nope, don't want to do this

this meant that modules that we hadn't seen before were always assumed
to have changed, which can cause arbitrary (and possibly dangerous)
modules to be reloaded.

this can happen when you call ->refresh, then load a module, then call
->refresh again. in this case, there's no way to tell if the module has
been modified since it was loaded, which means that we have to assume
one way or the other. assuming that it hasn't can cause changes to be
missed when ->refresh is called, but assuming that it has can cause
modules (such as Moose) to be reloaded when we really can't support it.

1 files changed, 0 insertions, 6 deletions

diff --git a/lib/Class/Refresh.pm b/lib/Class/Refresh.pm
index 2cbf0ca..24569ab 100644
--- a/lib/Class/Refresh.pm
+++ b/lib/Class/Refresh.pm
@@ -74,11 +74,6 @@ Returns a list of modules which have changed since the last call to C<refresh>.
 sub modified_modules {
     my $class = shift;
 
-    if (!%CACHE) {
-        $class->_update_cache_for($_) for keys %INC;
-        return;
-    }
-
     my @ret;
     for my $file (keys %CACHE) {
         # refresh files that are in our
@@ -94,7 +89,6 @@ sub modified_modules {
         }
         else {
             $class->_update_cache_for($file);
-            push @ret, $class->_file_to_mod($file);
         }
     }