my @art = $schema->resultset("Artist")->search({ }, { order_by => 'name DESC'});
-cmp_ok(@art, '==', 3, "Three artists returned");
+is(@art, 3, "Three artists returned");
my $art = $art[0];
is($art->name, 'We Are In Rehab', "Accessor update ok");
my %dirty = $art->get_dirty_columns();
-cmp_ok(scalar(keys(%dirty)), '==', 1, '1 dirty column');
+is(scalar(keys(%dirty)), 1, '1 dirty column');
ok(grep($_ eq 'name', keys(%dirty)), 'name is dirty');
is($art->get_column("name"), 'We Are In Rehab', 'And via get_column');
ok($art->update, 'Update run');
my %not_dirty = $art->get_dirty_columns();
-cmp_ok(scalar(keys(%not_dirty)), '==', 0, 'Nothing is dirty');
+is(scalar(keys(%not_dirty)), 0, 'Nothing is dirty');
eval {
my $ret = $art->make_column_dirty('name2');
ok(defined($@), 'Failed to make non-existent column dirty');
$art->make_column_dirty('name');
my %fake_dirty = $art->get_dirty_columns();
-cmp_ok(scalar(keys(%fake_dirty)), '==', 1, '1 fake dirty column');
+is(scalar(keys(%fake_dirty)), 1, '1 fake dirty column');
ok(grep($_ eq 'name', keys(%fake_dirty)), 'name is fake dirty');
my $record_jp = $schema->resultset("Artist")->search(undef, { join => 'cds' })->search(undef, { prefetch => 'cds' })->next;
@art = $schema->resultset("Artist")->search({ name => 'We Are In Rehab' });
-cmp_ok(@art, '==', 1, "Changed artist returned by search");
+is(@art, 1, "Changed artist returned by search");
-cmp_ok($art[0]->artistid, '==', 3,'Correct artist too');
+is($art[0]->artistid, 3,'Correct artist too');
lives_ok (sub { $art->delete }, 'Cascading delete on Ordered has_many works' ); # real test in ordered.t
@art = $schema->resultset("Artist")->search({ });
-cmp_ok(@art, '==', 2, 'And then there were two');
+is(@art, 2, 'And then there were two');
ok(!$art->in_storage, "It knows it's dead");
@art = $schema->resultset("Artist")->search({ });
-cmp_ok(@art, '==', 3, 'And now there are three again');
+is(@art, 3, 'And now there are three again');
my $new = $schema->resultset("Artist")->create({ artistid => 4 });
-cmp_ok($new->artistid, '==', 4, 'Create produced record ok');
+is($new->artistid, 4, 'Create produced record ok');
@art = $schema->resultset("Artist")->search({ });
-cmp_ok(@art, '==', 4, "Oh my god! There's four of them!");
+is(@art, 4, "Oh my god! There's four of them!");
$new->set_column('name' => 'Man With A Fork');
my $cd = $schema->resultset("CD")->find(1);
my %cols = $cd->get_columns;
-cmp_ok(keys %cols, '==', 6, 'get_columns number of columns ok');
+is(keys %cols, 6, 'get_columns number of columns ok');
is($cols{title}, 'Spoonful of bees', 'get_columns values ok');
my( $or_rs ) = $schema->resultset("CD")->search_rs($search, { join => 'tags',
order_by => 'cdid' });
-cmp_ok($or_rs->count, '==', 5, 'Search with OR ok');
+is($or_rs->count, 5, 'Search with OR ok');
my $distinct_rs = $schema->resultset("CD")->search($search, { join => 'tags', distinct => 1 });
-cmp_ok($distinct_rs->all, '==', 4, 'DISTINCT search with OR ok');
+is($distinct_rs->all, 4, 'DISTINCT search with OR ok');
SKIP: {
skip "SQLite < 3.2.6 doesn't understand COUNT(DISTINCT())", 2
if $is_broken_sqlite;
- my $tcount = $schema->resultset("Track")->search(
+ my $tcount = $schema->resultset('Track')->search(
{},
- {
- select => {count => {distinct => ['position', 'title']}},
- as => ['count']
+ {
+ select => [ qw/position title/ ]
}
);
- cmp_ok($tcount->next->get_column('count'), '==', 13, 'multiple column COUNT DISTINCT ok');
-
- $tcount = $schema->resultset("Track")->search(
- {},
- {
- columns => {count => {count => {distinct => ['position', 'title']}}},
- }
- );
- cmp_ok($tcount->next->get_column('count'), '==', 13, 'multiple column COUNT DISTINCT using column syntax ok');
+ is($tcount->count, 13, 'multiple column COUNT DISTINCT ok');
+
+ $tcount = $schema->resultset('Track')->search(
+ {},
+ {
+ group_by => [ qw/position title/ ]
+ }
+ );
+ is($tcount->count, 13, 'multiple column COUNT DISTINCT using column syntax ok');
}
my $tag_rs = $schema->resultset('Tag')->search(
my $rel_rs = $tag_rs->search_related('cd');
-cmp_ok($rel_rs->count, '==', 5, 'Related search ok');
+is($rel_rs->count, 5, 'Related search ok');
-cmp_ok($or_rs->next->cdid, '==', $rel_rs->next->cdid, 'Related object ok');
+is($or_rs->next->cdid, $rel_rs->next->cdid, 'Related object ok');
$or_rs->reset;
$rel_rs->reset;
my $tag = $schema->resultset('Tag')->search(
[ { 'me.tag' => 'Blue' } ], { cols=>[qw/tagid/] } )->next;
-cmp_ok($tag->has_column_loaded('tagid'), '==', 1, 'Has tagid loaded');
-cmp_ok($tag->has_column_loaded('tag'), '==', 0, 'Has not tag loaded');
+ok($tag->has_column_loaded('tagid'), 'Has tagid loaded');
+ok(!$tag->has_column_loaded('tag'), 'Has not tag loaded');
ok($schema->storage(), 'Storage available');
ok($schema->source('SourceNameArtists'), 'SourceNameArtists result source exists');
my @artsn = $schema->resultset('SourceNameArtists')->search({}, { order_by => 'name DESC' });
- cmp_ok(@artsn, '==', 4, "Four artists returned");
+ is(@artsn, 4, "Four artists returned");
# make sure subclasses that don't set source_name are ok
ok($schema->source('ArtistSubclass'), 'ArtistSubclass exists');
{
my $art_del = $schema->resultset("Artist")->find({ artistid => 1 });
lives_ok (sub { $art_del->delete }, 'Cascading delete on Ordered has_many works' ); # real test in ordered.t
- cmp_ok( $schema->resultset("CD")->search({artist => 1}), '==', 0, 'Cascading through has_many top level.');
- cmp_ok( $schema->resultset("CD_to_Producer")->search({cd => 1}), '==', 0, 'Cascading through has_many children.');
+ is( $schema->resultset("CD")->search({artist => 1}), 0, 'Cascading through has_many top level.');
+ is( $schema->resultset("CD_to_Producer")->search({cd => 1}), 0, 'Cascading through has_many children.');
}
# test column_info