ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/AnyEvent-MP/MP.pm
(Generate patch)

Comparing AnyEvent-MP/MP.pm (file contents):
Revision 1.34 by root, Wed Aug 5 23:50:46 2009 UTC vs.
Revision 1.44 by root, Wed Aug 12 21:39:58 2009 UTC

8 8
9 $NODE # contains this node's noderef 9 $NODE # contains this node's noderef
10 NODE # returns this node's noderef 10 NODE # returns this node's noderef
11 NODE $port # returns the noderef of the port 11 NODE $port # returns the noderef of the port
12 12
13 $SELF # receiving/own port id in rcv callbacks
14
15 # ports are message endpoints
16
17 # sending messages
13 snd $port, type => data...; 18 snd $port, type => data...;
19 snd $port, @msg;
20 snd @msg_with_first_element_being_a_port;
14 21
15 $SELF # receiving/own port id in rcv callbacks 22 # miniports
23 my $miniport = port { my @msg = @_; 0 };
16 24
25 # full ports
26 my $port = port;
17 rcv $port, smartmatch => $cb->($port, @msg); 27 rcv $port, smartmatch => $cb->(@msg);
18
19 # examples:
20 rcv $port2, ping => sub { snd $_[0], "pong"; 0 }; 28 rcv $port, ping => sub { snd $_[0], "pong"; 0 };
21 rcv $port1, pong => sub { warn "pong received\n" }; 29 rcv $port, pong => sub { warn "pong received\n"; 0 };
22 snd $port2, ping => $port1; 30
31 # remote ports
32 my $port = spawn $node, $initfunc, @initdata;
23 33
24 # more, smarter, matches (_any_ is exported by this module) 34 # more, smarter, matches (_any_ is exported by this module)
25 rcv $port, [child_died => $pid] => sub { ... 35 rcv $port, [child_died => $pid] => sub { ...
26 rcv $port, [_any_, _any_, 3] => sub { .. $_[2] is 3 36 rcv $port, [_any_, _any_, 3] => sub { .. $_[2] is 3
37
38 # monitoring
39 mon $port, $cb->(@msg) # callback is invoked on death
40 mon $port, $otherport # kill otherport on abnormal death
41 mon $port, $otherport, @msg # send message on death
27 42
28=head1 DESCRIPTION 43=head1 DESCRIPTION
29 44
30This module (-family) implements a simple message passing framework. 45This module (-family) implements a simple message passing framework.
31 46
90 105
91=cut 106=cut
92 107
93package AnyEvent::MP; 108package AnyEvent::MP;
94 109
95use AnyEvent::MP::Base; 110use AnyEvent::MP::Kernel;
96 111
97use common::sense; 112use common::sense;
98 113
99use Carp (); 114use Carp ();
100 115
101use AE (); 116use AE ();
102 117
103use base "Exporter"; 118use base "Exporter";
104 119
105our $VERSION = '0.1'; 120our $VERSION = $AnyEvent::MP::Kernel::VERSION;
121
106our @EXPORT = qw( 122our @EXPORT = qw(
107 NODE $NODE *SELF node_of _any_ 123 NODE $NODE *SELF node_of _any_
108 resolve_node initialise_node 124 resolve_node initialise_node
109 snd rcv mon kil reg psub 125 snd rcv mon kil reg psub spawn
110 port 126 port
111); 127);
112 128
113our $SELF; 129our $SELF;
114 130
297 $port 313 $port
298} 314}
299 315
300=item reg $port, $name 316=item reg $port, $name
301 317
302Registers the given port under the name C<$name>. If the name already 318=item reg $name
303exists it is replaced. 319
320Registers the given port (or C<$SELF><<< if missing) under the name
321C<$name>. If the name already exists it is replaced.
304 322
305A port can only be registered under one well known name. 323A port can only be registered under one well known name.
306 324
307A port automatically becomes unregistered when it is killed. 325A port automatically becomes unregistered when it is killed.
