Test::Class::Moose - Serious testing for serious Perl


version 0.99


    package TestsFor::DateTime;
    use Test::Class::Moose;
    use DateTime;

    # methods that begin with test_ are test methods.
    sub test_constructor {
        my $test = shift;
        $test->test_report->plan(3);    # strictly optional

        can_ok 'DateTime', 'new';
        my %args = (
            year  => 1967,
            month => 6,
            day   => 20,
        isa_ok my $date = DateTime->new(%args), 'DateTime';
        is $date->year, $args{year}, '... and the year should be correct';



See the Test::Class::Moose home page for a summary.

Test::Class::Moose is a powerful testing framework for Perl. Out of the box you get:

  • Reporting

  • Extensibility

  • Tagging tests

  • Parallel testing

  • Test inheritance

  • Write your tests using Moose

  • All the testing functions and behavior from Test::Most

  • Event handlers for startup, setup, teardown, and shutdown of test classes

Better docs will come later. You should already know how to use Moose and Test::Class.


Inheriting from Test::Class::Moose

Just use Test::Class::Moose. That's all. You'll get all Test::Most test functions, too, along with strict and warnings. You can use all Moose behavior, too.

When you use Test::Class::Moose it inserts itself as a parent class for your test class. This means that if you try to use extends in your test class you will break things unless you include Test::Class::Moose as a parent. We recommend that you use roles in your test classes instead.

Declare a test method

All method names that begin with test_ are test methods. Methods that do not are not test methods.

 sub test_this_is_a_method {
     my $test = shift;

     ok 1, 'whee!';

 sub this_is_not_a_test_method {
    my $test = shift;
    # but you can, of course, call it like normal

You may specify Test and Tests method attributes, just like in Test::Class and the method will automatically be a test method, even if does not start with test_:

    sub this_is_a_test : Test {
        pass 'we have a single test';

    sub another_test_method : Tests { # like "no_plan"
        # a bunch of tests

    sub yet_another_test_method : Tests(7) { # sets plan to 7 tests

Note: Prior to version 0.51, this feature only worked if you had the optional Sub::Attribute installed.


No plans needed. The test suite declares a plan of the number of test classes.

Each test class is a subtest declaring a plan of the number of test methods.

Each test method relies on an implicit done_testing call.

If you prefer, you can declare a plan in a test method:

    sub test_something {
        my $test = shift;

Or with a Tests attribute:

    sub test_something : Tests(3) {
        my $test = shift;

You may call plan() multiple times for a given test method. Each call to plan() will add that number of tests to the plan. For example, with a method modifier:

    before 'test_something' => sub {
        my $test = shift;

        # more tests

Please note that if you call plan, the plan will still show up at the end of the subtest run, but you'll get the desired failure if the number of tests run does not match the plan.

Inheriting from another Test::Class::Moose class

List it as the extends in the import list. If the base class does not use (or extend) Test::Class::Moose, then a compile-time error is thrown.

 package TestsFor::Some::Class::Subclass;
 use Test::Class::Moose extends => 'TestsFor::Some::Class';

 sub test_me {
     my $test  = shift;
     my $class = $test->test_class;
     ok 1, "I overrode my parent! ($class)";

 before 'test_this_baby' => sub {
     my $test  = shift;
     my $class = $test->test_class;
     pass "This should run before my parent method ($class)";

 sub this_should_not_run {
     my $test = shift;
     fail "We should never see this test";

 sub test_this_should_be_run {
     for ( 1 .. 5 ) {
         pass "This is test number $_ in this method";


Skipping Test::Most

By default, when you use Test::Class::Moose in your own test class, it exports all the subs from Test::Most into your class. If you'd prefer to import a different set of test tools, you can pass bare => 1 when using Test::Class::Moose:

 use Test::Class::Moose bare => 1;

When you pass this, Test::Class::Moose will not export Test::Most's subs into your class. You will have to explicitly import something like Test::More or Test2::Tools::Compare in order to actually perform tests.

