develooper Front page | perl.perl5.porters | Postings from April 2010

Re: Perl-5-12-0-RC4 test failures - CPANPLUS using system perl?

Thread Previous | Thread Next
From:
Darren Duncan
Date:
April 8, 2010 02:06
Subject:
Re: Perl-5-12-0-RC4 test failures - CPANPLUS using system perl?
Message ID:
4BBD1221.5030301@darrenduncan.net
P.S.  Again, CC replies directly to me if I'm to see/respond in a timely manner.

As a follow-up, I did 2 additional tests so far, just on the Intel machine:

1.  With the RC4, I ran "make test" rather than "make test_harness", in case the 
test parallelism had anything to do with the problems; the problems still occurred.

2.  I rebuilt RC2 again using the exact same configure settings as for the 
failing RC4, and RC2 passed all tests:

#####

All tests successful.

Test Summary Report
-------------------
../ext/POSIX/t/posix.t                                          (Wstat: 0 Tests: 
66 Failed: 0)
   TODO passed:   11
Files=1801, Tests=357458, 446 wallclock secs (54.72 usr 20.45 sys + 435.66 cusr 
68.46 csys = 579.29 CPU)
Result: PASS

#####

Although that POSIX line is still there, which looks odd.

So it seems fairly clear that something changed between RC2 and RC4 that broke. 
  Or alternately, that RC4 exposed an underlying problem (using System Perl) 
that wasn't exposed before, perhaps?

It would seem that I eliminated any other variables, because both of these, 
within an hour of each other, involved all of the same OS patches that might 
have changed since I tested RC2 the other day.

Thank you.

-- Darren Duncan

Darren Duncan wrote:
> So, as with the Perl 5.12.0 candidates #0 and #2, I have tested #4 on 
> two Mac OS X machines, one a 32-bit PPC dual-G4 running 10.5.8, the 
> other a 64-bit Intel Core2Duo running 10.6.3.  In the past, all of my 
> tests were successful, but today there were failures on just the Intel 
> machine.
> 
> P.S.  Any responses meant to be read by me in a timely manner should be 
> CC'd to my sender address (darren@darrenduncan.net) as I'm not 
> subscribed to perl5-porters and only read it from a delayed web archive 
> like nntp.perl.org.
> 
> While testing both candidate #2 and #4, on both machines, I used this 
> arrangement:
> 
>     sh Configure -des -Duserelocatableinc -Dusethreads 
> -Dprefix=/Volumes/<volname>/Perl/perl_5_12_0_rc<N>/
>     make
>     TEST_JOBS=3 make test_harness
>     make install
> 
> All tests passed for #2, as I recall, with this setup on both machines.
> 
> For #4, all tests passed on the PPC machine only:
> 
> #####
> 
> All tests successful.
> Files=1801, Tests=357262, 1423 wallclock secs (205.15 usr 72.00 sys + 
> 1476.28 cusr 274.90 csys = 2028.33 CPU)
> Result: PASS
> 
> #####
> 
> ... while the Intel machine passed almost all of the tests, with these 
> exceptions:
> 
> #####
> 
> Test Summary Report
> -------------------
> ../cpan/CPANPLUS-Dist-Build/t/02_CPANPLUS-Dist-Build.t          (Wstat: 
> 1280 Tests: 34 Failed: 5)
>   Failed tests:  5, 8, 18, 21, 33
>   Non-zero exit status: 5
> ../cpan/CPANPLUS/t/04_CPANPLUS-Module.t                         (Wstat: 
> 256 Tests: 124 Failed: 1)
>   Failed test:  109
>   Non-zero exit status: 1
> ../cpan/CPANPLUS/t/20_CPANPLUS-Dist-MM.t                        (Wstat: 
> 512 Tests: 43 Failed: 8)
>   Failed tests:  11-14, 16-18, 42
>   Non-zero exit status: 2
> ../ext/POSIX/t/posix.t                                          (Wstat: 
> 0 Tests: 66 Failed: 0)
>   TODO passed:   11
> Files=1801, Tests=357419, 453 wallclock secs (54.74 usr 20.25 sys + 
> 430.68 cusr 66.64 csys = 572.31 CPU)
> Result: FAIL
> make: *** [test_harness] Error 14
> 
> #####
> 
> For more details, below this email is the complete output of "make 
> test_harness" but with all the "... ok" lines stripped out.
> 
> The problem there seems to be that the test suite is trying to use the 
> System Perl 5.10.0 (which isn't on the PPC machine, which has 5.8.8 
> instead) rather than the Perl that I am compiling.
> 
> This seems to be a significant problem in general; any other 
> installations of Perl that might be on the system should be ignored, and 
> only the one we're building now should be used.
> 
> -- Darren Duncan
<snip>

Thread Previous | Thread Next


nntp.perl.org: Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at ask@perl.org | Group listing | About