[ViewVC] Diff of: cvs/AnyEvent-MP/MP.pm

Comparing AnyEvent-MP/MP.pm (file contents):
Revision 1.2 by root, Fri Jul 31 20:55:46 2009 UTC vs.
Revision 1.121 by root, Tue Feb 28 18:37:24 2012 UTC

1	=head1 NAME	1	=head1 NAME
2		2
3	AnyEvent::MP - multi-processing/message-passing framework	3	AnyEvent::MP - erlang-style multi-processing/message-passing framework
4		4
5	=head1 SYNOPSIS	5	=head1 SYNOPSIS
6		6
7	use AnyEvent::MP;	7	use AnyEvent::MP;
8		8
9	NODE # returns this node identifier
10	$NODE # contains this node identifier	9	$NODE # contains this node's node ID
		10	NODE # returns this node's node ID
11		11
		12	$SELF # receiving/own port id in rcv callbacks
		13
		14	# initialise the node so it can send/receive messages
		15	configure;
		16
		17	# ports are message destinations
		18
		19	# sending messages
12	snd $port, type => data...;	20	snd $port, type => data...;
		21	snd $port, @msg;
		22	snd @msg_with_first_element_being_a_port;
13		23
14	rcv $port, smartmatch => $cb->($port, @msg);	24	# creating/using ports, the simple way
		25	my $simple_port = port { my @msg = @_ };
15		26
16	# examples:	27	# creating/using ports, tagged message matching
		28	my $port = port;
17	rcv $port2, ping => sub { snd $_[0], "pong"; 0 };	29	rcv $port, ping => sub { snd $_[0], "pong" };
18	rcv $port1, pong => sub { warn "pong received\n" };	30	rcv $port, pong => sub { warn "pong received\n" };
19	snd $port2, ping => $port1;
20		31
21	# more, smarter, matches (_any_ is exported by this module)	32	# create a port on another node
22	rcv $port, [child_died => $pid] => sub { ...	33	my $port = spawn $node, $initfunc, @initdata;
23	rcv $port, [_any_, _any_, 3] => sub { .. $_[2] is 3	34
		35	# destroy a port again
		36	kil $port; # "normal" kill
		37	kil $port, my_error => "everything is broken"; # error kill
		38
		39	# monitoring
		40	mon $localport, $cb->(@msg) # callback is invoked on death
		41	mon $localport, $otherport # kill otherport on abnormal death
		42	mon $localport, $otherport, @msg # send message on death
		43
		44	# temporarily execute code in port context
		45	peval $port, sub { die "kill the port!" };
		46
		47	# execute callbacks in $SELF port context
		48	my $timer = AE::timer 1, 0, psub {
		49	die "kill the port, delayed";
		50	};
		51
		52	=head1 CURRENT STATUS
		53
		54	bin/aemp - stable.
		55	AnyEvent::MP - stable API, should work.
		56	AnyEvent::MP::Intro - explains most concepts.
		57	AnyEvent::MP::Kernel - mostly stable API.
		58	AnyEvent::MP::Global - stable API.
24		59
25	=head1 DESCRIPTION	60	=head1 DESCRIPTION
26		61
27	This module (-family) implements a simple message passing framework.	62	This module (-family) implements a simple message passing framework.
28		63
29	Despite its simplicity, you can securely message other processes running	64	Despite its simplicity, you can securely message other processes running
30	on the same or other hosts.	65	on the same or other hosts, and you can supervise entities remotely.
		66
		67	For an introduction to this module family, see the L<AnyEvent::MP::Intro>
		68	manual page and the examples under F<eg/>.
31		69
32	=head1 CONCEPTS	70	=head1 CONCEPTS
33		71
34	=over 4	72	=over 4
35		73
36	=item port	74	=item port
37		75
38	A port is something you can send messages to with the C<snd> function, and	76	Not to be confused with a TCP port, a "port" is something you can send
39	you can register C<rcv> handlers with. All C<rcv> handlers will receive	77	messages to (with the C<snd> function).
40	messages they match, messages will not be queued.
41		78
42	=item port id - C<pid@host#portname>	79	Ports allow you to register C<rcv> handlers that can match all or just
		80	some messages. Messages send to ports will not be queued, regardless of
		81	anything was listening for them or not.
43		82
44	A port id is always the node id, a hash-mark (C<#>) as separator, followed	83	Ports are represented by (printable) strings called "port IDs".
45	by a port name.
46		84
47	A port name can be a well known port (basically an identifier/bareword),	85	=item port ID - C<nodeid#portname>
48	or a generated name, consisting of node id, a dot (C<.>), and an	86
49	identifier.	87	A port ID is the concatenation of a node ID, a hash-mark (C<#>) as
		88	separator, and a port name (a printable string of unspecified format).
50		89
51	=item node	90	=item node
52		91
53	A node is a single process containing at least one port - the node	92	A node is a single process containing at least one port - the node port,
54	port. You can send messages to node ports to let them create new ports,	93	which enables nodes to manage each other remotely, and to create new
55	among other things.	94	ports.
56		95
57	Initially, nodes are either private (single-process only) or hidden	96	Nodes are either public (have one or more listening ports) or private
58	(connected to a father node only). Only when they epxlicitly "go public"	97	(no listening ports). Private nodes cannot talk to other private nodes
59	can you send them messages form unrelated other nodes.	98	currently, but all nodes can talk to public nodes.
60		99
61	Public nodes automatically connect to all other public nodes in a network	100	Nodes is represented by (printable) strings called "node IDs".
62	when they connect, creating a full mesh.
63		101
64	=item node id - C<host:port>, C<id@host>, C<id>	102	=item node ID - C<[A-Za-z0-9_\-.:]*>
65		103
66	A node ID is a string that either uniquely identifies a given node (For	104	A node ID is a string that uniquely identifies the node within a
67	private and hidden nodes), or contains a recipe on how to reach a given	105	network. Depending on the configuration used, node IDs can look like a
68	node (for public nodes).	106	hostname, a hostname and a port, or a random string. AnyEvent::MP itself
		107	doesn't interpret node IDs in any way except to uniquely identify a node.
