Do we really need this? (ob IANAL but...) it seems to me that by default everything is copyrighted so there is no need to mention it.

Also, from the docs, the module tests "That the said default copyright statement has at least one final year that matches the current year". The code seems to be supporting this. Why? If I have a module that hasn't been updated since 2006, why would the copyright mention 2011 at all? Does this also means that a test that passes on December 31st 2011 will stop passing on January 1st 2012? That seems a little extreme, not to mention impolite.

If I were to use this, it would probably be in a 'releasetest' target or somesuch -- something only run at release time. I see it as only applying at release time, so why run it at any other time.

--MidLifeXis

Do we really need this? (ob IANAL but...) it seems to me that by default everything is copyrighted so there is no need to mention it.

I have clarified the issues in my original post.

Also, from the docs, the module tests "That the said default copyright statement has at least one final year that matches the current year". The code seems to be supporting this. Why? If I have a module that hasn't been updated since 2006, why would the copyright mention 2011 at all? Does this also means that a test that passes on December 31st 2011 will stop passing on January 1st 2012? That seems a little extreme, not to mention impolite.

My assumption was that this would only be run by module authors just prior to release. So the default copyright should be uptodate. Individual files only have to match a subset of the default range.

It already exists, and I use it before every release. It is part of Module::CPANTS::Analyse as Module::CPANTS::Kwalitee::License.

That only seems to cover the license and that was much the easier part.

The year of a copyright does not have to be the present year.   Perhaps you could test that a copyright notice exists or not, but I find it difficult to consider how “copyright” could be a meaningful Test subject.

Edit:   Intended only as an observation, not “negative feedback!”

To me it looks like Fedora's policy makes sense: In cases where the licensing is unclear.... So if the license is clear, there should be no need to care about the copyright. The author field should be enough.

That said, if this helps the Debian packagers, then why not.

Is there somewhere a list of best practices for modules that would help reduce the packagers work load? Something like the list in your post, but maybe better formalized, ideally as a single test file, to be included, as t/debian.t for example, in distributions?

Fails to build: There is not a lot that can be done there as it is adequately captured by the CPAN testing.

Well, actually you could reduce the number of dependencies. This module doesn't need Perl6::Slurp at all, as it is easy to do without.

--- lib/Test/Copyright.pm.org   2011-06-15 11:36:34.179000011 +0200
+++ lib/Test/Copyright.pm       2011-06-15 11:35:08.949000008 +0200
@@ -10,7 +10,6 @@ use Test::More;
 use CPAN::Meta;
 use Software::LicenseUtils;
 use Readonly;
-use Perl6::Slurp;
 use UNIVERSAL::require;
 use Lingua::EN::NameParse;
 use Email::Address;
@@ -141,6 +140,13 @@ sub _software_licenses_ok {
     return @licenses;
 }

+sub slurp {
+    my $filename = shift;
+    open my $fh, "<", $filename or die "$filename: $!";
+    local $/;
+    return <$fh>;
+}
+
 sub _cpan_meta_ok {
     foreach my $file (@META_FILES) {
         if (-r $file) {
[download]

After that, and installing mod::

All tests successful.
Files=9, Tests=12,  0 wallclock secs ( 0.03 usr  0.02 sys +  0.47 cusr
+  0.04 csys =  0.56 CPU)
Result: PASS
[download]

Now let me use that on a module I consider perfect:

$ perl -MTest::Copyright -MTest::More -wle'copyright_ok (); done_testi
+ng'
not ok 1 - found CPAN::Meta file
#   Failed test 'found CPAN::Meta file'
#   at /pro/lib/perl5/site_perl/5.12.2/Test/Copyright.pm line 91.
ok 2 # skip No CPAN::Meta object
1..2
# Looks like you failed 1 test of 2.
Exit 1
$
[download]

Well, of course, there is no META.yml during testing. That is generated in a make dist.

I'm not here to add only negative critiques. I'm very much into any means to increase the quality (kwalitee) of my modules, and I'm also very curious about how others think that can be achieved. So far I have however seen no additional value in this module in the form it is currently presented.

FWIW, I have created a meta-target in most of my Makefile.PL's that does all quality checks I can imagine. Of course that is only run after I have used Module::Release to check the module against all available and supported perl builds (I have over 70).

