[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 1.52 by root, Fri Aug 14 15:13:20 2009 UTC

…		…
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 noderef
		10	NODE # returns this node's noderef
		11	NODE $port # returns the noderef of the port
11		12
		13	$SELF # receiving/own port id in rcv callbacks
		14
		15	# initialise the node so it can send/receive messages
		16	initialise_node; # -OR-
		17	initialise_node "localhost:4040"; # -OR-
		18	initialise_node "slave/", "localhost:4040"
		19
		20	# ports are message endpoints
		21
		22	# sending messages
12	snd $port, type => data...;	23	snd $port, type => data...;
		24	snd $port, @msg;
		25	snd @msg_with_first_element_being_a_port;
13		26
14	rcv $port, smartmatch => $cb->($port, @msg);	27	# creating/using ports, the simple way
		28	my $somple_port = port { my @msg = @_; 0 };
15		29
16	# examples:	30	# creating/using ports, tagged message matching
		31	my $port = port;
17	rcv $port2, ping => sub { snd $_[0], "pong"; 0 };	32	rcv $port, ping => sub { snd $_[0], "pong"; 0 };
18	rcv $port1, pong => sub { warn "pong received\n" };	33	rcv $port, pong => sub { warn "pong received\n"; 0 };
19	snd $port2, ping => $port1;
20		34
21	# more, smarter, matches (_any_ is exported by this module)	35	# create a port on another node
22	rcv $port, [child_died => $pid] => sub { ...	36	my $port = spawn $node, $initfunc, @initdata;
23	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
		42
		43	=head1 CURRENT STATUS
		44
		45	AnyEvent::MP - stable API, should work
		46	AnyEvent::MP::Intro - outdated
		47	AnyEvent::MP::Kernel - WIP
		48	AnyEvent::MP::Transport - mostly stable
		49
		50	stay tuned.
24		51
25	=head1 DESCRIPTION	52	=head1 DESCRIPTION
26		53
27	This module (-family) implements a simple message passing framework.	54	This module (-family) implements a simple message passing framework.
28		55
29	Despite its simplicity, you can securely message other processes running	56	Despite its simplicity, you can securely message other processes running
30	on the same or other hosts.	57	on the same or other hosts.
31		58
		59	For an introduction to this module family, see the L<AnyEvent::MP::Intro>
		60	manual page.
		61
		62	At the moment, this module family is severly broken and underdocumented,
		63	so do not use. This was uploaded mainly to reserve the CPAN namespace -
		64	stay tuned!
		65
32	=head1 CONCEPTS	66	=head1 CONCEPTS
33		67
34	=over 4	68	=over 4
35		69
36	=item port	70	=item port
37		71
38	A port is something you can send messages to with the C<snd> function, and	72	A port is something you can send messages to (with the C<snd> function).
39	you can register C<rcv> handlers with. All C<rcv> handlers will receive
40	messages they match, messages will not be queued.
41		73
		74	Some ports allow you to register C<rcv> handlers that can match specific
		75	messages. All C<rcv> handlers will receive messages they match, messages
		76	will not be queued.
		77
42	=item port id - C<pid@host#portname>	78	=item port id - C<noderef#portname>
43		79
44	A port id is always the node id, a hash-mark (C<#>) as separator, followed	80	A port id is normaly the concatenation of a noderef, a hash-mark (C<#>) as
45	by a port name.	81	separator, and a port name (a printable string of unspecified format). An
46		82	exception is the the node port, whose ID is identical to its node
47	A port name can be a well known port (basically an identifier/bareword),	83	reference.
48	or a generated name, consisting of node id, a dot (C<.>), and an
49	identifier.
50		84
51	=item node	85	=item node
52		86
53	A node is a single process containing at least one port - the node	87	A node is a single process containing at least one port - the node
54	port. You can send messages to node ports to let them create new ports,	88	port. You can send messages to node ports to find existing ports or to
55	among other things.	89	create new ports, among other things.
56		90
57	Initially, nodes are either private (single-process only) or hidden	91	Nodes are either private (single-process only), slaves (connected to a
58	(connected to a father node only). Only when they epxlicitly "go public"	92	master node only) or public nodes (connectable from unrelated nodes).
59	can you send them messages form unrelated other nodes.
60		93
61	Public nodes automatically connect to all other public nodes in a network	94	=item noderef - C<host:port,host:port...>, C<id@noderef>, C<id>
62	when they connect, creating a full mesh.
