The Perl Toolchain Summit needs more sponsors. If your company depends on Perl, please support this very important event.

NAME

IO::Async::Loop::AnyEvent - use IO::Async with AnyEvent

SYNOPSIS

   use IO::Async::Loop::AnyEvent;

   my $loop = IO::Async::Loop::AnyEvent->new();

   $loop->add( ... );

   $loop->add( IO::Async::Signal->new(
         name => 'HUP',
         on_receipt => sub { ... },
   ) );

   $loop->run;

DESCRIPTION

This subclass of IO::Async::Loop uses AnyEvent to perform its work.

CONSTRUCTOR

new

   $loop = IO::Async::Loop::AnyEvent->new

This function returns a new instance of a IO::Async::Loop::AnyEvent object.

BUGS

  • watch_idle and unwatch_idle don't work properly against AnyEvent::Impl::IOAsync. At least, the unit tests fail, and some scheduled CODErefs never get executed, and sit in the internal queue of the inner-nested IO::Async::Loop that AnyEvent::Impl::IOAsync itself constructed. An easy workaround here is simply to pick another AnyEvent model, by using the PERL_ANYEVENT_MODEL environment variable.

    That all said, I am honestly surprised this is the only thing that breaks, when IO::Async is nested upon AnyEvent itself running atop another IO::Async.

  • The implementation of the loop_once method requires the use of an undocumented AnyEvent method (one_event before version 6, _poll thereafter). This happens to work at the time of writing, but as it is undocumented it may be subject to change.

    The loop_forever method does not rely on this undocumented method, so should be safe from upstream changes. Furthremore, if AnyEvent rather than IO::Async remains ultimately in control of the runtime, by waiting on condvars, this should not be problematic.

AUTHOR

Paul Evans <leonerd@leonerd.org.uk>