Test2::Manual::Testing::Migrating - How to migrate existing tests from Test::More to Test2.
This tutorial covers the conversion of an existing test. This tutorial assumes you have a test written using Test::More.
This tutorial will be converting this example test one section at a time:
t/example.t
:
#####################
# Boilerplate
use strict;
use warnings;
use Test::More tests => 14;
use_ok 'Scalar::Util';
require_ok 'Exporter';
#####################
# Simple assertions (no changes)
ok(1, "pass");
is("apple", "apple", "Simple string compare");
like("foo bar baz", qr/bar/, "Regex match");
#####################
# Todo
{
local $TODO = "These are todo";
ok(0, "oops");
}
#####################
# Deep comparisons
is_deeply([1, 2, 3], [1, 2, 3], "Deep comparison");
#####################
# Comparing references
my $ref = [1];
is($ref, $ref, "Check that we have the same ref both times");
#####################
# Things that are gone
ok(eq_array([1], [1]), "array comparison");
ok(eq_hash({a => 1}, {a => 1}), "hash comparison");
ok(eq_set([1, 3, 2], [1, 2, 3]), "set comparison");
note explain([1, 2, 3]);
{
package THING;
sub new { bless({}, shift) }
}
my $thing = new_ok('THING');
#####################
# Tools that changed
isa_ok($thing, 'THING', '$thing');
can_ok(__PACKAGE__, qw/ok is/);
BEFORE:
use strict;
use warnings;
use Test::More tests => 14;
use_ok 'Scalar::Util';
require_ok 'Exporter';
AFTER:
use Test2::V0;
plan(11);
use Scalar::Util;
require Exporter;
Test2::V0 is the recommended bundle. In a full migration you will want to replace Test::More with the Test2::V0 bundle.
Note: You should always double check the latest Test2 to see if there is a new recommended bundle. When writing a new test you should always use the newest Test::V# module. Higher numbers are newer version.
When srand is on (default) it can cause problems with things like File::Temp which will end up attempting the same "random" filenames for every test process started on a given day (or sharing the same seed).
If this is a problem for you then please disable srand when loading Test2::V0:
use Test2::V0 -no_srand => 1;
use_ok()
has been removed. a use MODULE
statement will throw an exception on failure anyway preventing the test from passing.
If you REALLY want/need to assert that the file loaded you can use the ok module:
use ok 'Scalar::Util';
The main difference here is that there is a space instead of an underscore.
require_ok
has been removed just like use_ok
. There is no ok module equivalent here. Just use require
.
The Test2::V0 bundle turns strict and warnings on for you.
Test2 does not allow you to set the plan at import. In the old code you would pass tests => 11
as an import argument. In Test2 you either need to use the plan()
function to set the plan, or use done_testing()
at the end of the test.
If your test already uses done_testing()
you can keep that and no plan changes are necessary.
Note: We are also changing the plan from 14 to 11, that is because we dropped use_ok
, require_ok
, and we will be dropping one more later on. This is why done_testing()
is recommended over a set plan.
The vast majority of assertions will not need any changes:
#####################
# Simple assertions (no changes)
ok(1, "pass");
is("apple", "apple", "Simple string compare");
like("foo bar baz", qr/bar/, "Regex match");
{
local $TODO = "These are todo";
ok(0, "oops");
}
The $TODO
package variable is gone. You now have a todo()
function.
There are 2 ways this can be used:
todo "These are todo" => sub {
ok(0, "oops");
};
This is the cleanest way to do a todo. This will make all assertions inside the codeblock into TODO assertions.
{
my $TODO = todo "These are todo";
ok(0, "oops");
}
This is a system that emulates the old way. Instead of modifying a global $TODO
variable you create a todo object with the todo()
function and assign it to a lexical variable. Once the todo object falls out of scope the TODO ends.
is_deeply([1, 2, 3], [1, 2, 3], "Deep comparison");
Deep comparisons are easy, simply replace is_deeply()
with is()
.
is([1, 2, 3], [1, 2, 3], "Deep comparison");
my $ref = [1];
is($ref, $ref, "Check that we have the same ref both times");
The is()
function provided by Test::More forces both arguments into strings, which makes this a comparison of the reference addresses. Test2's is()
function is a deep comparison, so this will still pass, but fails to actually test what we want (that both references are the same exact ref, not just identical structures.)