63		95
64	=item node id - C<host:port>, C<id@host>, C<id>
65
66	A node ID is a string that either uniquely identifies a given node (For	96	A node reference is a string that either simply identifies the node (for
67	private and hidden nodes), or contains a recipe on how to reach a given	97	private and slave nodes), or contains a recipe on how to reach a given
68	node (for public nodes).	98	node (for public nodes).
69		99
		100	This recipe is simply a comma-separated list of C<address:port> pairs (for
		101	TCP/IP, other protocols might look different).
		102
		103	Node references come in two flavours: resolved (containing only numerical
		104	addresses) or unresolved (where hostnames are used instead of addresses).
		105
		106	Before using an unresolved node reference in a message you first have to
		107	resolve it.
		108
70	=back	109	=back
71		110
72	=head1 FUNCTIONS	111	=head1 VARIABLES/FUNCTIONS
73		112
74	=over 4	113	=over 4
75		114
76	=cut	115	=cut
77		116
78	package AnyEvent::MP;	117	package AnyEvent::MP;
79		118
80	use AnyEvent::MP::Util ();
81	use AnyEvent::MP::Node;	119	use AnyEvent::MP::Kernel;
82	use AnyEvent::MP::Transport;
83		120
84	use utf8;
85	use common::sense;	121	use common::sense;
86		122
87	use Carp ();	123	use Carp ();
88		124
89	use AE ();	125	use AE ();
90		126
91	use base "Exporter";	127	use base "Exporter";
92		128
93	our $VERSION = '0.0';	129	our $VERSION = $AnyEvent::MP::Kernel::VERSION;
94	our @EXPORT = qw(NODE $NODE $PORT snd rcv _any_);
95		130
96	our $DEFAULT_SECRET;	131	our @EXPORT = qw(
97	our $DEFAULT_PORT = "4040";	132	NODE $NODE *SELF node_of _any_
		133	resolve_node initialise_node
		134	snd rcv mon kil reg psub spawn
		135	port
		136	);
98		137
99	our $CONNECT_INTERVAL = 5; # new connect every 5s, at least	138	our $SELF;
100	our $CONNECT_TIMEOUT = 30; # includes handshake
101		139
102	sub default_secret {	140	sub _self_die() {
103	unless (defined $DEFAULT_SECRET) {	141	my $msg = $@;
104	if (open my $fh, "<$ENV{HOME}/.aemp-secret") {	142	$msg =~ s/\n+$// unless ref $msg;
105	sysread $fh, $DEFAULT_SECRET, -s $fh;	143	kil $SELF, die => $msg;
106	} else {
107	$DEFAULT_SECRET = AnyEvent::MP::Util::nonce 32;
108	}
109	}
110
111	$DEFAULT_SECRET
112	}	144	}
113		145
114	our $UNIQ = sprintf "%x.%x", $$, time; # per-process/node unique cookie	146	=item $thisnode = NODE / $NODE
115	our $PUBLIC = 0;
116	our $NODE;
117	our $PORT;
118		147
119	our %NODE; # node id to transport mapping, or "undef", for local node	148	The C<NODE> function returns, and the C<$NODE> variable contains the
120	our %PORT; # local ports	149	noderef of the local node. The value is initialised by a call to
121	our %LISTENER; # local transports	150	C<initialise_node>.
122		151
123	sub NODE() { $NODE }	152	=item $noderef = node_of $port
124		153
125	{	154	Extracts and returns the noderef from a port ID or a noderef.
126	use POSIX ();	155
127	my $nodename = (POSIX::uname)[1];	156	=item initialise_node $noderef, $seednode, $seednode...
128	$NODE = "$$\@$nodename";	157
		158	=item initialise_node "slave/", $master, $master...
		159
		160	Before a node can talk to other nodes on the network it has to initialise
		161	itself - the minimum a node needs to know is it's own name, and optionally
		162	it should know the noderefs of some other nodes in the network.
		163
		164	This function initialises a node - it must be called exactly once (or
		165	never) before calling other AnyEvent::MP functions.
		166
		167	All arguments (optionally except for the first) are noderefs, which can be
		168	either resolved or unresolved.
		169
		170	The first argument will be looked up in the configuration database first
		171	(if it is C<undef> then the current nodename will be used instead) to find
		172	the relevant configuration profile (see L<aemp>). If none is found then
		173	the default configuration is used. The configuration supplies additional
		174	seed/master nodes and can override the actual noderef.
		175
		176	There are two types of networked nodes, public nodes and slave nodes:
		177
		178	=over 4
		179
		180	=item public nodes
		181
		182	For public nodes, C<$noderef> (supplied either directly to
		183	C<initialise_node> or indirectly via a profile or the nodename) must be a
		184	noderef (possibly unresolved, in which case it will be resolved).