308 326
309=cut 327=cut
310 328
311sub reg(@) { 329sub reg(@) {
312 my ($port, $name) = @_; 330 my $port = @_ > 1 ? shift : $SELF || Carp::croak 'reg: called with one argument only, but $SELF not set,';
313 331
314 $REG{$name} = $port; 332 $REG{$_[0]} = $port;
315} 333}
316 334
317=item rcv $port, $callback->(@msg) 335=item rcv $port, $callback->(@msg)
318 336
319Replaces the callback on the specified miniport (after converting it to 337Replaces the callback on the specified miniport (after converting it to
324=item rcv $port, $smartmatch => $callback->(@msg), ... 342=item rcv $port, $smartmatch => $callback->(@msg), ...
325 343
326=item rcv $port, [$smartmatch...] => $callback->(@msg), ... 344=item rcv $port, [$smartmatch...] => $callback->(@msg), ...
327 345
328Register callbacks to be called on matching messages on the given full 346Register callbacks to be called on matching messages on the given full
329port (after converting it to one if required). 347port (after converting it to one if required) and return the port.
330 348
331The callback has to return a true value when its work is done, after 349The callback has to return a true value when its work is done, after
332which is will be removed, or a false value in which case it will stay 350which is will be removed, or a false value in which case it will stay
333registered. 351registered.
334 352
335The global C<$SELF> (exported by this module) contains C<$port> while 353The global C<$SELF> (exported by this module) contains C<$port> while
336executing the callback. 354executing the callback.
337 355
338Runtime errors wdurign callback execution will result in the port being 356Runtime errors during callback execution will result in the port being
339C<kil>ed. 357C<kil>ed.
340 358
341If the match is an array reference, then it will be matched against the 359If the match is an array reference, then it will be matched against the
342first elements of the message, otherwise only the first element is being 360first elements of the message, otherwise only the first element is being
343matched. 361matched.
346exported by this module) matches any single element of the message. 364exported by this module) matches any single element of the message.
347 365
348While not required, it is highly recommended that the first matching 366While not required, it is highly recommended that the first matching
349element is a string identifying the message. The one-string-only match is 367element is a string identifying the message. The one-string-only match is
350also the most efficient match (by far). 368also the most efficient match (by far).
369
370Example: create a port and bind receivers on it in one go.
371
372 my $port = rcv port,
373 msg1 => sub { ...; 0 },
374 msg2 => sub { ...; 0 },
375 ;
376
377Example: create a port, bind receivers and send it in a message elsewhere
378in one go:
379
380 snd $otherport, reply =>
381 rcv port,
382 msg1 => sub { ...; 0 },
383 ...
384 ;
351 385
352=cut 386=cut
353 387
354sub rcv($@) { 388sub rcv($@) {
355 my $port = shift; 389 my $port = shift;
462 } 496 }
463} 497}
464 498
465=item $guard = mon $port, $cb->(@reason) 499=item $guard = mon $port, $cb->(@reason)
466 500
467=item $guard = mon $port, $otherport 501=item $guard = mon $port, $rcvport
468 502
503=item $guard = mon $port
504
469=item $guard = mon $port, $otherport, @msg 505=item $guard = mon $port, $rcvport, @msg
470 506
471Monitor the given port and do something when the port is killed. 507Monitor the given port and do something when the port is killed or
508messages to it were lost, and optionally return a guard that can be used
509to stop monitoring again.
472 510
511C<mon> effectively guarantees that, in the absence of hardware failures,
512that after starting the monitor, either all messages sent to the port
513will arrive, or the monitoring action will be invoked after possible
514message loss has been detected. No messages will be lost "in between"
515(after the first lost message no further messages will be received by the
516port). After the monitoring action was invoked, further messages might get
517delivered again.
518
473In the first form, the callback is simply called with any number 519In the first form (callback), the callback is simply called with any
474of C<@reason> elements (no @reason means that the port was deleted 520number of C<@reason> elements (no @reason means that the port was deleted
475"normally"). Note also that I<< the callback B<must> never die >>, so use 521"normally"). Note also that I<< the callback B<must> never die >>, so use
476C<eval> if unsure. 522C<eval> if unsure.
