++ed by:

1 non-PAUSE user(s).

Adam Kennedy


Test::XT - Generate best practice release-only tests


  use Test::XT 'WriteXT';
  # Write some specific tests:
      # Generally safe and recommended for most distributions
      'Test::Pod'            => 'xt/pod.t',
      'Test::CPAN::Meta'     => 'xt/meta.t',
      'Test::MinimumVersion' => 'xt/minimumversion.t',
      'Test::HasVersion'     => 'xt/hasversion.t',
      # Controversial history and methodology, does not install reliably.
      # Forced use leads to cargo cult of worse-than-nothing empty method stubs.
      'Test::Pod::Coverage'  => 'xt/podcoverage.t',
      # May become unreliable over time as PPI and Perl::Critic change.
      # Safe when BOTH distribution and policy file are active and maintained.
      'Test::Perl::Critic'   => 'xt/critic.t',
      # Should only be used if you hand-maintain your MANIFEST file.
      # Can be presumptive about MANIFEST.SKIP in some situations.
      'Test::DistManifest'   => 'xt/distmanifest.t',
      # Does not install reliably, does not install AT ALL on Windows.
      'Test::CheckChanges'   => 'xt/changes.t',


A number of Test modules have been written over the years to support authors. Typically, these modules have standard short test scripts documented in them that you can cut and paste into your distribution.

Unfortunately almost all of these cut-and-paste test scripts are wrong.

Either the test script runs during install time, or it runs with an out-of-date version of the test module, or the author adds the test modules as an (unnecesary) dependency at install time, or for automated testing.

Test::XT is a module intended for use in code generators, release automation and other ancillary systems. It generates an appropriate test script for various testing modules that runs in the appropriate mode for each type of execution environment.

1. End User Install

At installation time, test scripts should never ever run, even if the test modules are installed and available.

2. Automated Testing

  # Enable automated testing

During automated testing we should run the tests, but only if the testing module are already installed and at the current/latest version.

However, we should not install dependencies during automated testing, because failing to install a testing dependency means less runs on your code when the entire point of the author tests is to improve the standard of testing, not reduce it.

3. Release/Author Testing

  # Enable author tests

All tests should be run at release time by the author. Despite this, the dependencies should NEVER be added to your Makefile.PL or Build.PL, because it is far too easy to accidentally have these extra dependencies bleed through into your published META.yml.

This would cause inaccuracies in tools that track dependencies across the entire repository via the META.yml files.


To programmatically extract a list of the relevant module files, or to get informtion about supported modules, you may directly access the underlying hash:


Please note that this interface is experimental and subject to change.



    test    => 'ok_changes',
    release => 0,
    comment => 'Test that Changes has an entry for current version',
    modules => {
      'Test::CheckChanges' => '0.08',

This function provides a simple way to write a single test to a file, following the usual template. The test data is a hash (Note: it's NOT a hash reference).

The example above writes a test to xt/somefile.t that loads Test::CheckChanges if available, calling the ok_changes() function if it is. A few knobs control how this works:

  • test is the name of the subroutine to run, which has to be exported from the test module.

  • release determines whether this is a release-only test, which means it is not executed during automated testing, even if the needed prerequisites are available.

  • comment is the default comment which briefly describes the test.

  • modules is a hash reference containing pairs of modules and their required versions. If no particular version is required, use 0.


      'Test::Pod'            => 'xt/pod.t',
      'Test::CPAN::Meta'     => 'xt/meta.t',
      'Test::MinimumVersion' => 'xt/minimumversion.t',
      'Test::Perl::Critic'   => 'xt/critic.t',

This provides a convenient way to write multiple test files using the default profile settings (such as which modules to require, what subroutine to call, whether this is a release-only test, and so on).

Example code:



  my $test = Test::XT->new(
    test    => 'all_pod_files_ok',
    release => 0, # is this a RELEASE test only?
    comment => 'Test that the syntax of our POD documentation is valid',
    modules => {
      'Pod::Simple' => '3.07',
      'Test::Pod'   => '1.26',
    default => 'pod.t',

This produces a new object that stores information about a given test module. It can then be transformed into output suitable for use in a stream (via write_string, which returns the test script as a string) or for writing directly to a file (using write).

For details on the available options, see WriteTest


  $test->module( 'Foo::Bar' => '1.23' );

Add a module dependency for the test script.


  $test->test( $command );

Change the name of the subroutine which is called to actually run the test.

Code example:



  $test->write( $path )

This method writes the test file to the provided path.

Note that this method throws an exception upon failure to open the file.

Code example:

  eval {
  print "Failure writing file" if $@;



This method writes the test script as a chunk of text and returns it. Note that this is the exact script written out to file via write.

Code example:

  print "Test script:\n";
  print $test->write_string;


This module is still missing support for lots of other author tests.


This module is stored in an Open Repository at the following address:


Write access to the repository is made available automatically to any published CPAN author, and to most other volunteers on request.

If you are able to submit your bug report in the form of new (failing) unit tests, or can apply your fix directly instead of submitting a patch, you are strongly encouraged to do so. The author currently maintains over 100 modules and it may take some time to deal with non-critical bug reports or patches.

This will guarantee that your issue will be addressed in the next release of the module.

If you cannot provide a direct test or fix, or don't have time to do so, then regular bug reports are still accepted and appreciated via the CPAN bug tracker.


For other issues, for commercial enhancement and support, or to have your write access enabled for the repository, contact the author at the email address above.


Adam Kennedy <adamk@cpan.org>

Jonathan Yu <jawnsy@cpan.org>


http://use.perl.org/~Alias/journal/38822, which explains why this style of testing is beneficial to you and the CPAN-at-large.


Copyright 2009-2011 Adam Kennedy

This program is free software; you can redistribute it and/or modify it under the same terms as Perl itself.

The full text of the license can be found in the LICENSE file included with this module.

Hosting generously
sponsored by Bytemark