How to use (interpret results from) Test::LeakTrace?

[Anonymous Monk]

In Release::Checklist, there's brief recommendation to use Test::LeakTrace as (re-phrasing) "help available to trace memory leaks to prevent crashes with out-of-memory after 4 days for long running process". OK.

use strict;
use warnings;
use Test::LeakTrace;
use Text::CSV_XS 'csv';

leaktrace {
    csv( in => \'1,2' );
};

This script produces a long list of 'leaks'. So what?

Then an example method_cache.pl, included with distribution, reports 'leaks' as well. I see nothing wrong with the code block. Shouldn't I use code like this?

The Test::LeakTrace is ++-ed and at stage 4 up river. Checking reverse dependencies, there's e.g. List::MoreUtils (up-river itself), but (maybe I'm wrong) dependency on Test::LeakTrace seems auto generated and not actually used.

[Anonymous Monk]

aargh, should have SS-ed: What does Test::LeakTrace do?, sorry

[Anonymous Monk]

And yet, "-6" is what the "no_leaks_ok" reports as result <= 0 i.e. the conclusion is drawn there are no leaks. But what are the ones on subsequent calls?

use strict;
use warnings;
use feature 'say';
use Test::LeakTrace;
use Text::CSV_XS 'csv';

say leaked_count { my $ret = csv( in => \'1,2' ); } for 1 .. 12;

__END__

25
-6
1
1
1
1
1
1
1
1
1
1