477 523
478In the second form, the other port will be C<kil>'ed with C<@reason>, iff 524In the second form (another port given), the other port (C<$rcvport>)
479a @reason was specified, i.e. on "normal" kils nothing happens, while 525will be C<kil>'ed with C<@reason>, iff a @reason was specified, i.e. on
480under all other conditions, the other port is killed with the same reason. 526"normal" kils nothing happens, while under all other conditions, the other
527port is killed with the same reason.
481 528
529The third form (kill self) is the same as the second form, except that
530C<$rvport> defaults to C<$SELF>.
531
482In the last form, a message of the form C<@msg, @reason> will be C<snd>. 532In the last form (message), a message of the form C<@msg, @reason> will be
533C<snd>.
534
535As a rule of thumb, monitoring requests should always monitor a port from
536a local port (or callback). The reason is that kill messages might get
537lost, just like any other message. Another less obvious reason is that
538even monitoring requests can get lost (for exmaple, when the connection
539to the other node goes down permanently). When monitoring a port locally
540these problems do not exist.
483 541
484Example: call a given callback when C<$port> is killed. 542Example: call a given callback when C<$port> is killed.
485 543
486 mon $port, sub { warn "port died because of <@_>\n" }; 544 mon $port, sub { warn "port died because of <@_>\n" };
487 545
488Example: kill ourselves when C<$port> is killed abnormally. 546Example: kill ourselves when C<$port> is killed abnormally.
489 547
490 mon $port, $self; 548 mon $port;
491 549
492Example: send us a restart message another C<$port> is killed. 550Example: send us a restart message when another C<$port> is killed.
493 551
494 mon $port, $self => "restart"; 552 mon $port, $self => "restart";
495 553
496=cut 554=cut
497 555
498sub mon { 556sub mon {
499 my ($noderef, $port) = split /#/, shift, 2; 557 my ($noderef, $port) = split /#/, shift, 2;
500 558
501 my $node = $NODE{$noderef} || add_node $noderef; 559 my $node = $NODE{$noderef} || add_node $noderef;
502 560
503 my $cb = shift; 561 my $cb = @_ ? shift : $SELF || Carp::croak 'mon: called with one argument only, but $SELF not set,';
504 562
505 unless (ref $cb) { 563 unless (ref $cb) {
506 if (@_) { 564 if (@_) {
507 # send a kill info message 565 # send a kill info message
508 my (@msg) = ($cb, @_); 566 my (@msg) = ($cb, @_);
539=cut 597=cut
540 598
541sub mon_guard { 599sub mon_guard {
542 my ($port, @refs) = @_; 600 my ($port, @refs) = @_;
543 601
602 #TODO: mon-less form?
603
544 mon $port, sub { 0 && @refs } 604 mon $port, sub { 0 && @refs }
545} 605}
546 606
547=item lnk $port1, $port2
548
549Link two ports. This is simply a shorthand for:
550
551 mon $port1, $port2;
552 mon $port2, $port1;
553
554It means that if either one is killed abnormally, the other one gets
555killed as well.
556
557=item kil $port[, @reason] 607=item kil $port[, @reason]
558 608
559Kill the specified port with the given C<@reason>. 609Kill the specified port with the given C<@reason>.
560 610
561If no C<@reason> is specified, then the port is killed "normally" (linked 611If no C<@reason> is specified, then the port is killed "normally" (linked
567Runtime errors while evaluating C<rcv> callbacks or inside C<psub> blocks 617Runtime errors while evaluating C<rcv> callbacks or inside C<psub> blocks
568will be reported as reason C<< die => $@ >>. 618will be reported as reason C<< die => $@ >>.
569 619
570Transport/communication errors are reported as C<< transport_error => 620Transport/communication errors are reported as C<< transport_error =>
571$message >>. 621$message >>.
622
623=cut
624
625=item $port = spawn $node, $initfunc[, @initdata]
626
627Creates a port on the node C<$node> (which can also be a port ID, in which
628case it's the node where that port resides).
629
630The port ID of the newly created port is return immediately, and it is
631permissible to immediately start sending messages or monitor the port.