Custom Test Toolkits

If you'd like to provide a custom set of test modules to all of your test classes, this is easily done with Import::Into:

  package MM::Test::Class::Moose;

  use strict;
  use warnings;
  use namespace::autoclean ();

  use Import::Into;
  use Test::Class::Moose ();
  use Test::Fatal;
  use Test::More;

  sub import {
      my @imports = qw(

      my $caller_level = 1;
      $_->import::into($caller_level) for @imports;

You could also create a kit in a separate module like My::Test::Kit using Test::Kit and then simply export that from your My::Test::Class::Moose module with Import::Into.


The test control methods are there to ensure that your tests are running with all of the resources they need. For example, database transactions might be started before a test method and rolled back after the test method. Fixtures needed for every test might be loaded and cleaned up. However you use them, it's important to understand when and how they're run.

1. test_startup — Runs once when the test class starts up
2. test_setup - Runs before each test method
3. test_teardown - Runs after each test method
4. test_shutdown - Runs once when the test class shuts down

Important: Do not run tests in test control methods. This will cause the test control method to fail (this is a feature, not a bug). If a test control method fails, the class/method will fail and testing for that class should stop.

The available test control methods are:


 sub test_startup {
    my $test = shift;
    $test->next::method;   # run this before your test_startup code
    # more startup

Runs at the start of each test class. Quite often the base class that you inherit from will have its own test_startup code running (such as starting a database transaction or connecting to an external resource). You almost always want to call $test->next::method before your own setup code. This ensures that the environment is set up to safely run your code. For example, if the parent test_startup starts a database transaction with the expectation that the test_teardown will end the database transactions, you can safely load database fixtures after that is run.

If you need to know the name of the class you're running this in (though usually you shouldn't), use $test->test_class, or you can do this:

    sub test_startup {
        my $test                 = shift;
        my $report               = $test->test_report;
        my $instance             = $report->current_instance->name;
        my $upcoming_test_method = $report->current_method->name;

The $test->test_report object is a Test::Class::Moose::Report::Instance object.


 sub test_setup {
    my $test = shift;
    $test->next::method;    # run this before your test_setup code
    # more setup

Runs at the start of each test method. If you must know the name of the test you're about to run, you can do this:

 sub test_setup {
    my $test = shift;
    my $test_method = $test->test_report->current_method->name;
    # do something with it


 sub test_teardown {
    my $test = shift;
    # more teardown
    $test->next::method;    # run this after your test_teardown code

Runs at the end of each test method.

By default, this is not run if the test class is skipped entirely. You can override the run_control_methods_on_skip in your class to return a true value in order to force this method to be run when the class is skipped.


 sub test_shutdown {
     my $test = shift;
     # more teardown
     $test->next::method;    # run this after your test_shutdown code

Runs at the end of each test class.

By default, this is not run if the test class is skipped entirely. You can override the run_control_methods_on_skip in your class to return a true value in order to force this method to be run when the class is skipped.

Overriding Test Control Methods

To override a test control method, just remember that this is OO:

 sub test_setup {
     my $test = shift;
     $test->next::method; # call parent test_setup
     # more setup code here


This feature is still considered experimental.

By default, each test class you create will be instantiated once. However, you can tell the Test::Class::Moose::Runner to create multiple instances of a test class.

To do this, simply consume the Test::Class::Moose::Role::ParameterizedInstances role in your test class. This role requires you to implement a _constructor_parameter_sets method in your test class. That method will be called as a class method. It is expected to return a list of key/value pairs. The keys are the name of the instance and the values are hashrefs of attributes to be passed to your test class's constructor. Here's a really dumb example:

 package TestsFor::PlainAndFancy;
 use Test::Class::Moose;
 with 'Test::Class::Moose::Role::ParameterizedInstances';

 has is_fancy => (
     is       => 'ro',
     isa      => 'Bool',
     required => 1,

 sub _constructor_parameter_sets {
     my $class = shift;
     return (
         "$class - plain" => { is_fancy => 0 },
         "$class - fancy" => { is_fancy => 1 },

 sub test_something { ... }

The test runner will run all the test methods in your class once per instance, and each instance will be run in its own subtest. You can dynamically decide to skip your test class completely by having _constructor_parameter_sets return an empty list.

Note that this feature has great potential for abuse, so use it cautiously. That said, there are cases where this feature can greatly simplify your test code.