		108
		109	=item binds - C<ip:port>
		110
		111	Nodes can only talk to each other by creating some kind of connection to
		112	each other. To do this, nodes should listen on one or more local transport
		113	endpoints - binds.
		114
		115	Currently, only standard C<ip:port> specifications can be used, which
		116	specify TCP ports to listen on. So a bind is basically just a tcp socket
		117	in listening mode thta accepts conenctions form other nodes.
		118
		119	=item seed nodes
		120
		121	When a node starts, it knows nothing about the network it is in - it
		122	needs to connect to at least one other node that is already in the
		123	network. These other nodes are called "seed nodes".
		124
		125	Seed nodes themselves are not special - they are seed nodes only because
		126	some other node I<uses> them as such, but any node can be used as seed
		127	node for other nodes, and eahc node cna use a different set of seed nodes.
		128
		129	In addition to discovering the network, seed nodes are also used to
		130	maintain the network - all nodes using the same seed node form are part of
		131	the same network. If a network is split into multiple subnets because e.g.
		132	the network link between the parts goes down, then using the same seed
		133	nodes for all nodes ensures that eventually the subnets get merged again.
		134
		135	Seed nodes are expected to be long-running, and at least one seed node
		136	should always be available. They should also be relatively responsive - a
		137	seed node that blocks for long periods will slow down everybody else.
		138
		139	For small networks, it's best if every node uses the same set of seed
		140	nodes. For large networks, it can be useful to specify "regional" seed
		141	nodes for most nodes in an area, and use all seed nodes as seed nodes for
		142	each other. What's important is that all seed nodes connections form a
		143	complete graph, so that the network cannot split into separate subnets
		144	forever.
		145
		146	Seed nodes are represented by seed IDs.
		147
		148	=item seed IDs - C<host:port>
		149
		150	Seed IDs are transport endpoint(s) (usually a hostname/IP address and a
		151	TCP port) of nodes that should be used as seed nodes.
		152
		153	=item global nodes
		154
		155	An AEMP network needs a discovery service - nodes need to know how to
		156	connect to other nodes they only know by name. In addition, AEMP offers a
		157	distributed "group database", which maps group names to a list of strings
		158	- for example, to register worker ports.
		159
		160	A network needs at least one global node to work, and allows every node to
		161	be a global node.
		162
		163	Any node that loads the L<AnyEvent::MP::Global> module becomes a global
		164	node and tries to keep connections to all other nodes. So while it can
		165	make sense to make every node "global" in small networks, it usually makes
		166	sense to only make seed nodes into global nodes in large networks (nodes
		167	keep connections to seed nodes and global nodes, so makign them the same
		168	reduces overhead).
69		169
70	=back	170	=back
71		171
72	=head1 FUNCTIONS	172	=head1 VARIABLES/FUNCTIONS
73		173
74	=over 4	174	=over 4
75		175
76	=cut	176	=cut
77		177
78	package AnyEvent::MP;	178	package AnyEvent::MP;
79		179
80	use AnyEvent::MP::Util ();	180	use AnyEvent::MP::Config ();
81	use AnyEvent::MP::Node;	181	use AnyEvent::MP::Kernel;
82	use AnyEvent::MP::Transport;	182	use AnyEvent::MP::Kernel qw(%NODE %PORT %PORT_DATA $UNIQ $RUNIQ $ID);
83		183
84	use utf8;
85	use common::sense;	184	use common::sense;
86		185
87	use Carp ();	186	use Carp ();
88		187
89	use AE ();	188	use AE ();
90		189
91	use base "Exporter";	190	use base "Exporter";
92		191
93	our $VERSION = '0.0';	192	our $VERSION = $AnyEvent::MP::Config::VERSION;
94	our @EXPORT = qw(NODE $NODE $PORT snd rcv _any_);
95		193
96	our $DEFAULT_SECRET;	194	our @EXPORT = qw(
97	our $DEFAULT_PORT = "4040";	195	NODE $NODE *SELF node_of after
		196	configure
		197	snd rcv mon mon_guard kil psub peval spawn cal
		198	port
		199	);
98		200
99	our $CONNECT_INTERVAL = 5; # new connect every 5s, at least	201	our $SELF;
100	our $CONNECT_TIMEOUT = 30; # includes handshake
101		202
102	sub default_secret {	203	sub _self_die() {
103	unless (defined $DEFAULT_SECRET) {	204	my $msg = $@;
104	if (open my $fh, "<$ENV{HOME}/.aemp-secret") {	205	$msg =~ s/\n+$// unless ref $msg;
105	sysread $fh, $DEFAULT_SECRET, -s $fh;	206	kil $SELF, die => $msg;
106	} else {	207	}
107	$DEFAULT_SECRET = AnyEvent::MP::Util::nonce 32;	208
		209	=item $thisnode = NODE / $NODE
		210
		211	The C<NODE> function returns, and the C<$NODE> variable contains, the node
		212	ID of the node running in the current process. This value is initialised by
		213	a call to C<configure>.
		214
		215	=item $nodeid = node_of $port
		216
		217	Extracts and returns the node ID from a port ID or a node ID.
		218
		219	=item configure $profile, key => value...
		220
		221	=item configure key => value...
		222
		223	Before a node can talk to other nodes on the network (i.e. enter
		224	"distributed mode") it has to configure itself - the minimum a node needs
		225	to know is its own name, and optionally it should know the addresses of
		226	some other nodes in the network to discover other nodes.
		227
		228	This function configures a node - it must be called exactly once (or
		229	never) before calling other AnyEvent::MP functions.