We now have the ref_is()
function that does what we really want, it ensures both references are the same reference. This function does the job better than the original, which could be thrown off by string overloading.
my $ref = [1];
ref_is($ref, $ref, "Check that we have the same ref both times");
ok(eq_array([1], [1]), "array comparison");
ok(eq_hash({a => 1}, {a => 1}), "hash comparison");
ok(eq_set([1, 3, 2], [1, 2, 3]), "set comparison");
note explain([1, 2, 3]);
{
package THING;
sub new { bless({}, shift) }
}
my $thing = new_ok('THING');
eq_array
, eq_hash
and eq_set
have been considered deprecated for a very long time, Test2 does not provide them at all. Instead you can just use is()
:
is([1], [1], "array comparison");
is({a => 1}, {a => 1}, "hash comparison");
eq_set
is a tad more complicated, see Test2::Tools::Compare for an explanation:
is([1, 3, 2], bag { item 1; item 2; item 3; end }, "set comparison");
explain()
has a rocky history. There have been arguments about how it should work. Test2 decided to simply not include explain()
to avoid the arguments. You can instead directly use Data::Dumper:
use Data::Dumper;
note Dumper([1, 2, 3]);
new_ok()
is gone. The implementation was complicated, and did not add much value:
{
package THING;
sub new { bless({}, shift) }
}
my $thing = THING->new;
ok($thing, "made a new thing");
The complete section after the conversion is:
is([1], [1], "array comparison");
is({a => 1}, {a => 1}, "hash comparison");
is([1, 3, 2], bag { item 1; item 2; item 3; end }, "set comparison");
use Data::Dumper;
note Dumper([1, 2, 3]);
{
package THING;
sub new { bless({}, shift) }
}
my $thing = THING->new;
ok($thing, "made a new thing");
isa_ok($thing, 'THING', '$thing');
can_ok(__PACKAGE__, qw/ok is/);
In Test::More these functions are very confusing, and most people use them wrong!
isa_ok()
from Test::More takes a thing, a class/reftype to check, and then uses the third argument as an alternative display name for the first argument (NOT a test name!).
can_ok()
from Test::More is not consistent with isa_ok
as all arguments after the first are subroutine names.
Test2 fixes this by making both functions consistent and obvious:
isa_ok($thing, ['THING'], 'got a THING');
can_ok(__PACKAGE__, [qw/ok is/], "have expected subs");
You will note that both functions take a thing, an arrayref as the second argument, then a test name as the third argument.
#####################
# Boilerplate
use Test2::V0;
plan(11);
use Scalar::Util;
require Exporter;
#####################
# Simple assertions (no changes)
ok(1, "pass");
is("apple", "apple", "Simple string compare");
like("foo bar baz", qr/bar/, "Regex match");
#####################
# Todo
todo "These are todo" => sub {
ok(0, "oops");
};
#####################
# Deep comparisons
is([1, 2, 3], [1, 2, 3], "Deep comparison");
#####################
# Comparing references
my $ref = [1];
ref_is($ref, $ref, "Check that we have the same ref both times");
#####################
# Things that are gone
is([1], [1], "array comparison");
is({a => 1}, {a => 1}, "hash comparison");
is([1, 3, 2], bag { item 1; item 2; item 3; end }, "set comparison");
use Data::Dumper;
note Dumper([1, 2, 3]);
{
package THING;
sub new { bless({}, shift) }
}
my $thing = THING->new;
#####################
# Tools that changed
isa_ok($thing, ['THING'], 'got a THING');
can_ok(__PACKAGE__, [qw/ok is/], "have expected subs");
Test2::Manual - Primary index of the manual.
The source code repository for Test2-Manual can be found at https://github.com/Test-More/Test2-Suite/.
Copyright 2018 Chad Granum <exodist@cpan.org>.
This program is free software; you can redistribute it and/or modify it under the same terms as Perl itself.
See http://dev.perl.org/licenses/