See the docs for Test::Class::Moose::Runner for details on running your test suite. If you'd like to get up and running quickly, here's a very simple test file you can use:

 use Test::Class::Moose::CLI;

Put this in a file like t/run-test-class.t. When you run it with prove it will load all the test classes defined in t/lib and run them sequentially.

See the documentation for Test::Class::Moose::CLI on the options you can pass when running tests.

Skipping Classes and Methods

If you wish to skip a class, set the reason in the test_startup method.

    sub test_startup {
        my $test = shift;
        $test->test_skip("I don't want to run this class");

If you are using test class instances, you can also make _constructor_parameter_sets return an empty list, which will result in the class being skipped.

Note that if you run test_skip, the test_shutdown method will also be skipped. This is due to the assumption that you might not have run any setup code and thus you don't need shutdown code. However, if you do need to run shutdown, you can override the run_control_methods_on_skip method to return true:

    sub run_control_methods_on_skip {1}

If you wish to skip an individual method, do so in the test_setup method.

    sub test_setup {
        my $test = shift;
        my $test_method = $test->test_report->current_method;

        if ( 'test_time_travel' eq $test_method->name ) {
            $test->test_skip("Time travel not yet available");

The "Tests" and "Test" Attributes

If you're comfortable with Test::Class, you know that test methods methods are declared in Test::Class with Test (for a method with a single test) or Tests, for a method with multiple tests. This also works for Test::Class::Moose. Test methods declared this way do not need to start with test_.

    sub something_we_want_to_check : Test {
        # this method may have only one test

    sub something_else_to_check : Tests {
        # this method may have multiple tests

    sub another_test_method : Tests(3) {
        # this method must have exactly 3 tests

If a test method overrides a parent test method and calls it, their plans will be added together:

    package TestsFor::Parent;

    use Test::Class::Moose;

    sub some_test : Tests(3) {
        # three tests

And later:

    package TestsFor::Child;

    use Test::Class::Moose extends => 'TestsFor::Parent';

    sub some_test : Tests(2) {
        my $test = shift;
        # 2 tests here

In the above example, TestsFor::Parent::some_test will run three tests, but TestsFor::Child::some_test will run five tests (two tests, plus the three from the parent).

Note that if a plan is explicitly declared, any modifiers or overriding methods calling the original method will also have to assert the number of tests to ensure the plan is correct. The above TestsFor::Parent and TestsFor::Child code would fail if the child's some_test method attribute was Tests without the number of tests asserted.

Do not use Test or Tests with test control methods because you don't run tests in those.

Tagging Methods

Sometimes you want to be able to assign metadata to help you better manage your test suite. You can do this with tags:

    sub test_save_poll_data : Tags(api network) {

Tags are strictly optional and you can provide one or more tags for each test method with a space separated list of tags. You can use this to filter your tests suite, if desired. For example, if your network goes down and all tests which rely on a network are tagged with network, you can skip those tests with this:

    Test::Class::Moose::Runner->new( exclude_tags => 'network' )->runtests;

Or maybe you want to run all api and database tests, but skip those marked deprecated:

        include_tags => [qw/api database/],
        exclude_tags => 'deprecated',

You can also inspect tags within your test classes:

    sub test_setup {
        my $test          = shift;
        my $method_to_run = $test->test_report->current_method;
        if ( $method_to_run->has_tag('db') ) {

Tagging support relies on Sub::Attribute. If this module is not available, include_tags and exclude_tags will be ignored, but a warning will be issued if those are seen. Prior to version 0.51, Sub::Attribute was optional. Now it's mandatory, so those features should always work.


... but probably shouldn't.

As a general rule, methods beginning with /^test_/ are reserved for Test::Class::Moose. This makes it easier to remember what you can and cannot override. However, any test with Test or Tests are test methods regardless of their names.


 my $report = $test->test_report;

Returns the Test::Class::Moose::Report object. Useful if you want to do your own reporting and not rely on the default output provided with the statistics boolean option.

You can also call it in test classes (most useful in the test_setup() method):

    sub test_setup {
        my $test = shift;
        my $report = $test->test_report;
        my $instance = $test->current_instance;
        my $method = $test->current_method; # the test method we're about to run
        if ( $method->name =~ /customer/ ) {
        # or better still
        if ( $method->has_tag('customer') ) {


 my $class = $test->test_class;

Returns the name for this test class. Useful if you rebless an object (such as applying a role at runtime) and don't want to lose the original class name.