		230
		231	The key/value pairs are basically the same ones as documented for the
		232	F<aemp> command line utility (sans the set/del prefix), with two additions:
		233
		234	=over 4
		235
		236	=item norc => $boolean (default false)
		237
		238	If true, then the rc file (e.g. F<~/.perl-anyevent-mp>) will I<not>
		239	be consulted - all configuraiton options must be specified in the
		240	C<configure> call.
		241
		242	=item force => $boolean (default false)
		243
		244	IF true, then the values specified in the C<configure> will take
		245	precedence over any values configured via the rc file. The default is for
		246	the rc file to override any options specified in the program.
		247
		248	=back
		249
		250	=over 4
		251
		252	=item step 1, gathering configuration from profiles
		253
		254	The function first looks up a profile in the aemp configuration (see the
		255	L<aemp> commandline utility). The profile name can be specified via the
		256	named C<profile> parameter or can simply be the first parameter). If it is
		257	missing, then the nodename (F<uname -n>) will be used as profile name.
		258
		259	The profile data is then gathered as follows:
		260
		261	First, all remaining key => value pairs (all of which are conveniently
		262	undocumented at the moment) will be interpreted as configuration
		263	data. Then they will be overwritten by any values specified in the global
		264	default configuration (see the F<aemp> utility), then the chain of
		265	profiles chosen by the profile name (and any C<parent> attributes).
		266
		267	That means that the values specified in the profile have highest priority
		268	and the values specified directly via C<configure> have lowest priority,
		269	and can only be used to specify defaults.
		270
		271	If the profile specifies a node ID, then this will become the node ID of
		272	this process. If not, then the profile name will be used as node ID. The
		273	special node ID of C<anon/> will be replaced by a random node ID.
		274
		275	=item step 2, bind listener sockets
		276
		277	The next step is to look up the binds in the profile, followed by binding
		278	aemp protocol listeners on all binds specified (it is possible and valid
		279	to have no binds, meaning that the node cannot be contacted form the
		280	outside. This means the node cannot talk to other nodes that also have no
		281	binds, but it can still talk to all "normal" nodes).
		282
		283	If the profile does not specify a binds list, then a default of C<*> is
		284	used, meaning the node will bind on a dynamically-assigned port on every
		285	local IP address it finds.
		286
		287	=item step 3, connect to seed nodes
		288
		289	As the last step, the seed ID list from the profile is passed to the
		290	L<AnyEvent::MP::Global> module, which will then use it to keep
		291	connectivity with at least one node at any point in time.
		292
		293	=back
		294
		295	Example: become a distributed node using the local node name as profile.
		296	This should be the most common form of invocation for "daemon"-type nodes.
		297
		298	configure
		299
		300	Example: become an anonymous node. This form is often used for commandline
		301	clients.
		302
		303	configure nodeid => "anon/";
		304
		305	Example: configure a node using a profile called seed, which is suitable
		306	for a seed node as it binds on all local addresses on a fixed port (4040,
		307	customary for aemp).
		308
		309	# use the aemp commandline utility
		310	# aemp profile seed nodeid anon/ binds '*:4040'
		311
		312	# then use it
		313	configure profile => "seed";
		314
		315	# or simply use aemp from the shell again:
		316	# aemp run profile seed
		317
		318	# or provide a nicer-to-remember nodeid
		319	# aemp run profile seed nodeid "$(hostname)"
		320
		321	=item $SELF
		322
		323	Contains the current port id while executing C<rcv> callbacks or C<psub>
		324	blocks.
		325
		326	=item *SELF, SELF, %SELF, @SELF...
		327
		328	Due to some quirks in how perl exports variables, it is impossible to
		329	just export C<$SELF>, all the symbols named C<SELF> are exported by this
		330	module, but only C<$SELF> is currently used.
		331
		332	=item snd $port, type => @data
		333
		334	=item snd $port, @msg
		335
		336	Send the given message to the given port, which can identify either a
		337	local or a remote port, and must be a port ID.
		338
		339	While the message can be almost anything, it is highly recommended to
		340	use a string as first element (a port ID, or some word that indicates a
		341	request type etc.) and to consist if only simple perl values (scalars,
		342	arrays, hashes) - if you think you need to pass an object, think again.
		343
		344	The message data logically becomes read-only after a call to this
		345	function: modifying any argument (or values referenced by them) is
		346	forbidden, as there can be considerable time between the call to C<snd>
		347	and the time the message is actually being serialised - in fact, it might
		348	never be copied as within the same process it is simply handed to the
		349	receiving port.
		350
		351	The type of data you can transfer depends on the transport protocol: when
		352	JSON is used, then only strings, numbers and arrays and hashes consisting
		353	of those are allowed (no objects). When Storable is used, then anything
		354	that Storable can serialise and deserialise is allowed, and for the local
		355	node, anything can be passed. Best rely only on the common denominator of
		356	these.
		357
		358	=item $local_port = port
		359
		360	Create a new local port object and returns its port ID. Initially it has
		361	no callbacks set and will throw an error when it receives messages.
		362
		363	=item $local_port = port { my @msg = @_ }
		364
		365	Creates a new local port, and returns its ID. Semantically the same as
		366	creating a port and calling C<rcv $port, $callback> on it.
		367
		368	The block will be called for every message received on the port, with the
		369	global variable C<$SELF> set to the port ID. Runtime errors will cause the
		370	port to be C<kil>ed. The message will be passed as-is, no extra argument
		371	(i.e. no port ID) will be passed to the callback.
		372
		373	If you want to stop/destroy the port, simply C<kil> it:
		374
		375	my $port = port {
		376	my @msg = @_;
108	}	377	...
