Re: Looking for David Burdick

[Khen1950fx]

You probably do not need the author's help. It installed for me. Try this:

#!/usr/bin/perl

use strict;
use warnings;
use CPAN;

my (@mods) = qw(
    Exporter
    Carp
    XSLoader
    AutoLoader
    Text::Template
    Module::Load
    MRO::Compat
    File::Temp
    Sub::Install
    Data::OptList
    lib
    ExtUtils::CBuilder
    Scalar::Util
    Test::More
    Params::Util
    IO::Handle
    Sub::Exporter
    Data::Section
    File::Spec
    ExtUtils::MakeMaker
    base
    Software::License
    Software::License::Boost_1_0
    Boost::Graph
);

foreach my $mod(@mods) {
    CPAN::Shell->install($mod);
}

exit 0;

[davido]

Not everyone is so lucky:

~/repos/Boost-Graph$ prove -b
t/Graph.t .. 1/? 
#   Failed test 'Depth First Search Levels (0 root): 4'
#   at t/Graph.t line 334.
#          got: '1'
#     expected: '4'

#   Failed test 'Depth First Search Levels (0 root): 5'
#   at t/Graph.t line 336.
#          got: '2'
#     expected: '5'

#   Failed test 'Depth First Search Levels (0 root): 7'
#   at t/Graph.t line 338.
#          got: '3'
#     expected: '7'

#   Failed test 'Depth First Search Levels (0 root) depth(7): 3'
#   at t/Graph.t line 339.
#          got: '2'
#     expected: '3'

#   Failed test 'Depth First Search Levels (0 root): 6'
#   at t/Graph.t line 340.
#          got: '4'
#     expected: '6'

#   Failed test 'Depth First Search Levels (0 root) depth(6): 2'
#   at t/Graph.t line 341.
#          got: '1'
#     expected: '2'

#   Failed test 'Depth First Search Levels (0 root): 1'
#   at t/Graph.t line 342.
#          got: '5'
#     expected: '1'

#   Failed test 'Depth First Search Levels (0 root) depth(1): 1'
#   at t/Graph.t line 343.
#          got: '2'
#     expected: '1'

#   Failed test 'Depth First Search Levels (0 root): 2'
#   at t/Graph.t line 344.
#          got: '7'
#     expected: '2'

#   Failed test 'Depth First Search Levels (0 root) depth(2): 2'
#   at t/Graph.t line 345.
#          got: '3'
#     expected: '2'

#   Failed test 'Depth First Search Levels (0 root): 3'
#   at t/Graph.t line 346.
#          got: '6'
#     expected: '3'
# Looks like you failed 11 tests of 95.
t/Graph.t .. Dubious, test returned 11 (wstat 2816, 0xb00)
Failed 11/95 subtests 

Test Summary Report
-------------------
t/Graph.t (Wstat: 2816 Tests: 95 Failed: 11)
  Failed tests:  75, 77, 79-87
  Non-zero exit status: 11
Files=1, Tests=95,  0 wallclock secs ( 0.03 usr  0.00 sys +  0.04 cusr
+  0.00 csys =  0.07 CPU)
Result: FAIL

...and...

~/repos/Boost-Graph$ perl -v

This is perl 5, version 18, subversion 1 (v5.18.1) built for x86_64-li
+nux
...

Update: The work is underway at https://github.com/duffee/Boost-Graph. I'm sure he would appreciate any fork/pull-requests that advance the cause.

[Ea]

ahAA! We've got a Heisenbug! I was thinking about this discussion after you mentioned you were running v5.18.1 and I was running v5.16.3. The failing tests depend on the order of elements in an array filled by keys. I installed perlbrew and perl-5.18.1 and failed a different set of tests.

The interesting thing is that with a perlbrew installed perl-5.16.3 (to prove that the failed tests were truly caused by the hash randomization feature introduced in v5.18), it still fails 2 tests and yet passes all tests on my fedora 19 packaged Perl 5.16.3 - wonders never cease.

In spite of it waking me up in the middle of the night, I feel quite pleased tracking that down. Think I'll treat myself to a lemon tart in the coffee shop while I re-think the test :)

[Ea]

Not if you're compiling with gcc 4.4 and above. We're trying to quash this bug on RT. It means that make doesn't succeed. The fact that we've got tests failing is a big step forward, mainly due to the work done by cjfields, dcmertens and davido.

Out of interest, what C compiler did you use to install Boost::Graph?

[marto]

Why are you including lots of core modules in this installation list? Did you actually install Boost::Graph without error using the method you suggest?

[DrHyde]

You should always declare in your Makefile.PL (or Build.PL, or other build instructions) *all* the modules your code depends on. That's because older versions of perl have fewer modules in core, and new versions of perl occasionally drop modules from core.

I presume that that is a list of all the modules used by the code. Including a few core modules can't do any harm.

[marto]

"You should always declare in your Makefile.PL (or Build.PL, or other build instructions) *all* the modules your code depends on. That's because older versions of perl have fewer modules in core"

I agree, but the modules in question have been core since 5/5.6.

"I presume that that is a list of all the modules used by the code"

Is it? Perhaps. Khen1950fx has a history of posts like this where there's no difference from doing cpan Module::Name, also forcing installs so potential problems are missed and even going so far as to create a module on cpan (infact I just found another, Task::Mechanistic, again pointless) to install another module. None of this makes any sense from a module installation perspecitve.

Update: s/installing/install/

[Khen1950fx]

Yes---It installed without error, which seems to be something that eludes you:).

[marto]

Eludes me? I specifically asked you this question. You answered. The other inditect question (also asked previously), of why you suggest people install modules in such a ridiculous fashion and why you pollute CPAN with modules which simply install other cpan modules remains unknown.