| Subject: | A fix for the random failure of the last test |
Hello,
The last test of t/basic.t fails randomly because the second sort you're
applying does not describe a total order. Both what you expect :
my @expected = (
[ [ "Jose", "Wales" ], [ "John", "Wall" ] ],
[ [ "Jose", "Wales" ], [ "John", "Wayne" ] ],
[ [ "John", "Wall" ], [ "John", "Wayne" ] ]
);
and what you get after the second sort (on my machine)
my @got = (
[ [ 'Jose', 'Wales' ], [ 'John', 'Wayne' ] ],
[ [ 'Jose', 'Wales' ], [ 'John', 'Wall' ] ],
[ [ 'John', 'Wall' ], [ 'John', 'Wayne' ] ]
);
are sorted according to
sort { ($a->[0][1] cmp $b->[0][1]) || ($a->[0][0] cmp $b->[0][0]) }
@array
but that's because your criterion considers that [ [ 'Jose', 'Wales' ],
[ 'John', 'Wayne' ] ] and [ 'Jose', 'Wales' ], [ 'John', 'Wall' ] ]
are equal.
The fix is just to also look at the second element. Patch attached. I
think the order may differ on different machines because
Math::Combinatorics make no guarantee on the order on which it iterates
on the pairs, and that using Test::More instead of Test::Leaner may
magically change that order.
Vincent.
| Subject: | lingua_en_similarnames_levenshtein_sort.patch |
--- t/basic.t 2011-08-24 19:26:02.000000000 +0200
+++ t/basic.t 2011-08-24 19:27:53.000000000 +0200
@@ -47,10 +47,8 @@
my @person_pair = sort { ($a->[1] cmp $b->[1]) || ($a->[0] cmp $b->[0]) } @{$person_pair};
push @people_sorted, \@person_pair;
}
-@people_sorted = sort { ($a->[0][1] cmp $b->[0][1]) || ($a->[0][0] cmp $b->[0][0]) } @people_sorted;
-# cpantest is regularly choking on this test and I don't now why
-# The sort above works reliably for me to get the results to match up with what's expected.
-#is_deeply($expected_list, \@people_sorted, 'People with Similar Names');
+@people_sorted = sort { ($a->[0][1] cmp $b->[0][1]) || ($a->[0][0] cmp $b->[0][0]) || ($a->[1][1] cmp $b->[1][1]) || ($a->[1][0] cmp $b->[1][0]) } @people_sorted;
+is_deeply($expected_list, \@people_sorted, 'People with Similar Names');
done_testing();