sisyphus / math-mpfr Goto Github PK

I am trying to configure GitHub Actions to run the test of this module.
It would be nice if you could include more exact instruction on how to install the external dependencies. (e.g apt-get install WHAT on debian/ubuntu)

The version of the pre-compiled MPFR library provided by Arch Linux (https://www.archlinux.org/packages/core/i686/mpfr/) has a suffix version, which makes the third test (from t/_1basic.t) to fail with the following message:

not ok 3 - Header and Library do not match

and the fifth test from t/constants.t to fail with:

not ok 5 3.1.3-p5 is not the same as 3.1.3

The value of Rmpfr_get_version is 3.1.3-p5, while the value of MPFR_VERSION_STRING is 3.1.3. I think this also affects other Linux distributions.

A simple fix would be removing the suffix from Rmpfr_get_version to match MPFR_VERSION_STRING, or maybe changing the test like this:

{
    my $v1 = Rmpfr_get_version();
    my $v2 = MPFR_VERSION_STRING;
    $v1 =~ s/-.*//;                 # remove any version suffix
    $v2 =~ s/-.*//;                 # =//=
    if($v1 eq $v2) {print "ok 3\n"}
    else {
        print "not ok 3 - Header and Library do not match ('$v1' ne '$v2')\n"
    }
}

As a result of Perl/perl5@b1e3ba27cd, the tests of Inf and NaN in this test file are now failing. Because of the accumulating nature of $check, that then causes a cascade of failures through the rest of the test file.

Please see the commit message for full details of what changed, and Perl/perl5#19550 for the original report. I suspect Math::MPFR just needs to tweak a couple of tests here.

We have the following bug reported to the Debian package of Math-MPFR
(https://bugs.debian.org/918673):

It doesn't seem to be a bug in the packaging, so you may want to take
a look. Thanks!

------8<-----------8<-----------8<-----------8<-----------8<-----

Source: libmath-mpfr-perl
Version: 4.05-1
Severity: serious
Tags: ftbfs

https://buildd.debian.org/status/package.php?p=libmath-mpfr-perl

...
Invalid type 'D' in pack at t/subnormal_doubles.t line 63.
t/subnormal_doubles.t ............ 
1..1
Dubious, test returned 255 (wstat 65280, 0xff00)
Failed 1/1 subtests 

Calling all cache freeing functions
Skipping tests 4, 5 and 6 - not relevant to this build
...
Test Summary Report
-------------------
t/subnormal_doubles.t          (Wstat: 65280 Tests: 0 Failed: 0)
  Non-zero exit status: 255
  Parse errors: Bad plan.  You planned 1 tests but ran 0.
Files=69, Tests=1137, 12 wallclock secs ( 0.29 usr  0.08 sys +  2.49 cusr  0.28 csys =  3.14 CPU)
Result: FAIL
Failed 1/69 test programs. 0/1137 subtests failed.
make[1]: *** [Makefile:1021: test_dynamic] Error 255

------8<-----------8<-----------8<-----------8<-----------8<-----

Thanks for considering,
gregor herrmann,
Debian Perl Group