109	}	378	kil $SELF;
110
111	$DEFAULT_SECRET
112	}
113
114	our $UNIQ = sprintf "%x.%x", $$, time; # per-process/node unique cookie
115	our $PUBLIC = 0;
116	our $NODE;
117	our $PORT;
118
119	our %NODE; # node id to transport mapping, or "undef", for local node
120	our %PORT; # local ports
121	our %LISTENER; # local transports
122
123	sub NODE() { $NODE }
124
125	{
126	use POSIX ();
127	my $nodename = (POSIX::uname)[1];
128	$NODE = "$$\@$nodename";
129	}
130
131	sub _ANY_() { 1 }
132	sub _any_() { \&_ANY_ }
133
134	sub add_node {
135	my ($noderef) = @_;
136
137	return $NODE{$noderef}
138	if exists $NODE{$noderef};
139
140	for (split /,/, $noderef) {
141	return $NODE{$noderef} = $NODE{$_}
142	if exists $NODE{$_};
143	}
144
145	# for indirect sends, use a different class
146	my $node = new AnyEvent::MP::Node::Direct $noderef;
147
148	$NODE{$_} = $node
149	for $noderef, split /,/, $noderef;
150
151	$node
152	}
153
154	sub snd($@) {
155	my ($noderef, $port) = split /#/, shift, 2;
156
157	add_node $noderef
158	unless exists $NODE{$noderef};
159
160	$NODE{$noderef}->send ([$port, [@_]]);
161	}
162
163	sub _inject {
164	my ($port, $msg) = @{+shift};
165
166	$port = $PORT{$port}
167	or return;
168
169	use Data::Dumper;
170	warn Dumper $msg;
171	}
172
173	sub normalise_noderef($) {
174	my ($noderef) = @_;
175
176	my $cv = AE::cv;
177	my @res;
178
179	$cv->begin (sub {
180	my %seen;
181	my @refs;
182	for (sort { $a->[0] <=> $b->[0] } @res) {
183	push @refs, $_->[1] unless $seen{$_->[1]}++
184	}
185	shift->send (join ",", @refs);
186	});	379	};
187		380
188	$noderef = $DEFAULT_PORT unless length $noderef;	381	=cut
189		382
190	my $idx;	383	sub rcv($@);
191	for my $t (split /,/, $noderef) {
192	my $pri = ++$idx;
193
194	#TODO: this should be outside normalise_noderef and in become_public
195	if ($t =~ /^\d*$/) {
196	my $nodename = (POSIX::uname)[1];
197		384
198	$cv->begin;	385	sub _kilme {
199	AnyEvent::Socket::resolve_sockaddr $nodename, $t \|\| "aemp=$DEFAULT_PORT", "tcp", 0, undef, sub {	386	die "received message on port without callback";
200	for (@_) {	387	}
201	my ($service, $host) = AnyEvent::Socket::unpack_sockaddr $_->[3];	388
202	push @res, [	389	sub port(;&) {
203	$pri += 1e-5,	390	my $id = "$UNIQ." . ++$ID;
204	AnyEvent::Socket::format_hostport AnyEvent::Socket::format_address $host, $service	391	my $port = "$NODE#$id";
205	];	392
		393	rcv $port, shift \|\| \&_kilme;
		394
		395	$port
		396	}
		397
		398	=item rcv $local_port, $callback->(@msg)
		399
		400	Replaces the default callback on the specified port. There is no way to
		401	remove the default callback: use C<sub { }> to disable it, or better
		402	C<kil> the port when it is no longer needed.
		403
		404	The global C<$SELF> (exported by this module) contains C<$port> while
		405	executing the callback. Runtime errors during callback execution will
		406	result in the port being C<kil>ed.
		407
		408	The default callback received all messages not matched by a more specific
		409	C<tag> match.
		410
		411	=item rcv $local_port, tag => $callback->(@msg_without_tag), ...
		412
		413	Register (or replace) callbacks to be called on messages starting with the
		414	given tag on the given port (and return the port), or unregister it (when
		415	C<$callback> is C<$undef> or missing). There can only be one callback
		416	registered for each tag.
		417
		418	The original message will be passed to the callback, after the first
		419	element (the tag) has been removed. The callback will use the same
		420	environment as the default callback (see above).
		421
		422	Example: create a port and bind receivers on it in one go.
		423
		424	my $port = rcv port,
		425	msg1 => sub { ... },
		426	msg2 => sub { ... },
		427	;
		428
		429	Example: create a port, bind receivers and send it in a message elsewhere
		430	in one go:
		431
		432	snd $otherport, reply =>
		433	rcv port,
		434	msg1 => sub { ... },
		435	...
		436	;
		437
		438	Example: temporarily register a rcv callback for a tag matching some port
		439	(e.g. for an rpc reply) and unregister it after a message was received.
