Going along the path of testing the examples in my distribution, I think it could be generalized. What do you think about this? Gabor
=head1 NAME Test::Example - Check if all the examples in the distribution work correctly =head1 SYNOPSIS use Test::Example; test_all_examples(); or use Test::Example; foreach my $file (glob 'myexamples/*.plx') { test_example( dir => 'myexamples', script => $file, stdout => "stdout/$file", stderr => "stderr/$file", ); } =head1 METHODS =head2 test_all_examples Goes over all the .pl files in the eg/ examples/ /sample/ (...?) directories runs each one of the scripts using L<test_example>. Options given to test_example are: test_example( dir => 'eg', # the name of the relevant directory script => 'scriptname.pl', # the name of the current .pl file stdin => 'scriptname.pl_stdin', stdout => 'scriptname.pl_stdout', stderr => 'scriptname.pl_stderr', ); =head2 test_all_examples_do The same as test_all_examples but =head2 test_example test_example( dir => 'myexamples', script => 'doit.pl', stdin => 'file_providing_stdin', stdout => 'file_listing_expected_output_of_doit', stderr => 'file_listing_expected_errors_of_doit', argv => ['command', 'line', 'arguments'], ); Before running doit.pl chdirs into the 'myexamples' directory. doit.pl is executed using system. The list of values provided as argv are supplied as command line parameters. Its STDIN is redirected from the file that is given as 'stdin'. Its STDOUT and STDERR are captured. In short, something like this: chdir 'myexamples'; system("$h{script} @{ $h{argv} } < $h{stdin} > temp_out 2> temp_err"); Once the script finished the content of temp_out is compared to the expeced output and the content of temp_err to the expected errors. If no 'stderr' key provided then the expectation is that nothing will be printed to STDERR. =head2 test_example_do The same as L</test_example> but instead of using C<system> to run the external script it will use C<do 'scriptname.pl'> =cut