-time options: MULTIPLICITY PERL_IMPLICIT_CONTEXT PERL_MALLOC_WRAP USE_ITHREADS USE_LARGE_FILES USE_PERLIO USE_REENTRANT_API Locally applied patches: fix for regcomp CVE-2007-5116 security vulnerability fix for double-free CVE-2008-1927 security vulnerability /Library/Perl/Updates/<version> comes before system perl directories installprivlib and installarchlib points to the Updates directory Built under darwin Compiled at Mar 11 2009 16:09:08 @INC: /Library/Perl/Updates/5.8.8/darwin-thread-multi-2level /Library/Perl/Updates/5.8.8 /System/Library/Perl/5.8.8/darwin-thread-multi-2level /System/Library/Perl/5.8.8 /Library/Perl/5.8.8/darwin-thread-multi-2level /Library/Perl/5.8.8 /Library/Perl /Network/Library/Perl/5.8.8/darwin-thread-multi-2level /Network/Library/Perl/5.8.8 /Network/Library/Perl /System/Library/Perl/Extras/5.8.8/darwin-thread-multi-2level /System/Library/Perl/Extras/5.8.8 /Library/Perl/5.8.6 /Library/Perl/5.8.1 . `gcc --version` i686-apple-darwin9-gcc-4.0.1 (GCC) 4.0.1 (Apple Inc. build 5465) Copyright (C) 2005 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. and attached is the output from running `prove` on the test suite. Originally the TRmpz_out_str.t failed all tests but that was just because the permissions of the out* files weren't set correctly. I fixed this with: `sudo chmod g+w out*` and they all passed. I have a hunch that the tests are failing because of the 64int support being set incorrectly. Bob

Message body is not shown because sender requested not to inline it.

Fri May 29 20:33:44 2009 sisyphus [...] cpan.org - Correspondence added

On Fri May 29 16:51:18 2009, bobjkuo@gmail.com wrote: Show quoted text

> `perl -V`

[snip] Show quoted text

> config_args='-ds -e -Dprefix=/usr -Dccflags=-g -pipe -Dldflags= > -Dman3ext=3pm -Duseithreads -Duseshrplib'

[snip] Show quoted text

> use64bitint=define use64bitall=define uselongdouble=undef

That's odd .... how do those configure args result in 'use64bitint' and 'use64bitall' being defined ? Show quoted text

> ivtype='long', ivsize=4, nvtype='double', nvsize=8,

And to top things off, despite 'use64bitint' and 'use64bitall' having been defined, it's just a regular 32-bit build of perl. I doubt that we'll ever want to build a 64-bit int build of Math-GMPz when ivsize < 8, so this patch to the Makefilel.PL should fix things portably: ############################## --- Makefile.PL Fri May 22 18:18:19 2009 +++ Makefile.PL_new Sat May 30 10:31:48 2009 @@ -20,10 +20,12 @@ if($use_64_bit_int == -1) {} elsif($use_64_bit_int == 1) {$defines .= " -DUSE_64_BIT_INT"} else { - if($Config::Config{use64bitint} eq 'define' || - $Config::Config{use64bitall} eq 'define' || - $Config::Config{usemorebits} eq 'define' || - $Config::Config{ivsize} == 8 ) {$defines .= " -DUSE_64_BIT_INT"} + unless($Config::Config{ivsize} < 8) { + if($Config::Config{use64bitint} eq 'define' || + $Config::Config{use64bitall} eq 'define' || + $Config::Config{usemorebits} eq 'define') + {$defines .= " -DUSE_64_BIT_INT"} + } } if($use_long_double == -1) {} ############################## (That patch is for the 0.27 Makefile.PL - but it should apply to the 0.26 Makefile.PL as well.) Show quoted text

> Originally the TRmpz_out_str.t failed all tests but that was just > because the permissions of the out* files weren't set correctly. I > fixed this with: > `sudo chmod g+w out*` > and they all passed.

Is there something I can do to fix that ? Would creating the out* files in the 't' directory rather than the top-level directory fix the problem ? Show quoted text

> I have a hunch that the tests are failing because of the 64int support > being set incorrectly.