		440
		441	rcv $port, $otherport => sub {
		442	my @reply = @_;
		443
		444	rcv $SELF, $otherport;
		445	};
		446
		447	=cut
		448
		449	sub rcv($@) {
		450	my $port = shift;
		451	my ($nodeid, $portid) = split /#/, $port, 2;
		452
		453	$NODE{$nodeid} == $NODE{""}
		454	or Carp::croak "$port: rcv can only be called on local ports, caught";
		455
		456	while (@_) {
		457	if (ref $_[0]) {
		458	if (my $self = $PORT_DATA{$portid}) {
		459	"AnyEvent::MP::Port" eq ref $self
		460	or Carp::croak "$port: rcv can only be called on message matching ports, caught";
		461
		462	$self->[0] = shift;
		463	} else {
		464	my $cb = shift;
		465	$PORT{$portid} = sub {
		466	local $SELF = $port;
		467	eval { &$cb }; _self_die if $@;
206	}	468	};
207	$cv->end;	469	}
		470	} elsif (defined $_[0]) {
		471	my $self = $PORT_DATA{$portid} \|\|= do {
		472	my $self = bless [$PORT{$portid} \|\| sub { }, { }, $port], "AnyEvent::MP::Port";
		473
		474	$PORT{$portid} = sub {
		475	local $SELF = $port;
		476
		477	if (my $cb = $self->[1]{$_[0]}) {
		478	shift;
		479	eval { &$cb }; _self_die if $@;
		480	} else {
		481	&{ $self->[0] };
		482	}
		483	};
		484
		485	$self
208	};	486	};
209		487
210	# my (undef, undef, undef, undef, @ipv4) = gethostbyname $nodename;	488	"AnyEvent::MP::Port" eq ref $self
211	#	489	or Carp::croak "$port: rcv can only be called on message matching ports, caught";
212	# for (@ipv4) {	490
213	# push @res, [	491	my ($tag, $cb) = splice @_, 0, 2;
214	# $pri,	492
215	# AnyEvent::Socket::format_hostport AnyEvent::Socket::format_address $_, $t \|\| $DEFAULT_PORT,	493	if (defined $cb) {
216	# ];	494	$self->[1]{$tag} = $cb;
217	# }
218	} else {	495	} else {
219	my ($host, $port) = AnyEvent::Socket::parse_hostport $t, "aemp=$DEFAULT_PORT"	496	delete $self->[1]{$tag};
220	or Carp::croak "$t: unparsable transport descriptor";
221
222	$cv->begin;
223	AnyEvent::Socket::resolve_sockaddr $host, $port, "tcp", 0, undef, sub {
224	for (@_) {
225	my ($service, $host) = AnyEvent::Socket::unpack_sockaddr $_->[3];
226	push @res, [
227	$pri += 1e-5,
228	AnyEvent::Socket::format_hostport AnyEvent::Socket::format_address $host, $service
229	];
230	}
231	$cv->end;
232	}	497	}
233	}	498	}
234	}	499	}
235		500
236	$cv->end;	501	$port
237
238	$cv
239	}	502	}
240		503
241	sub become_public {	504	=item peval $port, $coderef[, @args]
242	return if $PUBLIC;
243		505
244	my $noderef = join ",", ref $_[0] ? @{+shift} : shift;	506	Evaluates the given C<$codref> within the contetx of C<$port>, that is,
245	my @args = @_;	507	when the code throews an exception the C<$port> will be killed.
246		508
247	$NODE = (normalise_noderef $noderef)->recv;	509	Any remaining args will be passed to the callback. Any return values will
		510	be returned to the caller.
248		511
249	my $self = new AnyEvent::MP::Node::Self noderef => $NODE;	512	This is useful when you temporarily want to execute code in the context of
		513	a port.
250		514
251	$NODE{""} = $self; # empty string == local node	515	Example: create a port and run some initialisation code in it's context.
252		516
253	for my $t (split /,/, $NODE) {	517	my $port = port { ... };
254	$NODE{$t} = $self;
255		518
256	my ($host, $port) = AnyEvent::Socket::parse_hostport $t;	519	peval $port, sub {
		520	init
		521	or die "unable to init";
		522	};
257		523
258	$LISTENER{$t} = AnyEvent::MP::Transport::mp_server $host, $port,	524	=cut
259	@args,
260	on_error => sub {
261	die "on_error<@_>\n";#d#
262	},
263	on_connect => sub {
264	my ($tp) = @_;
265		525
266	$NODE{$tp->{remote_id}} = $_[0];	526	sub peval($$) {
267	},	527	local $SELF = shift;
268	sub {	528	my $cb = shift;
269	my ($tp) = @_;
270		529
271	$NODE{"$tp->{peerhost}:$tp->{peerport}"} = $tp;	530	if (wantarray) {
272	},	531	my @res = eval { &$cb };
273	;	532	_self_die if $@;
		533	@res
		534	} else {
		535	my $res = eval { &$cb };
		536	_self_die if $@;
		537	$res
274	}	538	}
		539	}
275		540
276	$PUBLIC = 1;	541	=item $closure = psub { BLOCK }
		542
		543	Remembers C<$SELF> and creates a closure out of the BLOCK. When the
		544	closure is executed, sets up the environment in the same way as in C<rcv>
		545	callbacks, i.e. runtime errors will cause the port to get C<kil>ed.
		546
		547	The effect is basically as if it returned C<< sub { peval $SELF, sub {
		548	BLOCK }, @_ } >>.
		549
		550	This is useful when you register callbacks from C<rcv> callbacks:
		551
		552	rcv delayed_reply => sub {
		553	my ($delay, @reply) = @_;
		554	my $timer = AE::timer $delay, 0, psub {
		555	snd @reply, $SELF;
		556	};
		557	};
		558
		559	=cut
		560
		561	sub psub(&) {
		562	my $cb = shift;
		563
		564	my $port = $SELF
		565	or Carp::croak "psub can only be called from within rcv or psub callbacks, not";
		566
		567	sub {
		568	local $SELF = $port;
		569
		570	if (wantarray) {
		571	my @res = eval { &$cb };
		572	_self_die if $@;
		573	@res
		574	} else {
		575	my $res = eval { &$cb };
		576	_self_die if $@;
		577	$res
		578	}
		579	}
		580	}
		581
		582	=item $guard = mon $port, $cb->(@reason) # call $cb when $port dies
		583
		584	=item $guard = mon $port, $rcvport # kill $rcvport when $port dies
		585
		586	=item $guard = mon $port # kill $SELF when $port dies
		587
		588	=item $guard = mon $port, $rcvport, @msg # send a message when $port dies
		589
		590	Monitor the given port and do something when the port is killed or
		591	messages to it were lost, and optionally return a guard that can be used
		592	to stop monitoring again.
		593
		594	In the first form (callback), the callback is simply called with any
		595	number of C<@reason> elements (no @reason means that the port was deleted
		596	"normally"). Note also that I<< the callback B<must> never die >>, so use
		597	C<eval> if unsure.
		598
		599	In the second form (another port given), the other port (C<$rcvport>)
		600	will be C<kil>'ed with C<@reason>, if a @reason was specified, i.e. on
		601	"normal" kils nothing happens, while under all other conditions, the other
		602	port is killed with the same reason.