$ make tgzdist
pod-spell-check --aspell
    ok 1 - CSV_XS.pm
    ok 2 - sandbox/i-ttt/lib/i/ttt.pm
    1..2
ok 1 - CommonMistakes
    ok 1 - CSV_XS.pm
    ok 2 - sandbox/i-ttt/lib/i/ttt.pm
    1..2
ok 2 - Spell-check with aspell
    1..0 # SKIP Ispell not selected
ok 3 # skip Ispell not selected
1..3
perl sandbox/genPPPort_h.pl
perl sandbox/genMETA.pl -c
Check if ChangeLog and README are still valid UTF8 ...
Check required and recommended module versions ...
Checking generated YAML ...
Checking if 5.006 is still OK as minimal version for examples
1..5
ok 1 - examples/speed.pl
ok 2 - examples/csv-check
ok 3 - examples/parser-xs.pl
ok 4 - examples/csv2xls
ok 5 - examples/csvdiff
rm -rf Text-CSV_XS-0.83
/pro/bin/perl "-MExtUtils::Manifest=manicopy,maniread" \
        -e "manicopy(maniread(),'Text-CSV_XS-0.83', 'best');"
mkdir Text-CSV_XS-0.83
mkdir Text-CSV_XS-0.83/examples
mkdir Text-CSV_XS-0.83/t
mkdir Text-CSV_XS-0.83/files
META.yml not found at -e line 1
Generating META.yml
perl sandbox/genMETA.pl
tar cvf Text-CSV_XS-0.83.tar Text-CSV_XS-0.83
Text-CSV_XS-0.83/
:
Text-CSV_XS-0.83/Makefile.PL
rm -rf Text-CSV_XS-0.83
gzip --best Text-CSV_XS-0.83.tar
/pro/bin/perl "-MExtUtils::Manifest=fullcheck" -e fullcheck
No such file: META.yml
Checked dist            Text-CSV_XS-0.83.tgz
Kwalitee rating         130.43% (30/23)
Ignoring metrics        is_prereq, prereq_matches_use, build_prereq_ma
+tches_use
$
[download]

When I extract the META.yml from the generated archive, just to see what Test::Copyright would do:

$ perl -MTest::Copyright -MTest::More -wle'copyright_ok (); done_testi
+ng'
ok 1 - found CPAN::Meta file isa CPAN::Meta
ok 2 - more than zero licenses
ok 3 - Found a good license object
ok 4 - more than zero recognised licenses
ok 5 - found license file: README
not ok 6 - Found license perl in file README
#   Failed test 'Found license perl in file README'
#   at /pro/lib/perl5/site_perl/5.12.2/Test/Copyright.pm line 208.
Use of uninitialized value $holder in anonymous hash ({}) at /pro/lib/
+perl5/site_perl/5.12.2/Test/Copyright.pm line 267, <DATA> line 102.
ok 7 - Found default copyright details
ok 8 - final copyright year is uptodate
Use of uninitialized value $holder in exists at /pro/lib/perl5/site_pe
+rl/5.12.2/Test/Copyright.pm line 314.
Use of uninitialized value $holder in hash element at /pro/lib/perl5/s
+ite_perl/5.12.2/Test/Copyright.pm line 318.
Use of uninitialized value $holder in exists at /pro/lib/perl5/site_pe
+rl/5.12.2/Test/Copyright.pm line 314.
Use of uninitialized value $holder in hash element at /pro/lib/perl5/s
+ite_perl/5.12.2/Test/Copyright.pm line 318.
ok 9 - File blib/lib/Text/CSV_XS.pm has copyright statement
ok 10 - Copyright for blib/lib/Text/CSV_XS.pm is described centrally
1..10
# Looks like you failed 1 test of 10.
[download]

I think that this can all use a bit of cleanup.


Enjoy, Have FUN! H.Merijn

$ perl Makefile.PL
Checking if your kit is complete...
Looks good
Warning: prerequisite Lingua::EN::NameParse 0 not found.
Warning: prerequisite Perl6::Slurp 0 not found.
Writing Makefile for Test::Copyright
[download]

Those are two modules that I do not have installed on my test machine (with over 7100 modules installed). That looks like something overly complicated for such a simple module and I'm for sure not going to install those to test/use this new module.