You may override this in a subclass. Currently returns all methods in a test class that start with test_ (except for the test control methods).

Please note that the behavior for include and exclude is also contained in this method. If you override it, you will need to account for those yourself.


Sadly, we have an import method. This is used to automatically provide you with all of the Test::Most behavior.


We use nested tests (subtests) at each level:

    # Executing tests for TestsFor::Basic::Subclass
        # TestsFor::Basic::Subclass->test_me()
            ok 1 - I overrode my parent! (TestsFor::Basic::Subclass)
        ok 1 - test_me
        # TestsFor::Basic::Subclass->test_this_baby()
            ok 1 - This should run before my parent method (TestsFor::Basic::Subclass)
            ok 2 - whee! (TestsFor::Basic::Subclass)
        ok 2 - test_this_baby
        # TestsFor::Basic::Subclass->test_this_should_be_run()
            ok 1 - This is test number 1 in this method
            ok 2 - This is test number 2 in this method
            ok 3 - This is test number 3 in this method
            ok 4 - This is test number 4 in this method
            ok 5 - This is test number 5 in this method
        ok 3 - test_this_should_be_run
    ok 1 - TestsFor::Basic::Subclass
    # Executing tests for TestsFor::Basic
        # TestsFor::Basic->test_me()
            ok 1 - test_me() ran (TestsFor::Basic)
            ok 2 - this is another test (TestsFor::Basic)
        ok 1 - test_me
        # TestsFor::Basic->test_this_baby()
            ok 1 - whee! (TestsFor::Basic)
        ok 2 - test_this_baby
    ok 2 - TestsFor::Basic
    # Test classes:    2
    # Test methods:    5
    # Total tests run: 11
    All tests successful.
    Files=1, Tests=2,  2 wallclock secs ( 0.03 usr  0.00 sys +  0.27 cusr  0.01 csys =  0.31 CPU)
    Result: PASS


See Test::Class::Moose::Report for more detailed information on reporting.

Reporting features are subject to change.

Sometimes you want more information about your test classes, it's time to do some reporting. Maybe you even want some tests for your reporting. If you do that, run the test suite in a subtest (because the plans will otherwise be wrong).

    #!/usr/bin/env perl
    use lib 'lib';
    use Test::Most;
    use Test::Class::Moose::Load qw(t/lib);
    use Test::Class::Moose::Runner;

    my $test_suite = Test::Class::Moose::Runner->new;

    subtest 'run the test suite' => sub {
    my $report = $test_suite->test_report;

    foreach my $class ( $report->all_test_instances ) {
        my $class_name = $class->name;
        ok !$class->is_skipped, "$class_name was not skipped";
        ok $class->passed, "$class_name passed";

        subtest "$class_name methods" => sub {
            foreach my $method ( $class->all_test_methods ) {
                my $method_name = $method->name;
                ok $method->passed, "$method_name passed";

                ok !$method->is_skipped, "$method_name was not skipped";
                cmp_ok $method->num_tests, '>', 0,
                  '... and some tests should have been run';
                diag "Run time for $method_name: ".$method->time->duration;
        my $time   = $class->time;
        diag "Run time for $class_name: ".$class->time->duration;

        my $real   = $time->real;
        my $user   = $time->user;
        my $system = $time->system;
        # do with these as you will
    diag "Number of test classes: "   . $report->num_test_classes;
    diag "Number of test instances: " . $report->num_test_instances;
    diag "Number of test methods: "   . $report->num_test_methods;
    diag "Number of tests:        "   . $report->num_tests;


If you just want to output reporting information, you do not need to run the test suite in a subtest:

    my $test_suite = Test::Class::Moose::Runner->new->runtests;
    my $report     = $test_suite->test_report;

Or even shorter:

    my $report = Test::Class::Moose::Runner->new->runtests->test_report;


If you would like Test::Class::Moose to take care of loading your classes for you, see Test::Class::Moose::Role::AutoUse in this distribution.


Version 0.79

  • The Test::Class::Moose::Config class's args method is now deprecated. This was a holdover from when Test::Class::Moose was both a parent class for your test classes and the test class runner.