		603
		604	The third form (kill self) is the same as the second form, except that
		605	C<$rvport> defaults to C<$SELF>.
		606
		607	In the last form (message), a message of the form C<@msg, @reason> will be
		608	C<snd>.
		609
		610	Monitoring-actions are one-shot: once messages are lost (and a monitoring
		611	alert was raised), they are removed and will not trigger again.
		612
		613	As a rule of thumb, monitoring requests should always monitor a port from
		614	a local port (or callback). The reason is that kill messages might get
		615	lost, just like any other message. Another less obvious reason is that
		616	even monitoring requests can get lost (for example, when the connection
		617	to the other node goes down permanently). When monitoring a port locally
		618	these problems do not exist.
		619
		620	C<mon> effectively guarantees that, in the absence of hardware failures,
		621	after starting the monitor, either all messages sent to the port will
		622	arrive, or the monitoring action will be invoked after possible message
		623	loss has been detected. No messages will be lost "in between" (after
		624	the first lost message no further messages will be received by the
		625	port). After the monitoring action was invoked, further messages might get
		626	delivered again.
		627
		628	Inter-host-connection timeouts and monitoring depend on the transport
		629	used. The only transport currently implemented is TCP, and AnyEvent::MP
		630	relies on TCP to detect node-downs (this can take 10-15 minutes on a
		631	non-idle connection, and usually around two hours for idle connections).
		632
		633	This means that monitoring is good for program errors and cleaning up
		634	stuff eventually, but they are no replacement for a timeout when you need
		635	to ensure some maximum latency.
		636
		637	Example: call a given callback when C<$port> is killed.
		638
		639	mon $port, sub { warn "port died because of <@_>\n" };
		640
		641	Example: kill ourselves when C<$port> is killed abnormally.
		642
		643	mon $port;
		644
		645	Example: send us a restart message when another C<$port> is killed.
		646
		647	mon $port, $self => "restart";
		648
		649	=cut
		650
		651	sub mon {
		652	my ($nodeid, $port) = split /#/, shift, 2;
		653
		654	my $node = $NODE{$nodeid} \|\| add_node $nodeid;
		655
		656	my $cb = @_ ? shift : $SELF \|\| Carp::croak 'mon: called with one argument only, but $SELF not set,';
		657
		658	unless (ref $cb) {
		659	if (@_) {
		660	# send a kill info message
		661	my (@msg) = ($cb, @_);
		662	$cb = sub { snd @msg, @_ };
		663	} else {
		664	# simply kill other port
		665	my $port = $cb;
		666	$cb = sub { kil $port, @_ if @_ };
		667	}
		668	}
		669
		670	$node->monitor ($port, $cb);
		671
		672	defined wantarray
		673	and ($cb += 0, AnyEvent::Util::guard { $node->unmonitor ($port, $cb) })
		674	}
		675
		676	=item $guard = mon_guard $port, $ref, $ref...
		677
		678	Monitors the given C<$port> and keeps the passed references. When the port
		679	is killed, the references will be freed.
		680
		681	Optionally returns a guard that will stop the monitoring.
		682
		683	This function is useful when you create e.g. timers or other watchers and
		684	want to free them when the port gets killed (note the use of C<psub>):
		685
		686	$port->rcv (start => sub {
		687	my $timer; $timer = mon_guard $port, AE::timer 1, 1, psub {
		688	undef $timer if 0.9 < rand;
		689	});
		690	});
		691
		692	=cut
		693
		694	sub mon_guard {
		695	my ($port, @refs) = @_;
		696
		697	#TODO: mon-less form?
		698
		699	mon $port, sub { 0 && @refs }
		700	}
		701
		702	=item kil $port[, @reason]
		703
		704	Kill the specified port with the given C<@reason>.
		705
		706	If no C<@reason> is specified, then the port is killed "normally" -
		707	monitor callback will be invoked, but the kil will not cause linked ports
		708	(C<mon $mport, $lport> form) to get killed.
		709
		710	If a C<@reason> is specified, then linked ports (C<mon $mport, $lport>
		711	form) get killed with the same reason.
		712
		713	Runtime errors while evaluating C<rcv> callbacks or inside C<psub> blocks
		714	will be reported as reason C<< die => $@ >>.
		715
		716	Transport/communication errors are reported as C<< transport_error =>
		717	$message >>.
		718
		719	=cut
		720
		721	=item $port = spawn $node, $initfunc[, @initdata]
		722
		723	Creates a port on the node C<$node> (which can also be a port ID, in which
		724	case it's the node where that port resides).
		725
		726	The port ID of the newly created port is returned immediately, and it is
		727	possible to immediately start sending messages or to monitor the port.
		728
		729	After the port has been created, the init function is called on the remote
		730	node, in the same context as a C<rcv> callback. This function must be a
		731	fully-qualified function name (e.g. C<MyApp::Chat::Server::init>). To
		732	specify a function in the main program, use C<::name>.
		733
		734	If the function doesn't exist, then the node tries to C<require>
		735	the package, then the package above the package and so on (e.g.
		736	C<MyApp::Chat::Server>, C<MyApp::Chat>, C<MyApp>) until the function
		737	exists or it runs out of package names.
		738
		739	The init function is then called with the newly-created port as context
		740	object (C<$SELF>) and the C<@initdata> values as arguments. It I<must>
		741	call one of the C<rcv> functions to set callbacks on C<$SELF>, otherwise
		742	the port might not get created.
		743
		744	A common idiom is to pass a local port, immediately monitor the spawned
		745	port, and in the remote init function, immediately monitor the passed
		746	local port. This two-way monitoring ensures that both ports get cleaned up
		747	when there is a problem.
		748
		749	C<spawn> guarantees that the C<$initfunc> has no visible effects on the
		750	caller before C<spawn> returns (by delaying invocation when spawn is
		751	called for the local node).