		185
		186	After resolving, the node will bind itself on all endpoints and try to
		187	connect to all additional C<$seednodes> that are specified. Seednodes are
		188	optional and can be used to quickly bootstrap the node into an existing
		189	network.
		190
		191	=item slave nodes
		192
		193	When the C<$noderef> (either as given or overriden by the config file)
		194	is the special string C<slave/>, then the node will become a slave
		195	node. Slave nodes cannot be contacted from outside and will route most of
		196	their traffic to the master node that they attach to.
		197
		198	At least one additional noderef is required (either by specifying it
		199	directly or because it is part of the configuration profile): The node
		200	will try to connect to all of them and will become a slave attached to the
		201	first node it can successfully connect to.
		202
		203	=back
		204
		205	This function will block until all nodes have been resolved and, for slave
		206	nodes, until it has successfully established a connection to a master
		207	server.
		208
		209	Example: become a public node listening on the guessed noderef, or the one
		210	specified via C<aemp> for the current node. This should be the most common
		211	form of invocation for "daemon"-type nodes.
		212
		213	initialise_node;
		214
		215	Example: become a slave node to any of the the seednodes specified via
		216	C<aemp>. This form is often used for commandline clients.
		217
		218	initialise_node "slave/";
		219
		220	Example: become a slave node to any of the specified master servers. This
		221	form is also often used for commandline clients.
		222
		223	initialise_node "slave/", "master1", "192.168.13.17", "mp.example.net";
		224
		225	Example: become a public node, and try to contact some well-known master
		226	servers to become part of the network.
		227
		228	initialise_node undef, "master1", "master2";
		229
		230	Example: become a public node listening on port C<4041>.
		231
		232	initialise_node 4041;
		233
		234	Example: become a public node, only visible on localhost port 4044.
		235
		236	initialise_node "localhost:4044";
		237
		238	=item $cv = resolve_node $noderef
		239
		240	Takes an unresolved node reference that may contain hostnames and
		241	abbreviated IDs, resolves all of them and returns a resolved node
		242	reference.
		243
		244	In addition to C<address:port> pairs allowed in resolved noderefs, the
		245	following forms are supported:
		246
		247	=over 4
		248
		249	=item the empty string
		250
		251	An empty-string component gets resolved as if the default port (4040) was
		252	specified.
		253
		254	=item naked port numbers (e.g. C<1234>)
		255
		256	These are resolved by prepending the local nodename and a colon, to be
		257	further resolved.
		258
		259	=item hostnames (e.g. C<localhost:1234>, C<localhost>)
		260
		261	These are resolved by using AnyEvent::DNS to resolve them, optionally
		262	looking up SRV records for the C<aemp=4040> port, if no port was
		263	specified.
		264
		265	=back
		266
		267	=item $SELF
		268
		269	Contains the current port id while executing C<rcv> callbacks or C<psub>
		270	blocks.
		271
		272	=item SELF, %SELF, @SELF...
		273
		274	Due to some quirks in how perl exports variables, it is impossible to
		275	just export C<$SELF>, all the symbols called C<SELF> are exported by this
		276	module, but only C<$SELF> is currently used.
		277
		278	=item snd $port, type => @data
		279
		280	=item snd $port, @msg
		281
		282	Send the given message to the given port ID, which can identify either
		283	a local or a remote port, and must be a port ID.
		284
		285	While the message can be about anything, it is highly recommended to use a
		286	string as first element (a port ID, or some word that indicates a request
		287	type etc.).
		288
		289	The message data effectively becomes read-only after a call to this
		290	function: modifying any argument is not allowed and can cause many
		291	problems.
		292
		293	The type of data you can transfer depends on the transport protocol: when
		294	JSON is used, then only strings, numbers and arrays and hashes consisting
		295	of those are allowed (no objects). When Storable is used, then anything
		296	that Storable can serialise and deserialise is allowed, and for the local
		297	node, anything can be passed.
		298
		299	=item $local_port = port
		300
		301	Create a new local port object and returns its port ID. Initially it has
		302	no callbacks set and will throw an error when it receives messages.
		303
		304	=item $local_port = port { my @msg = @_ }
		305
		306	Creates a new local port, and returns its ID. Semantically the same as
		307	creating a port and calling C<rcv $port, $callback> on it.
		308
		309	The block will be called for every message received on the port, with the
		310	global variable C<$SELF> set to the port ID. Runtime errors will cause the
		311	port to be C<kil>ed. The message will be passed as-is, no extra argument
		312	(i.e. no port ID) will be passed to the callback.