Version 0.77

  • The passing of the report object as an argument to test methods and test control methods is now deprecated. You can get the report from the test class object itself via the $test->test_report method.

  • The Test::Class::Moose->runtests method has been removed. Use Test::Class::Moose::Runner to run your test classes.

  • The Test::Class::Moose::Role::Parallel role has been removed. This has not done anything except issue a warning since version 0.55.

Version 0.75

  • The test_teardown method is no longer run when a test is skipped unless run_control_methods_on_skip returns a true value. The test_teardown method was never intended to be run unconditionally.

  • Parallel testing now parallelizes test classes rather than individual test instances. This is only relevant if your test suite contains parameterized test classes. This is slightly less efficient, but made the internal test running code much simpler and made it possible to fix reporting for parallel test runs.

  • The Test::Class::Moose::Config builder method has been removed.

  • The Test::Class::Moose::Runner builder method has been removed.

Version 0.67

Version 0.55

  • Running tests with Test::Class::Moose is deprecated - use Test::Class::Moose::Runner

    As of version 0.55, running tests and being a test class have been separated. Your test classes should continue to use Test::Class::Moose, but your test runner script should use Test::Class::Moose::Runner:

     use Test::Class::Moose::Load 't/lib';
     use Test::Class::Moose::Runner;

    Calling Test::Class::Moose->new->runtests still works, but is deprecated and will issue a warning.

  • Parallel testing is totally different

    The Test::Class::Moose::Role::Parallel role won't do anything other than issue a warning. See the Test::Class::Moose::Runner docs for details on running tests in parallel.

  • The Test::Class::Moose::Report all_test_classes method is deprecated

    This has been replaced with the all_test_instances method. The all_test_classes method is still present for backwards compatibility, but it simply calls all_test_instances under the hood.

  • The Test::Class::Moose::Report::Class class is gone

    It has been replaced by the Test::Class::Moose::Report::Instance class, which has the same API.

  • The Test::Class::Moose::Report::Method class_report method has been renamed

    This is now called instance_report.

Version 0.40

  • test_reporting

    As of version 0.40, the long deprecated method test_reporting has now been removed.

  • $report argument to methods deprecated

    Prior to version 0.40, you used to have a second argument to all test methods and test control methods:

        sub test_something {
            my ( $test, $report ) = @_;

    This was annoying. It was doubly annoying in test control methods in case you forgot it:

        sub test_setup {
            my ( $test, $report ) = @_;
            $test->next::method; # oops, needed $report

    That second argument is still passed, but it's deprecated. It's now recommended that you call the $test->test_report method to get that. Instead of this:

        sub test_froblinator {
            my ( $test, $report ) = @_;

    You write this:

        sub test_froblinator {
            my $test = shift;


  • Callbacks for tags (for example, 'critical' tags could bailout)

  • New test phases - start and end suite, not just start and end class/method


You can find documentation for this module with the perldoc command.

    perldoc Test::Class::Moose

You can also look for information at:



Bugs may be submitted at

I am also usually active on IRC as 'autarch' on irc://


The source code repository for Test-Class-Moose can be found at


  • Curtis "Ovid" Poe <>

  • Dave Rolsky <>


  • Andy Jack <>

  • Christopher Layne <>

  • Chuck Adams <>

  • Denny de la Haye <>

  • Desmond Daignault <>

  • Doug Bell <>

  • Gregory Oschwald <>

  • Harald Jörg <>

  • Iain Loasby <>

  • Jeremy Krieg <>

  • Jonathan C. Otsuka <>

  • Jonathan Stowe <>

  • jrubinator <>

  • Karen Etheridge <>

  • Larry Leszczynski <>

  • mark-5 <>

  • mephinet <>

  • Neil Bowers <>

  • Olaf Alders <>

  • Paul Boyd <>

  • Paul Williams <>

  • Petrea Corneliu Stefan <>

  • Steven Humphrey <>

  • Stuckdownawell <>

  • Tim Vroom <>

  • Tom Beresford <>

  • Tom Heady <>

  • Udo Oji <>


This software is copyright (c) 2012 - 2021 by Curtis "Ovid" Poe.

This is free software; you can redistribute it and/or modify it under the same terms as the Perl 5 programming language system itself.

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