		752
		753	Example: spawn a chat server port on C<$othernode>.
		754
		755	# this node, executed from within a port context:
		756	my $server = spawn $othernode, "MyApp::Chat::Server::connect", $SELF;
		757	mon $server;
		758
		759	# init function on C<$othernode>
		760	sub connect {
		761	my ($srcport) = @_;
		762
		763	mon $srcport;
		764
		765	rcv $SELF, sub {
		766	...
		767	};
		768	}
		769
		770	=cut
		771
		772	sub _spawn {
		773	my $port = shift;
		774	my $init = shift;
		775
		776	# rcv will create the actual port
		777	local $SELF = "$NODE#$port";
		778	eval {
		779	&{ load_func $init }
		780	};
		781	_self_die if $@;
		782	}
		783
		784	sub spawn(@) {
		785	my ($nodeid, undef) = split /#/, shift, 2;
		786
		787	my $id = "$RUNIQ." . ++$ID;
		788
		789	$_[0] =~ /::/
		790	or Carp::croak "spawn init function must be a fully-qualified name, caught";
		791
		792	snd_to_func $nodeid, "AnyEvent::MP::_spawn" => $id, @_;
		793
		794	"$nodeid#$id"
		795	}
		796
		797
		798	=item after $timeout, @msg
		799
		800	=item after $timeout, $callback
		801
		802	Either sends the given message, or call the given callback, after the
		803	specified number of seconds.
		804
		805	This is simply a utility function that comes in handy at times - the
		806	AnyEvent::MP author is not convinced of the wisdom of having it, though,
		807	so it may go away in the future.
		808
		809	=cut
		810
		811	sub after($@) {
		812	my ($timeout, @action) = @_;
		813
		814	my $t; $t = AE::timer $timeout, 0, sub {
		815	undef $t;
		816	ref $action[0]
		817	? $action[0]()
		818	: snd @action;
		819	};
		820	}
		821
		822	=item cal $port, @msg, $callback[, $timeout]
		823
		824	A simple form of RPC - sends a message to the given C<$port> with the
		825	given contents (C<@msg>), but adds a reply port to the message.
		826
		827	The reply port is created temporarily just for the purpose of receiving
		828	the reply, and will be C<kil>ed when no longer needed.
		829
		830	A reply message sent to the port is passed to the C<$callback> as-is.
		831
		832	If an optional time-out (in seconds) is given and it is not C<undef>,
		833	then the callback will be called without any arguments after the time-out
		834	elapsed and the port is C<kil>ed.
		835
		836	If no time-out is given (or it is C<undef>), then the local port will
		837	monitor the remote port instead, so it eventually gets cleaned-up.
		838
		839	Currently this function returns the temporary port, but this "feature"
		840	might go in future versions unless you can make a convincing case that
		841	this is indeed useful for something.
		842
		843	=cut
		844
		845	sub cal(@) {
		846	my $timeout = ref $_[-1] ? undef : pop;
		847	my $cb = pop;
		848
		849	my $port = port {
		850	undef $timeout;
		851	kil $SELF;
		852	&$cb;
		853	};
		854
		855	if (defined $timeout) {
		856	$timeout = AE::timer $timeout, 0, sub {
		857	undef $timeout;
		858	kil $port;
		859	$cb->();
		860	};
		861	} else {
		862	mon $_[0], sub {
		863	kil $port;
		864	$cb->();
		865	};
		866	}
		867
		868	push @_, $port;
		869	&snd;
		870
		871	$port
277	}	872	}
278		873
279	=back	874	=back
280		875
		876	=head1 AnyEvent::MP vs. Distributed Erlang
		877
		878	AnyEvent::MP got lots of its ideas from distributed Erlang (Erlang node
		879	== aemp node, Erlang process == aemp port), so many of the documents and
		880	programming techniques employed by Erlang apply to AnyEvent::MP. Here is a
		881	sample:
		882
		883	http://www.erlang.se/doc/programming_rules.shtml
		884	http://erlang.org/doc/getting_started/part_frame.html # chapters 3 and 4
		885	http://erlang.org/download/erlang-book-part1.pdf # chapters 5 and 6
		886	http://erlang.org/download/armstrong_thesis_2003.pdf # chapters 4 and 5
		887
		888	Despite the similarities, there are also some important differences:
		889
		890	=over 4
		891
		892	=item * Node IDs are arbitrary strings in AEMP.
		893
		894	Erlang relies on special naming and DNS to work everywhere in the same
		895	way. AEMP relies on each node somehow knowing its own address(es) (e.g. by
		896	configuration or DNS), and possibly the addresses of some seed nodes, but
		897	will otherwise discover other nodes (and their IDs) itself.
		898
		899	=item * Erlang has a "remote ports are like local ports" philosophy, AEMP
		900	uses "local ports are like remote ports".
		901
		902	The failure modes for local ports are quite different (runtime errors
		903	only) then for remote ports - when a local port dies, you I<know> it dies,
		904	when a connection to another node dies, you know nothing about the other
		905	port.
		906
		907	Erlang pretends remote ports are as reliable as local ports, even when
		908	they are not.
		909
		910	AEMP encourages a "treat remote ports differently" philosophy, with local
		911	ports being the special case/exception, where transport errors cannot
		912	occur.
		913
		914	=item * Erlang uses processes and a mailbox, AEMP does not queue.
		915
		916	Erlang uses processes that selectively receive messages out of order, and
		917	therefore needs a queue. AEMP is event based, queuing messages would serve
		918	no useful purpose. For the same reason the pattern-matching abilities
		919	of AnyEvent::MP are more limited, as there is little need to be able to
		920	filter messages without dequeuing them.
		921
		922	This is not a philosophical difference, but simply stems from AnyEvent::MP
		923	being event-based, while Erlang is process-based.
		924
		925	You cna have a look at L<Coro::MP> for a more Erlang-like process model on
		926	top of AEMP and Coro threads.
		927
		928	=item * Erlang sends are synchronous, AEMP sends are asynchronous.