		313
		314	If you want to stop/destroy the port, simply C<kil> it:
		315
		316	my $port = port {
		317	my @msg = @_;
		318	...
		319	kil $SELF;
		320	};
		321
		322	=cut
		323
		324	sub rcv($@);
		325
		326	sub _kilme {
		327	die "received message on port without callback";
129	}	328	}
130		329
131	sub _ANY_() { 1 }	330	sub port(;&) {
132	sub _any_() { \&_ANY_ }	331	my $id = "$UNIQ." . $ID++;
		332	my $port = "$NODE#$id";
133		333
134	sub add_node {	334	rcv $port, shift \|\| \&_kilme;
135	my ($noderef) = @_;
136		335
137	return $NODE{$noderef}	336	$port
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	}	337	}
153		338
		339	=item rcv $local_port, $callback->(@msg)
		340
		341	Replaces the default callback on the specified port. There is no way to
		342	remove the default callback: use C<sub { }> to disable it, or better
		343	C<kil> the port when it is no longer needed.
		344
		345	The global C<$SELF> (exported by this module) contains C<$port> while
		346	executing the callback. Runtime errors during callback execution will
		347	result in the port being C<kil>ed.
		348
		349	The default callback received all messages not matched by a more specific
		350	C<tag> match.
		351
		352	=item rcv $local_port, tag => $callback->(@msg_without_tag), ...
		353
		354	Register callbacks to be called on messages starting with the given tag on
		355	the given port (and return the port), or unregister it (when C<$callback>
		356	is C<$undef>).
		357
		358	The original message will be passed to the callback, after the first
		359	element (the tag) has been removed. The callback will use the same
		360	environment as the default callback (see above).
		361
		362	Example: create a port and bind receivers on it in one go.
		363
		364	my $port = rcv port,
		365	msg1 => sub { ... },
		366	msg2 => sub { ... },
		367	;
		368
		369	Example: create a port, bind receivers and send it in a message elsewhere
		370	in one go:
		371
		372	snd $otherport, reply =>
		373	rcv port,
		374	msg1 => sub { ... },
		375	...
		376	;
		377
		378	=cut
		379
154	sub snd($@) {	380	sub rcv($@) {
		381	my $port = shift;
155	my ($noderef, $port) = split /#/, shift, 2;	382	my ($noderef, $portid) = split /#/, $port, 2;
156		383
157	add_node $noderef	384	($NODE{$noderef} \|\| add_node $noderef) == $NODE{""}
158	unless exists $NODE{$noderef};	385	or Carp::croak "$port: rcv can only be called on local ports, caught";
159		386
160	$NODE{$noderef}->send ([$port, [@_]]);	387	while (@_) {
161	}	388	if (ref $_[0]) {
		389	if (my $self = $PORT_DATA{$portid}) {
		390	"AnyEvent::MP::Port" eq ref $self
		391	or Carp::croak "$port: rcv can only be called on message matching ports, caught";
162		392
163	sub _inject {	393	$self->[2] = shift;
164	my ($port, $msg) = @{+shift};	394	} else {
165		395	my $cb = shift;
166	$port = $PORT{$port}	396	$PORT{$portid} = sub {
167	or return;	397	local $SELF = $port;
168		398	eval { &$cb }; _self_die if $@;
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	});
187
188	$noderef = $DEFAULT_PORT unless length $noderef;
189
190	my $idx;
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
198	$cv->begin;
199	AnyEvent::Socket::resolve_sockaddr $nodename, $t \|\| "aemp=$DEFAULT_PORT", "tcp", 0, undef, sub {
200	for (@_) {
201	my ($service, $host) = AnyEvent::Socket::unpack_sockaddr $_->[3];
202	push @res, [
203	$pri += 1e-5,
204	AnyEvent::Socket::format_hostport AnyEvent::Socket::format_address $host, $service
205	];
206	}	399	};
207	$cv->end;	400	}
		401	} elsif (defined $_[0]) {
		402	my $self = $PORT_DATA{$portid} \|\|= do {
		403	my $self = bless [$PORT{$port} \|\| sub { }, { }, $port], "AnyEvent::MP::Port";
		404
		405	$PORT{$portid} = sub {
		406	local $SELF = $port;
		407
		408	if (my $cb = $self->[1]{$_[0]}) {
		409	shift;
		410	eval { &$cb }; _self_die if $@;
		411	} else {
		412	&{ $self->[0] };
		413	}
		414	};
		415
		416	$self
208	};	417	};
209		418
210	# my (undef, undef, undef, undef, @ipv4) = gethostbyname $nodename;	419	"AnyEvent::MP::Port" eq ref $self
211	#	420	or Carp::croak "$port: rcv can only be called on message matching ports, caught";
212	# for (@ipv4) {	421
213	# push @res, [	422	my ($tag, $cb) = splice @_, 0, 2;
214	# $pri,	423
215	# AnyEvent::Socket::format_hostport AnyEvent::Socket::format_address $_, $t \|\| $DEFAULT_PORT,	424	if (defined $cb) {
216	# ];	425	$self->[1]{$tag} = $cb;
217	# }
218	} else {	426	} else {
219	my ($host, $port) = AnyEvent::Socket::parse_hostport $t, "aemp=$DEFAULT_PORT"	427	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	}	428	}
233	}	429	}
234	}	430	}
235		431
236	$cv->end;	432	$port
237
238	$cv
239	}	433	}
240		434
241	sub become_public {	435	=item $closure = psub { BLOCK }
242	return if $PUBLIC;
243		436
244	my $noderef = join ",", ref $_[0] ? @{+shift} : shift;	437	Remembers C<$SELF> and creates a closure out of the BLOCK. When the
245	my @args = @_;	438	closure is executed, sets up the environment in the same way as in C<rcv>
		439	callbacks, i.e. runtime errors will cause the port to get C<kil>ed.