Yes - I think that's mainly it. In new.t and not_zero.t, we'll also need to replace occurrences of: if(defined($Config::Config{use64bitint})) { with if(Math::GMPz::_has_longlong()) { Do those changes (to Makefile.PL, new.t and not_zero.t) fix the failing tests with Math-GMPz-0.26 ? How does Math::Primality then fare ? If those changes fix the failing Math-GMPz tests, I'll apply them to 0.28, which will be released once we've got *all* issues sorted out. You might also give Math-GMPz-0.27 a run - though you'll need to make the same changes to it, too. (Probably would be good to check that version 0.27 hasn't introduced any *additional* problems.) Cheers, Rob

Fri May 29 20:33:44 2009 The RT System itself - Status changed from 'new' to 'open'

Sat May 30 20:19:19 2009 sisyphus [...] cpan.org - Correspondence added

RT-Send-CC:

sisyphus [...] cpan.org

On Fri May 29 20:33:44 2009, SISYPHUS wrote: Show quoted text

> I doubt that we'll ever want to build a 64-bit int build of Math-GMPz > when ivsize < 8, so this patch to the Makefilel.PL should fix things > portably:

Actually, instead of using that original patch I proposed, I'll apply the following patch and simplify things in the process: ############################## --- Makefile.PL Fri May 22 18:18:19 2009 +++ Makefile.PL_new Sun May 31 10:12:58 2009 @@ -20,18 +20,17 @@ if($use_64_bit_int == -1) {} elsif($use_64_bit_int == 1) {$defines .= " -DUSE_64_BIT_INT"} else { - if($Config::Config{use64bitint} eq 'define' || - $Config::Config{use64bitall} eq 'define' || - $Config::Config{usemorebits} eq 'define' || - $Config::Config{ivsize} == 8 ) {$defines .= " -DUSE_64_BIT_INT"} + unless($Config::Config{ivsize} < 8) { + $defines .= " -DUSE_64_BIT_INT"; + } } if($use_long_double == -1) {} elsif($use_long_double == 1) {$defines .= " -DUSE_LONG_DOUBLE"} else { - if($Config::Config{uselongdouble} eq 'define' || - $Config::Config{usemorebits} eq 'define' || - $Config::Config{nvsize} > 8 ) {$defines .= " -DUSE_LONG_DOUBLE"} + if($Config::Config{nvsize} > 8 ) { + $defines .= " -DUSE_LONG_DOUBLE"; + } } $defines =~ /-DUSE_64_BIT_INT/ ? print "Building with 64-bit'long long' support\n" : ############################## Show quoted text

> In new.t and not_zero.t, we'll also need to replace occurrences of: > > if(defined($Config::Config{use64bitint})) { > with > if(Math::GMPz::_has_longlong()) {

Duh ... not_zero.t doesn't have any occurrences of "if(defined ($Config::Config{use64bitint})) {", but it does have one occurrence of "defined($Config::Config{use64bitint})" which needs to be replaced with "Math::GMPz::_has_longlong()". Cheers, Rob

Tue Jun 02 03:38:47 2009 bobjkuo [...] gmail.com - Correspondence added

Subject:	Re: [rt.cpan.org #46499] v. 0.26 test suite bugs on Mac OS X 10.5
Date:	Tue, 2 Jun 2009 02:32:08 -0500
To:	bug-Math-GMPz [...] rt.cpan.org
From:	Bob Kuo <bobjkuo [...] gmail.com>

That did the trick for both v0.26 and v0.27. I've modified the code and both Jonathan Leto and I are hosting the changes on Github. You can see the commit here: http://github.com/bubaflub/Math--GMPz/commit/ee258b6326df6b0277031655bec2862528cdee97. I've updated the README with a brief blurb about Mac OS X, fixed some minor typos there, updated the two tests and the Makefile.PL. So I think this officially closes the bug. b

Tue Jun 09 07:26:24 2009 sisyphus [...] cpan.org - Correspondence added

Fixed with release of Math-GMPz-0.28.

Tue Jun 09 07:26:26 2009 sisyphus [...] cpan.org - Status changed from 'open' to 'resolved'