=pod =head1 NAME Mutex - Various locking implementations supporting processes and threads =head1 VERSION This document describes Mutex version 1.006 =head1 SYNOPSIS use Mutex; my $mutex = Mutex->new; { use MCE::Flow max_workers => 4; mce_flow sub { $mutex->lock; # access shared resource my $wid = MCE->wid; MCE->say($wid); sleep 1; $mutex->unlock; }; } { use MCE::Hobo; MCE::Hobo->create('work', $_) for 1..4; MCE::Hobo->waitall; } { use threads; threads->create('work', $_) for 5..8; $_->join for ( threads->list ); } sub work { my ($id) = @_; $mutex->lock; # access shared resource print $id, "\n"; sleep 1; $mutex->unlock; } =head1 DESCRIPTION This module, a standalone version of L, implements locking methods that can be used to coordinate access to shared data from multiple workers spawned as processes or threads. The inspiration for this module came from reading Mutex for Ruby. =head1 API DOCUMENTATION =head2 Mutex->new ( ) =head2 Mutex->new ( impl => "Channel" ) =head2 Mutex->new ( impl => "Flock", [ path => "/tmp/file.lock" ] ) =head2 Mutex->new ( path => "/tmp/file.lock" ) Creates a new mutex. Channel locking (the default), unless C is given, is through a pipe or socket depending on the platform. The advantage of channel locking is not having to re-establish handles inside new processes and threads. For Fcntl-based locking, it is the responsibility of the caller to remove the C, associated with the mutex, when path is given. Otherwise, it establishes a C internally including removal on scope exit. =head2 $mutex->impl ( void ) Returns the implementation used for the mutex. $m1 = Mutex->new( ); $m1->impl(); # Channel $m2 = Mutex->new( path => /tmp/my.lock ); $m2->impl(); # Flock $m3 = Mutex->new( impl => "Channel" ); $m3->impl(); # Channel $m4 = Mutex->new( impl => "Flock" ); $m4->impl(); # Flock =head2 $mutex->lock ( void ) =head2 $mutex->lock_exclusive ( void ) Attempts to grab an exclusive lock and waits if not available. Multiple calls to mutex->lock by the same process or thread is safe. The mutex will remain locked until mutex->unlock is called. The method C is an alias for C. ( my $mutex = Mutex->new( path => $0 ) )->lock_exclusive; =head2 $mutex->lock_shared ( void ) Like C, but attempts to grab a shared lock instead. For non-Fcntl implementations, C is an alias for C. =head2 $mutex->unlock ( void ) Releases the lock. A held lock by an exiting process or thread is released automatically. =head2 $mutex->synchronize ( sub { ... }, @_ ) =head2 $mutex->enter ( sub { ... }, @_ ) Obtains a lock, runs the code block, and releases the lock after the block completes. Optionally, the method is C aware. my $val = $mutex->synchronize( sub { # access shared resource return 'scalar'; }); my @ret = $mutex->enter( sub { # access shared resource return @list; }); The method C is an alias for C. =head2 $mutex->timedwait ( timeout ) Blocks until obtaining an exclusive lock. A false value is returned if the timeout is reached, and a true value otherwise. The default is 1 second when omitting timeout. my $mutex = Mutex->new( path => $0 ); # terminate script if a previous instance is still running exit unless $mutex->timedwait( 2 ); ... =head1 REQUIREMENTS Perl 5.8.1 or later. =head1 AUTHOR Mario E. Roy, Smarioeroy AT gmail DOT comE> =head1 COPYRIGHT AND LICENSE Copyright (C) 2017-2020 by Mario E. Roy Mutex is released under the same license as Perl. See L for more information. =cut