246		440
247	$NODE = (normalise_noderef $noderef)->recv;	441	This is useful when you register callbacks from C<rcv> callbacks:
248		442
249	my $self = new AnyEvent::MP::Node::Self noderef => $NODE;	443	rcv delayed_reply => sub {
250		444	my ($delay, @reply) = @_;
251	$NODE{""} = $self; # empty string == local node	445	my $timer = AE::timer $delay, 0, psub {
252		446	snd @reply, $SELF;
253	for my $t (split /,/, $NODE) {
254	$NODE{$t} = $self;
255
256	my ($host, $port) = AnyEvent::Socket::parse_hostport $t;
257
258	$LISTENER{$t} = AnyEvent::MP::Transport::mp_server $host, $port,
259	@args,
260	on_error => sub {
261	die "on_error<@_>\n";#d#
262	},
263	on_connect => sub {
264	my ($tp) = @_;
265
266	$NODE{$tp->{remote_id}} = $_[0];
267	},
268	sub {
269	my ($tp) = @_;
270
271	$NODE{"$tp->{peerhost}:$tp->{peerport}"} = $tp;
272	},
273	;	447	};
		448	};
		449
		450	=cut
		451
		452	sub psub(&) {
		453	my $cb = shift;
		454
		455	my $port = $SELF
		456	or Carp::croak "psub can only be called from within rcv or psub callbacks, not";
		457
		458	sub {
		459	local $SELF = $port;
		460
		461	if (wantarray) {
		462	my @res = eval { &$cb };
		463	_self_die if $@;
		464	@res
		465	} else {
		466	my $res = eval { &$cb };
		467	_self_die if $@;
		468	$res
		469	}
274	}	470	}
275
276	$PUBLIC = 1;
277	}	471	}
		472
		473	=item $guard = mon $port, $cb->(@reason)
		474
		475	=item $guard = mon $port, $rcvport
		476
		477	=item $guard = mon $port
		478
		479	=item $guard = mon $port, $rcvport, @msg
		480
		481	Monitor the given port and do something when the port is killed or
		482	messages to it were lost, and optionally return a guard that can be used
		483	to stop monitoring again.
		484
		485	C<mon> effectively guarantees that, in the absence of hardware failures,
		486	that after starting the monitor, either all messages sent to the port
		487	will arrive, or the monitoring action will be invoked after possible
		488	message loss has been detected. No messages will be lost "in between"
		489	(after the first lost message no further messages will be received by the
		490	port). After the monitoring action was invoked, further messages might get
		491	delivered again.
		492
		493	In the first form (callback), the callback is simply called with any
		494	number of C<@reason> elements (no @reason means that the port was deleted
		495	"normally"). Note also that I<< the callback B<must> never die >>, so use
		496	C<eval> if unsure.
		497
		498	In the second form (another port given), the other port (C<$rcvport>)
		499	will be C<kil>'ed with C<@reason>, iff a @reason was specified, i.e. on
		500	"normal" kils nothing happens, while under all other conditions, the other
		501	port is killed with the same reason.