		929
		930	Sending messages in Erlang is synchronous and blocks the process until
		931	a conenction has been established and the message sent (and so does not
		932	need a queue that can overflow). AEMP sends return immediately, connection
		933	establishment is handled in the background.
		934
		935	=item * Erlang suffers from silent message loss, AEMP does not.
		936
		937	Erlang implements few guarantees on messages delivery - messages can get
		938	lost without any of the processes realising it (i.e. you send messages a,
		939	b, and c, and the other side only receives messages a and c).
		940
		941	AEMP guarantees (modulo hardware errors) correct ordering, and the
		942	guarantee that after one message is lost, all following ones sent to the
		943	same port are lost as well, until monitoring raises an error, so there are
		944	no silent "holes" in the message sequence.
		945
		946	If you want your software to be very reliable, you have to cope with
		947	corrupted and even out-of-order messages in both Erlang and AEMP. AEMP
		948	simply tries to work better in common error cases, such as when a network
		949	link goes down.
		950
		951	=item * Erlang can send messages to the wrong port, AEMP does not.
		952
		953	In Erlang it is quite likely that a node that restarts reuses an Erlang
		954	process ID known to other nodes for a completely different process,
		955	causing messages destined for that process to end up in an unrelated
		956	process.
		957
		958	AEMP does not reuse port IDs, so old messages or old port IDs floating
		959	around in the network will not be sent to an unrelated port.
		960
		961	=item * Erlang uses unprotected connections, AEMP uses secure
		962	authentication and can use TLS.
		963
		964	AEMP can use a proven protocol - TLS - to protect connections and
		965	securely authenticate nodes.
		966
		967	=item * The AEMP protocol is optimised for both text-based and binary
		968	communications.
		969
		970	The AEMP protocol, unlike the Erlang protocol, supports both programming
		971	language independent text-only protocols (good for debugging), and binary,
		972	language-specific serialisers (e.g. Storable). By default, unless TLS is
		973	used, the protocol is actually completely text-based.
		974
		975	It has also been carefully designed to be implementable in other languages
		976	with a minimum of work while gracefully degrading functionality to make the
		977	protocol simple.
		978
		979	=item * AEMP has more flexible monitoring options than Erlang.
		980
		981	In Erlang, you can chose to receive I<all> exit signals as messages or
		982	I<none>, there is no in-between, so monitoring single Erlang processes is
		983	difficult to implement.
		984
		985	Monitoring in AEMP is more flexible than in Erlang, as one can choose
		986	between automatic kill, exit message or callback on a per-port basis.
		987
		988	=item * Erlang tries to hide remote/local connections, AEMP does not.
		989
		990	Monitoring in Erlang is not an indicator of process death/crashes, in the
		991	same way as linking is (except linking is unreliable in Erlang).
		992
		993	In AEMP, you don't "look up" registered port names or send to named ports
		994	that might or might not be persistent. Instead, you normally spawn a port
		995	on the remote node. The init function monitors you, and you monitor the
		996	remote port. Since both monitors are local to the node, they are much more
		997	reliable (no need for C<spawn_link>).
		998
		999	This also saves round-trips and avoids sending messages to the wrong port
		1000	(hard to do in Erlang).
		1001
		1002	=back
		1003
		1004	=head1 RATIONALE
		1005
		1006	=over 4
		1007
		1008	=item Why strings for port and node IDs, why not objects?
		1009
		1010	We considered "objects", but found that the actual number of methods
		1011	that can be called are quite low. Since port and node IDs travel over
		1012	the network frequently, the serialising/deserialising would add lots of
		1013	overhead, as well as having to keep a proxy object everywhere.
		1014
		1015	Strings can easily be printed, easily serialised etc. and need no special
		1016	procedures to be "valid".
		1017
		1018	And as a result, a port with just a default receiver consists of a single
		1019	code reference stored in a global hash - it can't become much cheaper.
		1020
		1021	=item Why favour JSON, why not a real serialising format such as Storable?
		1022
		1023	In fact, any AnyEvent::MP node will happily accept Storable as framing
		1024	format, but currently there is no way to make a node use Storable by
		1025	default (although all nodes will accept it).
		1026
		1027	The default framing protocol is JSON because a) JSON::XS is many times
		1028	faster for small messages and b) most importantly, after years of
		1029	experience we found that object serialisation is causing more problems
		1030	than it solves: Just like function calls, objects simply do not travel
		1031	easily over the network, mostly because they will always be a copy, so you
		1032	always have to re-think your design.
		1033
		1034	Keeping your messages simple, concentrating on data structures rather than
		1035	objects, will keep your messages clean, tidy and efficient.
		1036
		1037	=back
		1038
281	=head1 SEE ALSO	1039	=head1 SEE ALSO
		1040
		1041	L<AnyEvent::MP::Intro> - a gentle introduction.
		1042
		1043	L<AnyEvent::MP::Kernel> - more, lower-level, stuff.
		1044
		1045	L<AnyEvent::MP::Global> - network maintenance and port groups, to find
		1046	your applications.
		1047
		1048	L<AnyEvent::MP::DataConn> - establish data connections between nodes.
		1049
		1050	L<AnyEvent::MP::LogCatcher> - simple service to display log messages from
		1051	all nodes.
282		1052
283	L<AnyEvent>.	1053	L<AnyEvent>.
284		1054
285	=head1 AUTHOR	1055	=head1 AUTHOR
286		1056

Diff Legend

-–
+Removed lines
-+
+Added lines
-<
+Changed lines
->
+Changed lines

Comparing AnyEvent-MP/MP.pm (file contents): Revision 1.2 by root, Fri Jul 31 20:55:46 2009 UTC vs. Revision 1.121 by root, Tue Feb 28 18:37:24 2012 UTC

Diff Legend

Comparing AnyEvent-MP/MP.pm (file contents):
Revision 1.2 by root, Fri Jul 31 20:55:46 2009 UTC vs.
Revision 1.121 by root, Tue Feb 28 18:37:24 2012 UTC