stevieb has asked for the wisdom of the Perl Monks concerning the following question:

Update: I've solved the issue. Everything defined/created in the app that is NOT referenced within at least one route goes out of scope when the app is started. If something is referenced within a route even just once, it lives. Even my forking code works now :)

I'm still working on my Dancer2 application, but now I'm at the point where I need it to periodically to something, even without client interaction. Essentially, I want the Dancer2 app to behave like a normal always-running background script while a client browser is not attached.

So, what I need to do is have this app periodically do something without tying up the main application. I wrote a *very* basic async timed event module Async::Event::Interval (yeah, for fun, mostly). It works, and does what I need it to do (the event is a forked process). Here's an example:

use warnings;
use strict;

use Async::Event::Interval;

my $event = Async::Event::Interval->new(
    1, # interval, in seconds
    sub {
        print "event...\n";
    },
);
$event->start;

sleep 1; # wait for first event to fire
print "main is going to sleep...\n";
sleep 5;
print "done\n";

[download]

Output:

event...
main is going to sleep...
event...
event...
event...
event...
event...
done

[download]

I fetch the PID of the event process, and it shows up just dandy as running in the proc list:

$ ps ax | grep 22834
22834 pts/2    S+     0:00 perl t/evt.pl

[download]

However, no matter where it is I create and start the event in a Dancer2 app, it fires after the first interval is expired, but then in the proc list I see:

$ps ax | grep 22732
22732 pts/2    Z+     0:00 [/home/spek/repo] <defunct>

[download]

Can anyone explain what Dancer2 is doing that forces a child process to become defunct, or how I can get around/avoid this problem without having to have a second background app running to do the side-work that is needed? If anyone thinks a different async type software will work, I'll ditch my own, as it was just for playing around anyways.

As a side note, I have no idea why it's showing /home/spek/repo in the proc entry, as my current working dir is ~/repo/app-envui when I start the webapp.

Replies are listed 'Best First'.
Re: Async timed events with Dancer2
by Anonymous Monk on Sep 25, 2016 at 02:17 UTC
[reply]
      Thanks. I'll give that a try. If that doesn't work, I'll need to get a better understanding of how Dancer2 works under the hood.
[reply]
Re: Async timed events with Dancer2
by stevieb (Canon) on Sep 25, 2016 at 14:32 UTC

    OK, so an update here. I have removed the fork() stuff. The problem is that when I create an object in my webapp, it goes out of scope immediately as the webapp is presented with plackup. Here's a much trimmed down repro example:

    # Scope.pm in webapp root dir

package Scope;

sub new {
    return bless {}, shift
}

sub DESTROY {
    print "going out of scope...\n";
}

# webapp

package MyApp;

use lib '.';
use Scope;
use Dancer2;
use Dancer2::Plugin::Database;

our $VERSION = '0.1';

my $obj = Scope->new;

    [download]

    Now run it:

    plackup -R lib,bin bin/app.pl

    [download]

    Here's what happens when I run it:

    
Watching lib bin bin/lib bin/app.pl for file updates.
going out of scope
HTTP::Server::PSGI: Accepting connections at http://0:5000/

    [download]

    I've also asked for guidance on the #dancer IRC channel. I'm hoping someone with experience with Dancer2 (or perhaps web frameworks in general) can help me get a grasp on getting around this issue (as I pour through resources I'm finding online elsewhere).

[reply]
[d/l]
[select]

Back to Seekers of Perl Wisdom