Warnings when warnings enabled(!)

[pghmcfc]

I've not yet been able to track down where it's happening but in some environments I'm getting warnings enabled when the Module::Load test suite is running, and that results in lots of warnings appearing, e.g.:

Use of uninitialized value $cnt:: in concatenation (.) or string at t/02_Module-Load.t line 23.
Use of uninitialized value in hash slice at /home/stfn/.cpan/build/Module-Load-0.28-fjZxuz/blib/lib/Module/Load.pm line 20.

A full example of this can be seen at http://www.cpantesters.org/cpan/report/d259755c-7d04-11e3-a289-770048f56e94

Explicitly turning off these warnings seems to work:

--- lib/Module/Load.pm
+++ lib/Module/Load.pm
@@ -3,6 +3,7 @@ package Module::Load;
 $VERSION = '0.28';

 use strict;
+use warnings;
 use File::Spec ();

 sub import {
@@ -10,6 +11,7 @@ sub import {
     my $h; shift;

     {   no strict 'refs';
+        no warnings 'uninitialized';

         @_ or (
             *{"${who}::load"} = \&load, # compat to prev version
--- t/02_Module-Load.t
+++ t/02_Module-Load.t
@@ -2,6 +2,7 @@

 use Test::More;
 use strict;
+use warnings;

 #
 # Module::Load; test new features:
@@ -18,6 +19,7 @@ use strict;
 my ($afx, $cnt, $r, $tcode) = ('TestXYZ_', 0);

 sub _reset{
+    no warnings 'uninitialized';
     undef %{Data::Dumper::};
     undef %{XYZ::Module::};
     eval "undef %{$afx$cnt::}";

I think a better fix would be to make sure that the variables in question are always initialized though.

[magnolia-k]

https://github.com/magnolia-k/module-load/commit/f40620369cdc227798de2ce52e9889c1cbf49929

I wrote the code not using 'no warnings'.

Isn't there any this one well?

[pghmcfc]

Sorry, I don't understand the question.

[magnolia-k]

I think that I should not use 'no warnings' if possible.

[pghmcfc]

I agree, hence my comment about "a better fix" in the original report.

[magnolia-k]

I am sorry, I had overlooked the last paragraph. I also agree.

[toddr]

Looking at the supplied diff, the current code doesn't seem to match the suggestion and at best the patch has essentially already been applied. I suggest closing this case.