632
633After the port has been created, the init function is
634called. This function must be a fully-qualified function name
635(e.g. C<MyApp::Chat::Server::init>). To specify a function in the main
636program, use C<::name>.
637
638If the function doesn't exist, then the node tries to C<require>
639the package, then the package above the package and so on (e.g.
640C<MyApp::Chat::Server>, C<MyApp::Chat>, C<MyApp>) until the function
641exists or it runs out of package names.
642
643The init function is then called with the newly-created port as context
644object (C<$SELF>) and the C<@initdata> values as arguments.
645
646A common idiom is to pass your own port, monitor the spawned port, and
647in the init function, monitor the original port. This two-way monitoring
648ensures that both ports get cleaned up when there is a problem.
649
650Example: spawn a chat server port on C<$othernode>.
651
652 # this node, executed from within a port context:
653 my $server = spawn $othernode, "MyApp::Chat::Server::connect", $SELF;
654 mon $server;
655
656 # init function on C<$othernode>
657 sub connect {
658 my ($srcport) = @_;
659
660 mon $srcport;
661
662 rcv $SELF, sub {
663 ...
664 };
665 }
666
667=cut
668
669sub _spawn {
670 my $port = shift;
671 my $init = shift;
672
673 local $SELF = "$NODE#$port";
674 eval {
675 &{ load_func $init }
676 };
677 _self_die if $@;
678}
679
680sub spawn(@) {
681 my ($noderef, undef) = split /#/, shift, 2;
682
683 my $id = "$RUNIQ." . $ID++;
684
685 $_[0] =~ /::/
686 or Carp::croak "spawn init function must be a fully-qualified name, caught";
687
688 ($NODE{$noderef} || add_node $noderef)
689 ->send (["", "AnyEvent::MP::_spawn" => $id, @_]);
690
691 "$noderef#$id"
692}
572 693
573=back 694=back
574 695
575=head1 NODE MESSAGES 696=head1 NODE MESSAGES
576 697
618 739
619=back 740=back
620 741
621=head1 AnyEvent::MP vs. Distributed Erlang 742=head1 AnyEvent::MP vs. Distributed Erlang
622 743
623AnyEvent::MP got lots of its ideas from distributed erlang (erlang node 744AnyEvent::MP got lots of its ideas from distributed Erlang (Erlang node
624== aemp node, erlang process == aemp port), so many of the documents and 745== aemp node, Erlang process == aemp port), so many of the documents and
625programming techniques employed by erlang apply to AnyEvent::MP. Here is a 746programming techniques employed by Erlang apply to AnyEvent::MP. Here is a
626sample: 747sample:
627 748
628 http://www.erlang.se/doc/programming_rules.shtml 749 http://www.Erlang.se/doc/programming_rules.shtml
629 http://erlang.org/doc/getting_started/part_frame.html # chapters 3 and 4 750 http://Erlang.org/doc/getting_started/part_frame.html # chapters 3 and 4
630 http://erlang.org/download/erlang-book-part1.pdf # chapters 5 and 6 751 http://Erlang.org/download/Erlang-book-part1.pdf # chapters 5 and 6
631 http://erlang.org/download/armstrong_thesis_2003.pdf # chapters 4 and 5 752 http://Erlang.org/download/armstrong_thesis_2003.pdf # chapters 4 and 5
632 753
633Despite the similarities, there are also some important differences: 754Despite the similarities, there are also some important differences:
634 755
635=over 4 756=over 4
636 757
647 768
648Erlang uses processes that selctively receive messages, and therefore 769Erlang uses processes that selctively receive messages, and therefore
649needs a queue. AEMP is event based, queuing messages would serve no useful 770needs a queue. AEMP is event based, queuing messages would serve no useful
650purpose. 771purpose.
651 772
652(But see L<Coro::MP> for a more erlang-like process model on top of AEMP). 773(But see L<Coro::MP> for a more Erlang-like process model on top of AEMP).
653 774
654=item * Erlang sends are synchronous, AEMP sends are asynchronous. 775=item * Erlang sends are synchronous, AEMP sends are asynchronous.
