stevieb9/async-event-interval

async-event-interval's Issues

Check if 'destroy' in events shared hash works if program crashes

Repeated error handling in start() can be deduplicated after cb execution regardless of interval or not

start() check for is running is faulty

it's checking $self->{stop}, which is wrong. We need to set a {started} flag, add a status() method that checks for started, and return that.

_rand_shm_key() may produce dups

Which would result in overwriting a previous entry in the %entries hash.

start() should accept the callback params, not new()

When in new(), the same parameters are sent in on every restart.

Send them to start(), then pass them through to _event().

add ability to return a value from the event closure

send in a variable name as a param, and set it in the callback. then add a fetch() or equivalent method to fetch this data

Allow shared_scalar() to take a custom shm key

This will allow two separate scripts to load up the same variable. Needs also to have a way to identify which script should destroy the memory segment and semaphores.

Two events in one script appear to leak a segment

Test script 81-fork_dup_rand_key.t in IPC::Shareable.

use warnings;
use strict;

use Async::Event::Interval;
use Data::Dumper;
use IPC::Shareable;
use Test::More;

BEGIN {
    if (! $ENV{CI_TESTING}) {
        plan skip_all => "Not on a legit CI platform...";
    }
}

warn "Segs Before: " . IPC::Shareable::ipcs() . "\n" if $ENV{PRINT_SEGS};
my $segs_before = IPC::Shareable::ipcs();

{
    tie my %shared_data, 'IPC::Shareable', {
        key     => 'fork rand dup keys',
        create  => 1,
        destroy => 1
    };

    $shared_data{$$}{called}++;

    my $event_one = Async::Event::Interval->new(0, sub {$shared_data{$$}{called}++});
    my $event_two = Async::Event::Interval->new(0, sub {$shared_data{$$}{called}++});

    $event_one->start;
    $event_two->start;

    sleep 1;

    $event_one->stop;
    $event_two->stop;

    my $one_pid = $event_one->pid;
    my $two_pid = $event_two->pid;

    is exists $shared_data{$one_pid}{called}, 1, "Event one got a rand shm key ok";
    is exists $shared_data{$two_pid}{called}, 1, "Adding srand() ensures _shm_key_rand() gives out rand key in fork()";

    (tied %shared_data)->clean_up_all;
}

IPC::Shareable::_end;
Async::Event::Interval::_end;

warn "Segs After: " . IPC::Shareable::ipcs() . "\n" if $ENV{PRINT_SEGS};
my $segs_after = IPC::Shareable::ipcs();

is $segs_after, $segs_before, "2nd event segs were cleaned up ok";

done_testing();

Output:

Segs Before:       81
ok 1 - Event one got a rand shm key ok
ok 2 - Adding srand() ensures _shm_key_rand() gives out rand key in fork()
Segs After:       82
not ok 3 - 2nd event segs were cleaned up ok
#   Failed test '2nd event segs were cleaned up ok'
#   at t/81-fork_dup_rand_keys.t line 53.
#          got: '      82'
#     expected: '      81'
1..3
# Looks like you failed 1 test of 3.

Failed test 'done_testing() was already called at t/45-params.t

On some of my smoker systems the test suite fails like this:

#   Failed test 'done_testing() was already called at t/45-params.t line 26'
#   at t/45-params.t line 26.
t/45-params.t .......... 
All 3 subtests passed 

t/15-interval.t may fail

See subject. A list of failed CPAN Testers test reports with this problem may be found here:
http://analysis.cpantesters.org/reports_by_field?SUBMIT_xxx=Submit&distv=Async-Event-Interval-1.00&field=fail%3At%2F15-interval.t&field=meta%3Aosname%2Bperl

interval loop in _pm() needs a 'continue' hook to break the loop so it can finish()

"timeout" option

Add another param in new() that sets a timeout. If the event hasn't done something in X time, end itself

Many test failures

The last Async-Event-Interval releases show many test failures, see http://matrix.cpantesters.org/?dist=Async-Event-Interval+1.08
Looking at the individual test reports I don't see any patterns, also the failures look quite different.

Add new() param that allows executing callback immediately or after the first interval