		502
		503	The third form (kill self) is the same as the second form, except that
		504	C<$rvport> defaults to C<$SELF>.
		505
		506	In the last form (message), a message of the form C<@msg, @reason> will be
		507	C<snd>.
		508
		509	As a rule of thumb, monitoring requests should always monitor a port from
		510	a local port (or callback). The reason is that kill messages might get
		511	lost, just like any other message. Another less obvious reason is that
		512	even monitoring requests can get lost (for exmaple, when the connection
		513	to the other node goes down permanently). When monitoring a port locally
		514	these problems do not exist.
		515
		516	Example: call a given callback when C<$port> is killed.
		517
		518	mon $port, sub { warn "port died because of <@_>\n" };
		519
		520	Example: kill ourselves when C<$port> is killed abnormally.
		521
		522	mon $port;
		523
		524	Example: send us a restart message when another C<$port> is killed.
		525
		526	mon $port, $self => "restart";
		527
		528	=cut
		529
		530	sub mon {
		531	my ($noderef, $port) = split /#/, shift, 2;
		532
		533	my $node = $NODE{$noderef} \|\| add_node $noderef;
		534
		535	my $cb = @_ ? shift : $SELF \|\| Carp::croak 'mon: called with one argument only, but $SELF not set,';
		536
		537	unless (ref $cb) {
		538	if (@_) {
		539	# send a kill info message
		540	my (@msg) = ($cb, @_);
		541	$cb = sub { snd @msg, @_ };
		542	} else {
		543	# simply kill other port
		544	my $port = $cb;
		545	$cb = sub { kil $port, @_ if @_ };
		546	}
		547	}
		548
		549	$node->monitor ($port, $cb);
		550
		551	defined wantarray
		552	and AnyEvent::Util::guard { $node->unmonitor ($port, $cb) }
		553	}
		554
		555	=item $guard = mon_guard $port, $ref, $ref...
		556
		557	Monitors the given C<$port> and keeps the passed references. When the port
		558	is killed, the references will be freed.
		559
		560	Optionally returns a guard that will stop the monitoring.
		561
		562	This function is useful when you create e.g. timers or other watchers and
		563	want to free them when the port gets killed:
		564
		565	$port->rcv (start => sub {
		566	my $timer; $timer = mon_guard $port, AE::timer 1, 1, sub {
		567	undef $timer if 0.9 < rand;
		568	});
		569	});
		570
		571	=cut
		572
		573	sub mon_guard {
		574	my ($port, @refs) = @_;
		575
		576	#TODO: mon-less form?
		577
		578	mon $port, sub { 0 && @refs }
		579	}
		580
		581	=item kil $port[, @reason]
		582
		583	Kill the specified port with the given C<@reason>.
		584
		585	If no C<@reason> is specified, then the port is killed "normally" (linked
		586	ports will not be kileld, or even notified).
		587
		588	Otherwise, linked ports get killed with the same reason (second form of
		589	C<mon>, see below).
		590
		591	Runtime errors while evaluating C<rcv> callbacks or inside C<psub> blocks
		592	will be reported as reason C<< die => $@ >>.
		593
		594	Transport/communication errors are reported as C<< transport_error =>
		595	$message >>.
		596
		597	=cut
		598
		599	=item $port = spawn $node, $initfunc[, @initdata]
		600
		601	Creates a port on the node C<$node> (which can also be a port ID, in which
		602	case it's the node where that port resides).
		603
		604	The port ID of the newly created port is return immediately, and it is
		605	permissible to immediately start sending messages or monitor the port.
		606
		607	After the port has been created, the init function is
		608	called. This function must be a fully-qualified function name
		609	(e.g. C<MyApp::Chat::Server::init>). To specify a function in the main
		610	program, use C<::name>.
		611
		612	If the function doesn't exist, then the node tries to C<require>
		613	the package, then the package above the package and so on (e.g.
		614	C<MyApp::Chat::Server>, C<MyApp::Chat>, C<MyApp>) until the function
		615	exists or it runs out of package names.
		616
		617	The init function is then called with the newly-created port as context
		618	object (C<$SELF>) and the C<@initdata> values as arguments.
		619
		620	A common idiom is to pass your own port, monitor the spawned port, and
		621	in the init function, monitor the original port. This two-way monitoring
		622	ensures that both ports get cleaned up when there is a problem.
		623
		624	Example: spawn a chat server port on C<$othernode>.
		625
		626	# this node, executed from within a port context:
		627	my $server = spawn $othernode, "MyApp::Chat::Server::connect", $SELF;
		628	mon $server;
		629
		630	# init function on C<$othernode>
		631	sub connect {
		632	my ($srcport) = @_;
		633
		634	mon $srcport;
		635
		636	rcv $SELF, sub {
		637	...
		638	};
		639	}
		640
		641	=cut
		642
		643	sub _spawn {
		644	my $port = shift;
		645	my $init = shift;
		646
		647	local $SELF = "$NODE#$port";
		648	eval {
		649	&{ load_func $init }
		650	};
		651	_self_die if $@;
		652	}
		653
		654	sub spawn(@) {
		655	my ($noderef, undef) = split /#/, shift, 2;
		656
		657	my $id = "$RUNIQ." . $ID++;
		658
		659	$_[0] =~ /::/
		660	or Carp::croak "spawn init function must be a fully-qualified name, caught";
		661
		662	($NODE{$noderef} \|\| add_node $noderef)
		663	->send (["", "AnyEvent::MP::_spawn" => $id, @_]);
		664
		665	"$noderef#$id"
		666	}
		667
		668	=back
		669
		670	=head1 NODE MESSAGES
		671
		672	Nodes understand the following messages sent to them. Many of them take
		673	arguments called C<@reply>, which will simply be used to compose a reply
		674	message - C<$reply[0]> is the port to reply to, C<$reply[1]> the type and
		675	the remaining arguments are simply the message data.
		676
		677	While other messages exist, they are not public and subject to change.
		678
		679	=over 4
		680
		681	=cut
		682
		683	=item lookup => $name, @reply
		684
		685	Replies with the port ID of the specified well-known port, or C<undef>.
		686
		687	=item devnull => ...
		688
		689	Generic data sink/CPU heat conversion.
		690
		691	=item relay => $port, @msg
		692
		693	Simply forwards the message to the given port.
		694
		695	=item eval => $string[ @reply]
		696
		697	Evaluates the given string. If C<@reply> is given, then a message of the
		698	form C<@reply, $@, @evalres> is sent.
		699
		700	Example: crash another node.
		701
		702	snd $othernode, eval => "exit";
		703
		704	=item time => @reply
		705
		706	Replies the the current node time to C<@reply>.
		707
		708	Example: tell the current node to send the current time to C<$myport> in a
		709	C<timereply> message.
		710
		711	snd $NODE, time => $myport, timereply => 1, 2;
		712	# => snd $myport, timereply => 1, 2, <time>
		713
		714	=back
		715
		716	=head1 AnyEvent::MP vs. Distributed Erlang
		717
		718	AnyEvent::MP got lots of its ideas from distributed Erlang (Erlang node
		719	== aemp node, Erlang process == aemp port), so many of the documents and
		720	programming techniques employed by Erlang apply to AnyEvent::MP. Here is a
		721	sample:
		722
		723	http://www.Erlang.se/doc/programming_rules.shtml
		724	http://Erlang.org/doc/getting_started/part_frame.html # chapters 3 and 4
		725	http://Erlang.org/download/Erlang-book-part1.pdf # chapters 5 and 6
		726	http://Erlang.org/download/armstrong_thesis_2003.pdf # chapters 4 and 5
		727
		728	Despite the similarities, there are also some important differences:
		729
		730	=over 4
		731
		732	=item * Node references contain the recipe on how to contact them.
		733
		734	Erlang relies on special naming and DNS to work everywhere in the
		735	same way. AEMP relies on each node knowing it's own address(es), with
		736	convenience functionality.
		737
		738	This means that AEMP requires a less tightly controlled environment at the
		739	cost of longer node references and a slightly higher management overhead.
		740
		741	=item Erlang has a "remote ports are like local ports" philosophy, AEMP
		742	uses "local ports are like remote ports".
		743
		744	The failure modes for local ports are quite different (runtime errors
		745	only) then for remote ports - when a local port dies, you I<know> it dies,
		746	when a connection to another node dies, you know nothing about the other
		747	port.
		748
		749	Erlang pretends remote ports are as reliable as local ports, even when
		750	they are not.
		751
		752	AEMP encourages a "treat remote ports differently" philosophy, with local
		753	ports being the special case/exception, where transport errors cannot
		754	occur.
		755
		756	=item * Erlang uses processes and a mailbox, AEMP does not queue.
		757
		758	Erlang uses processes that selectively receive messages, and therefore
		759	needs a queue. AEMP is event based, queuing messages would serve no
		760	useful purpose. For the same reason the pattern-matching abilities of
		761	AnyEvent::MP are more limited, as there is little need to be able to
		762	filter messages without dequeing them.
		763
		764	(But see L<Coro::MP> for a more Erlang-like process model on top of AEMP).
		765
		766	=item * Erlang sends are synchronous, AEMP sends are asynchronous.
		767
		768	Sending messages in Erlang is synchronous and blocks the process (and
		769	so does not need a queue that can overflow). AEMP sends are immediate,
		770	connection establishment is handled in the background.
		771
		772	=item * Erlang suffers from silent message loss, AEMP does not.
		773
		774	Erlang makes few guarantees on messages delivery - messages can get lost
		775	without any of the processes realising it (i.e. you send messages a, b,
		776	and c, and the other side only receives messages a and c).
		777
		778	AEMP guarantees correct ordering, and the guarantee that there are no
		779	holes in the message sequence.
		780
		781	=item * In Erlang, processes can be declared dead and later be found to be
		782	alive.
		783
		784	In Erlang it can happen that a monitored process is declared dead and
		785	linked processes get killed, but later it turns out that the process is
		786	still alive - and can receive messages.
		787
		788	In AEMP, when port monitoring detects a port as dead, then that port will
		789	eventually be killed - it cannot happen that a node detects a port as dead
		790	and then later sends messages to it, finding it is still alive.
		791
		792	=item * Erlang can send messages to the wrong port, AEMP does not.
		793
		794	In Erlang it is quite likely that a node that restarts reuses a process ID
		795	known to other nodes for a completely different process, causing messages
		796	destined for that process to end up in an unrelated process.
		797
		798	AEMP never reuses port IDs, so old messages or old port IDs floating
		799	around in the network will not be sent to an unrelated port.
		800
		801	=item * Erlang uses unprotected connections, AEMP uses secure
		802	authentication and can use TLS.
		803
		804	AEMP can use a proven protocol - SSL/TLS - to protect connections and
		805	securely authenticate nodes.
		806
		807	=item * The AEMP protocol is optimised for both text-based and binary
		808	communications.
		809
		810	The AEMP protocol, unlike the Erlang protocol, supports both
		811	language-independent text-only protocols (good for debugging) and binary,
		812	language-specific serialisers (e.g. Storable).
		813
		814	It has also been carefully designed to be implementable in other languages
		815	with a minimum of work while gracefully degrading fucntionality to make the
		816	protocol simple.
		817
		818	=item * AEMP has more flexible monitoring options than Erlang.
		819
		820	In Erlang, you can chose to receive I<all> exit signals as messages
		821	or I<none>, there is no in-between, so monitoring single processes is
		822	difficult to implement. Monitoring in AEMP is more flexible than in
		823	Erlang, as one can choose between automatic kill, exit message or callback
		824	on a per-process basis.
		825
		826	=item * Erlang tries to hide remote/local connections, AEMP does not.
		827
		828	Monitoring in Erlang is not an indicator of process death/crashes,
		829	as linking is (except linking is unreliable in Erlang).
		830
		831	In AEMP, you don't "look up" registered port names or send to named ports
		832	that might or might not be persistent. Instead, you normally spawn a port
		833	on the remote node. The init function monitors the you, and you monitor
		834	the remote port. Since both monitors are local to the node, they are much
		835	more reliable.
		836
		837	This also saves round-trips and avoids sending messages to the wrong port
		838	(hard to do in Erlang).
		839
		840	=back
		841
		842	=head1 RATIONALE
		843
		844	=over 4
		845
		846	=item Why strings for ports and noderefs, why not objects?
		847
		848	We considered "objects", but found that the actual number of methods
		849	thatc an be called are very low. Since port IDs and noderefs travel over
		850	the network frequently, the serialising/deserialising would add lots of
		851	overhead, as well as having to keep a proxy object.
		852
		853	Strings can easily be printed, easily serialised etc. and need no special
		854	procedures to be "valid".
		855
		856	And a a miniport consists of a single closure stored in a global hash - it
		857	can't become much cheaper.
		858
		859	=item Why favour JSON, why not real serialising format such as Storable?
		860
		861	In fact, any AnyEvent::MP node will happily accept Storable as framing
		862	format, but currently there is no way to make a node use Storable by
		863	default.
		864
		865	The default framing protocol is JSON because a) JSON::XS is many times
		866	faster for small messages and b) most importantly, after years of
		867	experience we found that object serialisation is causing more problems
		868	than it gains: Just like function calls, objects simply do not travel
		869	easily over the network, mostly because they will always be a copy, so you
		870	always have to re-think your design.
		871
		872	Keeping your messages simple, concentrating on data structures rather than
		873	objects, will keep your messages clean, tidy and efficient.
278		874
279	=back	875	=back
280		876
281	=head1 SEE ALSO	877	=head1 SEE ALSO
282		878

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 1.52 by root, Fri Aug 14 15:13:20 2009 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 1.52 by root, Fri Aug 14 15:13:20 2009 UTC