655 776
656Sending messages in erlang is synchronous and blocks the process. AEMP 777Sending messages in Erlang is synchronous and blocks the process. AEMP
657sends are immediate, connection establishment is handled in the 778sends are immediate, connection establishment is handled in the
658background. 779background.
659 780
660=item * Erlang can silently lose messages, AEMP cannot. 781=item * Erlang can silently lose messages, AEMP cannot.
661 782
664and c, and the other side only receives messages a and c). 785and c, and the other side only receives messages a and c).
665 786
666AEMP guarantees correct ordering, and the guarantee that there are no 787AEMP guarantees correct ordering, and the guarantee that there are no
667holes in the message sequence. 788holes in the message sequence.
668 789
669=item * In erlang, processes can be declared dead and later be found to be 790=item * In Erlang, processes can be declared dead and later be found to be
670alive. 791alive.
671 792
672In erlang it can happen that a monitored process is declared dead and 793In Erlang it can happen that a monitored process is declared dead and
673linked processes get killed, but later it turns out that the process is 794linked processes get killed, but later it turns out that the process is
674still alive - and can receive messages. 795still alive - and can receive messages.
675 796
676In AEMP, when port monitoring detects a port as dead, then that port will 797In AEMP, when port monitoring detects a port as dead, then that port will
677eventually be killed - it cannot happen that a node detects a port as dead 798eventually be killed - it cannot happen that a node detects a port as dead
678and then later sends messages to it, finding it is still alive. 799and then later sends messages to it, finding it is still alive.
679 800
680=item * Erlang can send messages to the wrong port, AEMP does not. 801=item * Erlang can send messages to the wrong port, AEMP does not.
681 802
682In erlang it is quite possible that a node that restarts reuses a process 803In Erlang it is quite possible that a node that restarts reuses a process
683ID known to other nodes for a completely different process, causing 804ID known to other nodes for a completely different process, causing
684messages destined for that process to end up in an unrelated process. 805messages destined for that process to end up in an unrelated process.
685 806
686AEMP never reuses port IDs, so old messages or old port IDs floating 807AEMP never reuses port IDs, so old messages or old port IDs floating
687around in the network will not be sent to an unrelated port. 808around in the network will not be sent to an unrelated port.
693securely authenticate nodes. 814securely authenticate nodes.
694 815
695=item * The AEMP protocol is optimised for both text-based and binary 816=item * The AEMP protocol is optimised for both text-based and binary
696communications. 817communications.
697 818
698The AEMP protocol, unlike the erlang protocol, supports both 819The AEMP protocol, unlike the Erlang protocol, supports both
699language-independent text-only protocols (good for debugging) and binary, 820language-independent text-only protocols (good for debugging) and binary,
700language-specific serialisers (e.g. Storable). 821language-specific serialisers (e.g. Storable).
701 822
702It has also been carefully designed to be implementable in other languages 823It has also been carefully designed to be implementable in other languages
703with a minimum of work while gracefully degrading fucntionality to make the 824with a minimum of work while gracefully degrading fucntionality to make the
704protocol simple. 825protocol simple.
705 826
827=item * AEMP has more flexible monitoring options than Erlang.
828
829In Erlang, you can chose to receive I<all> exit signals as messages
830or I<none>, there is no in-between, so monitoring single processes is
831difficult to implement. Monitoring in AEMP is more flexible than in
832Erlang, as one can choose between automatic kill, exit message or callback
833on a per-process basis.
834
835=item * Erlang tries to hide remote/local connections, AEMP does not.
836
837Monitoring in Erlang is not an indicator of process death/crashes,
838as linking is (except linking is unreliable in Erlang).
839
840In AEMP, you don't "look up" registered port names or send to named ports
841that might or might not be persistent. Instead, you normally spawn a port
842on the remote node. The init function monitors the you, and you monitor
843the remote port. Since both monitors are local to the node, they are much
844more reliable.
845
846This also saves round-trips and avoids sending messages to the wrong port
847(hard to do in Erlang).
848
706=back 849=back
707 850
708=head1 SEE ALSO 851=head1 SEE ALSO
709 852
710L<AnyEvent>. 853L<AnyEvent>.

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines