ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/AnyEvent-MP/MP.pm

Comparing AnyEvent-MP/MP.pm (file contents):
Revision 1.7 by root, Sat Aug 1 15:04:30 2009 UTC vs.
Revision 1.128 by root, Sun Mar 4 14:28:44 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.
31		66
32	At the moment, this module family is severly brokena nd underdocumented,	67	For an introduction to this module family, see the L<AnyEvent::MP::Intro>
33	so do not use. This was uploaded mainly to resreve the CPAN namespace -	68	manual page and the examples under F<eg/>.
34	stay tuned!
35		69
36	=head1 CONCEPTS	70	=head1 CONCEPTS
37		71
38	=over 4	72	=over 4
39		73
40	=item port	74	=item port
41		75
42	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
43	you can register C<rcv> handlers with. All C<rcv> handlers will receive	77	messages to (with the C<snd> function).
44	messages they match, messages will not be queued.
45		78
		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.
		82
		83	Ports are represented by (printable) strings called "port IDs".
		84
46	=item port id - C<noderef#portname>	85	=item port ID - C<nodeid#portname>
47		86
48	A port id is always the noderef, a hash-mark (C<#>) as separator, followed	87	A port ID is the concatenation of a node ID, a hash-mark (C<#>)
49	by a port name (a printable string of unspecified format).	88	as separator, and a port name (a printable string of unspecified
		89	format created by AnyEvent::MP).
50		90
51	=item node	91	=item node
52		92
53	A node is a single process containing at least one port - the node	93	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,	94	which enables nodes to manage each other remotely, and to create new
55	among other things.	95	ports.
56		96
57	Initially, nodes are either private (single-process only) or hidden	97	Nodes are either public (have one or more listening ports) or private
58	(connected to a master node only). Only when they epxlicitly "become	98	(no listening ports). Private nodes cannot talk to other private nodes
59	public" can you send them messages from unrelated other nodes.	99	currently, but all nodes can talk to public nodes.
60		100
61	=item noderef - C<host:port,host:port...>, C<id@noderef>, C<id>	101	Nodes is represented by (printable) strings called "node IDs".
62		102
		103	=item node ID - C<[A-Za-z0-9_\-.:]*>
		104
63	A noderef is a string that either uniquely identifies a given node (for	105	A node ID is a string that uniquely identifies the node within a
64	private and hidden nodes), or contains a recipe on how to reach a given	106	network. Depending on the configuration used, node IDs can look like a
65	node (for public nodes).	107	hostname, a hostname and a port, or a random string. AnyEvent::MP itself
		108	doesn't interpret node IDs in any way except to uniquely identify a node.
		109
		110	=item binds - C<ip:port>
		111
		112	Nodes can only talk to each other by creating some kind of connection to
		113	each other. To do this, nodes should listen on one or more local transport
		114	endpoints - binds.
		115
		116	Currently, only standard C<ip:port> specifications can be used, which
		117	specify TCP ports to listen on. So a bind is basically just a tcp socket
		118	in listening mode thta accepts conenctions form other nodes.
		119
		120	=item seed nodes
		121
		122	When a node starts, it knows nothing about the network it is in - it
		123	needs to connect to at least one other node that is already in the
		124	network. These other nodes are called "seed nodes".
		125
		126	Seed nodes themselves are not special - they are seed nodes only because
		127	some other node I<uses> them as such, but any node can be used as seed
		128	node for other nodes, and eahc node cna use a different set of seed nodes.
		129
		130	In addition to discovering the network, seed nodes are also used to
		131	maintain the network - all nodes using the same seed node form are part of
		132	the same network. If a network is split into multiple subnets because e.g.
		133	the network link between the parts goes down, then using the same seed
		134	nodes for all nodes ensures that eventually the subnets get merged again.
		135
		136	Seed nodes are expected to be long-running, and at least one seed node
		137	should always be available. They should also be relatively responsive - a
		138	seed node that blocks for long periods will slow down everybody else.
		139
		140	For small networks, it's best if every node uses the same set of seed
		141	nodes. For large networks, it can be useful to specify "regional" seed
		142	nodes for most nodes in an area, and use all seed nodes as seed nodes for
		143	each other. What's important is that all seed nodes connections form a
		144	complete graph, so that the network cannot split into separate subnets
		145	forever.
		146
		147	Seed nodes are represented by seed IDs.
		148
		149	=item seed IDs - C<host:port>
		150
		151	Seed IDs are transport endpoint(s) (usually a hostname/IP address and a
		152	TCP port) of nodes that should be used as seed nodes.
		153
		154	=item global nodes
		155
		156	An AEMP network needs a discovery service - nodes need to know how to
		157	connect to other nodes they only know by name. In addition, AEMP offers a
		158	distributed "group database", which maps group names to a list of strings
		159	- for example, to register worker ports.
		160
		161	A network needs at least one global node to work, and allows every node to
		162	be a global node.
		163
		164	Any node that loads the L<AnyEvent::MP::Global> module becomes a global
		165	node and tries to keep connections to all other nodes. So while it can
		166	make sense to make every node "global" in small networks, it usually makes
		167	sense to only make seed nodes into global nodes in large networks (nodes
		168	keep connections to seed nodes and global nodes, so makign them the same
		169	reduces overhead).
66		170
67	=back	171	=back
68		172
69	=head1 VARIABLES/FUNCTIONS	173	=head1 VARIABLES/FUNCTIONS
70		174
…		…
72		176
73	=cut	177	=cut
74		178
75	package AnyEvent::MP;	179	package AnyEvent::MP;
76		180
77	use AnyEvent::MP::Util ();	181	use AnyEvent::MP::Config ();
78	use AnyEvent::MP::Node;	182	use AnyEvent::MP::Kernel;
79	use AnyEvent::MP::Transport;	183	use AnyEvent::MP::Kernel qw(%NODE %PORT %PORT_DATA $UNIQ $RUNIQ $ID);
80		184
81	use utf8;
82	use common::sense;	185	use common::sense;
83		186
84	use Carp ();	187	use Carp ();
85		188
86	use AE ();	189	use AE ();
		190	use Guard ();
87		191
88	use base "Exporter";	192	use base "Exporter";
89		193
90	our $VERSION = '0.01';	194	our $VERSION = $AnyEvent::MP::Config::VERSION;
91	our @EXPORT = qw(NODE $NODE $PORT snd rcv _any_);
92		195
93	our $DEFAULT_SECRET;	196	our @EXPORT = qw(
94	our $DEFAULT_PORT = "4040";	197	NODE $NODE *SELF node_of after
		198	configure
		199	snd rcv mon mon_guard kil psub peval spawn cal
		200	port
		201	db_set db_del db_reg
		202	db_mon db_family db_keys db_values
		203	);
95		204
96	our $CONNECT_INTERVAL = 5; # new connect every 5s, at least	205	our $SELF;
97	our $CONNECT_TIMEOUT = 30; # includes handshake
98		206
99	sub default_secret {	207	sub _self_die() {
100	unless (defined $DEFAULT_SECRET) {	208	my $msg = $@;
101	if (open my $fh, "<$ENV{HOME}/.aemp-secret") {	209	$msg =~ s/\n+$// unless ref $msg;
102	sysread $fh, $DEFAULT_SECRET, -s $fh;	210	kil $SELF, die => $msg;
103	} else {
104	$DEFAULT_SECRET = AnyEvent::MP::Util::nonce 32;
105	}
106	}
107
108	$DEFAULT_SECRET
109	}	211	}
110		212
111	=item NODE / $NODE	213	=item $thisnode = NODE / $NODE
112		214
113	The C<NODE ()> function and the C<$NODE> variable contain the noderef of	215	The C<NODE> function returns, and the C<$NODE> variable contains, the node
114	the local node. The value is initialised by a call to C<become_public> or	216	ID of the node running in the current process. This value is initialised by
115	C<become_slave>, after which all local port identifiers become invalid.	217	a call to C<configure>.
116		218
117	=cut	219	=item $nodeid = node_of $port
118		220
119	our $UNIQ = sprintf "%x.%x", $$, time; # per-process/node unique cookie	221	Extracts and returns the node ID from a port ID or a node ID.
120	our $ID = "a0";
121	our $PUBLIC = 0;
122	our $NODE;
123	our $PORT;
124		222
125	our %NODE; # node id to transport mapping, or "undef", for local node	223	=item configure $profile, key => value...
126	our %PORT; # local ports
127	our %LISTENER; # local transports
128		224
129	sub NODE() { $NODE }	225	=item configure key => value...
130		226
131	{	227	Before a node can talk to other nodes on the network (i.e. enter
132	use POSIX ();	228	"distributed mode") it has to configure itself - the minimum a node needs
133	my $nodename = (POSIX::uname)[1];	229	to know is its own name, and optionally it should know the addresses of
134	$NODE = "$$\@$nodename";	230	some other nodes in the network to discover other nodes.
135	}
136		231
137	sub _ANY_() { 1 }	232	This function configures a node - it must be called exactly once (or
138	sub _any_() { \&_ANY_ }	233	never) before calling other AnyEvent::MP functions.
139		234
140	sub add_node {	235	The key/value pairs are basically the same ones as documented for the
141	my ($noderef) = @_;	236	F<aemp> command line utility (sans the set/del prefix), with these additions:
142		237
143	return $NODE{$noderef}	238	=over 4
144	if exists $NODE{$noderef};
145		239
146	for (split /,/, $noderef) {	240	=item norc => $boolean (default false)
147	return $NODE{$noderef} = $NODE{$_}
148	if exists $NODE{$_};
149	}
150		241
151	# for indirect sends, use a different class	242	If true, then the rc file (e.g. F<~/.perl-anyevent-mp>) will I<not>
152	my $node = new AnyEvent::MP::Node::Direct $noderef;	243	be consulted - all configuraiton options must be specified in the
		244	C<configure> call.
153		245
154	$NODE{$_} = $node	246	=item force => $boolean (default false)
155	for $noderef, split /,/, $noderef;
156		247
157	$node	248	IF true, then the values specified in the C<configure> will take
158	}	249	precedence over any values configured via the rc file. The default is for
		250	the rc file to override any options specified in the program.
159		251
		252	=item secure => $pass->($nodeid)
		253
		254	In addition to specifying a boolean, you can specify a code reference that
		255	is called for every remote execution attempt - the execution request is
		256	granted iff the callback returns a true value.
		257
		258	See F<semp setsecure> for more info.
		259
		260	=back
		261
		262	=over 4
		263
		264	=item step 1, gathering configuration from profiles
		265
		266	The function first looks up a profile in the aemp configuration (see the
		267	L<aemp> commandline utility). The profile name can be specified via the
		268	named C<profile> parameter or can simply be the first parameter). If it is
		269	missing, then the nodename (F<uname -n>) will be used as profile name.
		270
		271	The profile data is then gathered as follows:
		272
		273	First, all remaining key => value pairs (all of which are conveniently
		274	undocumented at the moment) will be interpreted as configuration
		275	data. Then they will be overwritten by any values specified in the global
		276	default configuration (see the F<aemp> utility), then the chain of
		277	profiles chosen by the profile name (and any C<parent> attributes).
		278
		279	That means that the values specified in the profile have highest priority
		280	and the values specified directly via C<configure> have lowest priority,
		281	and can only be used to specify defaults.
		282
		283	If the profile specifies a node ID, then this will become the node ID of
		284	this process. If not, then the profile name will be used as node ID, with
		285	a unique randoms tring (C</%u>) appended.
		286
		287	The node ID can contain some C<%> sequences that are expanded: C<%n>
		288	is expanded to the local nodename, C<%u> is replaced by a random
		289	strign to make the node unique. For example, the F<aemp> commandline
		290	utility uses C<aemp/%n/%u> as nodename, which might expand to
		291	C<aemp/cerebro/ZQDGSIkRhEZQDGSIkRhE>.
		292
		293	=item step 2, bind listener sockets
		294
		295	The next step is to look up the binds in the profile, followed by binding
		296	aemp protocol listeners on all binds specified (it is possible and valid
		297	to have no binds, meaning that the node cannot be contacted form the
		298	outside. This means the node cannot talk to other nodes that also have no
		299	binds, but it can still talk to all "normal" nodes).
		300
		301	If the profile does not specify a binds list, then a default of C<*> is
		302	used, meaning the node will bind on a dynamically-assigned port on every
		303	local IP address it finds.
		304
		305	=item step 3, connect to seed nodes
		306
		307	As the last step, the seed ID list from the profile is passed to the
		308	L<AnyEvent::MP::Global> module, which will then use it to keep
		309	connectivity with at least one node at any point in time.
		310
		311	=back
		312
		313	Example: become a distributed node using the local node name as profile.
		314	This should be the most common form of invocation for "daemon"-type nodes.
		315
		316	configure
		317
		318	Example: become a semi-anonymous node. This form is often used for
		319	commandline clients.
		320
		321	configure nodeid => "myscript/%n/%u";
		322
		323	Example: configure a node using a profile called seed, which is suitable
		324	for a seed node as it binds on all local addresses on a fixed port (4040,
		325	customary for aemp).
		326
		327	# use the aemp commandline utility
		328	# aemp profile seed binds '*:4040'
		329
		330	# then use it
		331	configure profile => "seed";
		332
		333	# or simply use aemp from the shell again:
		334	# aemp run profile seed
		335
		336	# or provide a nicer-to-remember nodeid
		337	# aemp run profile seed nodeid "$(hostname)"
		338
		339	=item $SELF
		340
		341	Contains the current port id while executing C<rcv> callbacks or C<psub>
		342	blocks.
		343
		344	=item *SELF, SELF, %SELF, @SELF...
		345
		346	Due to some quirks in how perl exports variables, it is impossible to
		347	just export C<$SELF>, all the symbols named C<SELF> are exported by this
		348	module, but only C<$SELF> is currently used.
		349
160	=item snd $portid, type => @data	350	=item snd $port, type => @data
161		351
162	=item snd $portid, @msg	352	=item snd $port, @msg
163		353
164	Send the given message to the given port ID, which can identify either a	354	Send the given message to the given port, which can identify either a
165	local or a remote port.	355	local or a remote port, and must be a port ID.
166		356
167	While the message can be about anything, it is highly recommended to use	357	While the message can be almost anything, it is highly recommended to
168	a constant string as first element.	358	use a string as first element (a port ID, or some word that indicates a
		359	request type etc.) and to consist if only simple perl values (scalars,
		360	arrays, hashes) - if you think you need to pass an object, think again.
169		361
170	The message data effectively becomes read-only after a call to this	362	The message data logically becomes read-only after a call to this
171	function: modifying any argument is not allowed and can cause many	363	function: modifying any argument (or values referenced by them) is
172	problems.	364	forbidden, as there can be considerable time between the call to C<snd>
		365	and the time the message is actually being serialised - in fact, it might
		366	never be copied as within the same process it is simply handed to the
		367	receiving port.
173		368
174	The type of data you can transfer depends on the transport protocol: when	369	The type of data you can transfer depends on the transport protocol: when
175	JSON is used, then only strings, numbers and arrays and hashes consisting	370	JSON is used, then only strings, numbers and arrays and hashes consisting
176	of those are allowed (no objects). When Storable is used, then anything	371	of those are allowed (no objects). When Storable is used, then anything
177	that Storable can serialise and deserialise is allowed, and for the local	372	that Storable can serialise and deserialise is allowed, and for the local
178	node, anything can be passed.	373	node, anything can be passed. Best rely only on the common denominator of
		374	these.
179		375
180	=cut	376	=item $local_port = port
181		377
182	sub snd(@) {	378	Create a new local port object and returns its port ID. Initially it has
183	my ($noderef, $port) = split /#/, shift, 2;	379	no callbacks set and will throw an error when it receives messages.
184		380
185	add_node $noderef	381	=item $local_port = port { my @msg = @_ }
186	unless exists $NODE{$noderef};
187		382
188	$NODE{$noderef}->send (["$port", [@_]]);	383	Creates a new local port, and returns its ID. Semantically the same as
		384	creating a port and calling C<rcv $port, $callback> on it.
		385
		386	The block will be called for every message received on the port, with the
		387	global variable C<$SELF> set to the port ID. Runtime errors will cause the
		388	port to be C<kil>ed. The message will be passed as-is, no extra argument
		389	(i.e. no port ID) will be passed to the callback.
		390
		391	If you want to stop/destroy the port, simply C<kil> it:
		392
		393	my $port = port {
		394	my @msg = @_;
		395	...
		396	kil $SELF;
		397	};
		398
		399	=cut
		400
		401	sub rcv($@);
		402
		403	sub _kilme {
		404	die "received message on port without callback";
189	}	405	}
190		406
		407	sub port(;&) {
		408	my $id = $UNIQ . ++$ID;
		409	my $port = "$NODE#$id";
		410
		411	rcv $port, shift || \&_kilme;
		412
		413	$port
		414	}
		415
191	=item rcv $portid, type => $callback->(@msg)	416	=item rcv $local_port, $callback->(@msg)
192		417
193	=item rcv $portid, $smartmatch => $callback->(@msg)	418	Replaces the default callback on the specified port. There is no way to
		419	remove the default callback: use C<sub { }> to disable it, or better
		420	C<kil> the port when it is no longer needed.
194		421
195	=item rcv $portid, [$smartmatch...] => $callback->(@msg)	422	The global C<$SELF> (exported by this module) contains C<$port> while
		423	executing the callback. Runtime errors during callback execution will
		424	result in the port being C<kil>ed.
196		425
197	Register a callback on the port identified by C<$portid>, which I<must> be	426	The default callback received all messages not matched by a more specific
198	a local port.	427	C<tag> match.
199		428
200	The callback has to return a true value when its work is done, after	429	=item rcv $local_port, tag => $callback->(@msg_without_tag), ...
201	which is will be removed, or a false value in which case it will stay
202	registered.
203		430
204	If the match is an array reference, then it will be matched against the	431	Register (or replace) callbacks to be called on messages starting with the
205	first elements of the message, otherwise only the first element is being	432	given tag on the given port (and return the port), or unregister it (when
206	matched.	433	C<$callback> is C<$undef> or missing). There can only be one callback
		434	registered for each tag.
207		435
208	Any element in the match that is specified as C<_any_> (a function	436	The original message will be passed to the callback, after the first
209	exported by this module) matches any single element of the message.	437	element (the tag) has been removed. The callback will use the same
		438	environment as the default callback (see above).
210		439
211	While not required, it is highly recommended that the first matching	440	Example: create a port and bind receivers on it in one go.
212	element is a string identifying the message. The one-string-only match is	441
213	also the most efficient match (by far).	442	my $port = rcv port,
		443	msg1 => sub { ... },
		444	msg2 => sub { ... },
		445	;
		446
		447	Example: create a port, bind receivers and send it in a message elsewhere
		448	in one go:
		449
		450	snd $otherport, reply =>
		451	rcv port,
		452	msg1 => sub { ... },
		453	...
		454	;
		455
		456	Example: temporarily register a rcv callback for a tag matching some port
		457	(e.g. for an rpc reply) and unregister it after a message was received.
		458
		459	rcv $port, $otherport => sub {
		460	my @reply = @_;
		461
		462	rcv $SELF, $otherport;
		463	};
214		464
215	=cut	465	=cut
216		466
217	sub rcv($@) {	467	sub rcv($@) {
218	my ($port, $match, $cb) = @_;	468	my $port = shift;
219
220	my $port = $PORT{$port}
221	or do {
222	my ($noderef, $lport) = split /#/, $port;	469	my ($nodeid, $portid) = split /#/, $port, 2;
223	"AnyEvent::MP::Node::Self" eq ref $NODE{$noderef}	470
		471	$NODE{$nodeid} == $NODE{""}
224	or Carp::croak "$port: can only rcv on local ports";	472	or Carp::croak "$port: rcv can only be called on local ports, caught";
225		473
226	$PORT{$lport}	474	while (@_) {
227	or Carp::croak "$port: port does not exist";	475	if (ref $_[0]) {
228		476	if (my $self = $PORT_DATA{$portid}) {
229	$PORT{$port} = $PORT{$lport} # also return	477	"AnyEvent::MP::Port" eq ref $self
230	};	478	or Carp::croak "$port: rcv can only be called on message matching ports, caught";
231		479
232	if (!ref $match) {	480	$self->[0] = shift;
233	push @{ $port->{rc0}{$match} }, [$cb];
234	} elsif (("ARRAY" eq ref $match && !ref $match->[0])) {
235	my ($type, @match) = @$match;
236	@match
237	? push @{ $port->{rcv}{$match->[0]} }, [$cb, \@match]
238	: push @{ $port->{rc0}{$match->[0]} }, [$cb];
239	} else {	481	} else {
240	push @{ $port->{any} }, [$cb, $match];	482	my $cb = shift;
241	}	483	$PORT{$portid} = sub {
242	}	484	local $SELF = $port;
243		485	eval { &$cb }; _self_die if $@;
244	sub _inject {
245	my ($port, $msg) = @{+shift};
246
247	$port = $PORT{$port}
248	or return;
249
250	@_ = @$msg;
251
252	for (@{ $port->{rc0}{$msg->[0]} }) {
253	$_ && &{$_->[0]}
254	&& undef $_;
255	}
256
257	for (@{ $port->{rcv}{$msg->[0]} }) {
258	$_ && [@_[1..$#{$_->[1]}]] ~~ $_->[1]
259	&& &{$_->[0]}
260	&& undef $_;
261	}
262
263	for (@{ $port->{any} }) {
264	$_ && [@_[0..$#{$_->[1]}]] ~~ $_->[1]
265	&& &{$_->[0]}
266	&& undef $_;
267	}
268	}
269
270	sub normalise_noderef($) {
271	my ($noderef) = @_;
272
273	my $cv = AE::cv;
274	my @res;
275
276	$cv->begin (sub {
277	my %seen;
278	my @refs;
279	for (sort { $a->[0] <=> $b->[0] } @res) {
280	push @refs, $_->[1] unless $seen{$_->[1]}++
281	}
282	shift->send (join ",", @refs);
283	});
284
285	$noderef = $DEFAULT_PORT unless length $noderef;
286
287	my $idx;
288	for my $t (split /,/, $noderef) {
289	my $pri = ++$idx;
290
291	#TODO: this should be outside normalise_noderef and in become_public
292	if ($t =~ /^\d*$/) {
293	my $nodename = (POSIX::uname)[1];
294
295	$cv->begin;
296	AnyEvent::Socket::resolve_sockaddr $nodename, $t || "aemp=$DEFAULT_PORT", "tcp", 0, undef, sub {
297	for (@_) {
298	my ($service, $host) = AnyEvent::Socket::unpack_sockaddr $_->[3];
299	push @res, [
300	$pri += 1e-5,
301	AnyEvent::Socket::format_hostport AnyEvent::Socket::format_address $host, $service
302	];
303	}	486	};
304	$cv->end;	487	}
		488	} elsif (defined $_[0]) {
		489	my $self = $PORT_DATA{$portid} ||= do {
		490	my $self = bless [$PORT{$portid} || sub { }, { }, $port], "AnyEvent::MP::Port";
		491
		492	$PORT{$portid} = sub {
		493	local $SELF = $port;
		494
		495	if (my $cb = $self->[1]{$_[0]}) {
		496	shift;
		497	eval { &$cb }; _self_die if $@;
		498	} else {
		499	&{ $self->[0] };
		500	}
		501	};
		502
		503	$self
305	};	504	};
306		505
307	# my (undef, undef, undef, undef, @ipv4) = gethostbyname $nodename;	506	"AnyEvent::MP::Port" eq ref $self
308	#	507	or Carp::croak "$port: rcv can only be called on message matching ports, caught";
309	# for (@ipv4) {	508
310	# push @res, [	509	my ($tag, $cb) = splice @_, 0, 2;
311	# $pri,	510
312	# AnyEvent::Socket::format_hostport AnyEvent::Socket::format_address $_, $t || $DEFAULT_PORT,	511	if (defined $cb) {
313	# ];	512	$self->[1]{$tag} = $cb;
314	# }
315	} else {	513	} else {
316	my ($host, $port) = AnyEvent::Socket::parse_hostport $t, "aemp=$DEFAULT_PORT"	514	delete $self->[1]{$tag};
317	or Carp::croak "$t: unparsable transport descriptor";
318
319	$cv->begin;
320	AnyEvent::Socket::resolve_sockaddr $host, $port, "tcp", 0, undef, sub {
321	for (@_) {
322	my ($service, $host) = AnyEvent::Socket::unpack_sockaddr $_->[3];
323	push @res, [
324	$pri += 1e-5,
325	AnyEvent::Socket::format_hostport AnyEvent::Socket::format_address $host, $service
326	];
327	}
328	$cv->end;
329	}	515	}
330	}	516	}
331	}	517	}
332		518
333	$cv->end;	519	$port
334
335	$cv
336	}	520	}
337		521
338	sub become_public {	522	=item peval $port, $coderef[, @args]
339	return if $PUBLIC;
340		523
341	my $noderef = join ",", ref $_[0] ? @{+shift} : shift;	524	Evaluates the given C<$codref> within the contetx of C<$port>, that is,
342	my @args = @_;	525	when the code throews an exception the C<$port> will be killed.
343		526
344	$NODE = (normalise_noderef $noderef)->recv;	527	Any remaining args will be passed to the callback. Any return values will
		528	be returned to the caller.
345		529
346	for my $t (split /,/, $NODE) {	530	This is useful when you temporarily want to execute code in the context of
347	$NODE{$t} = $NODE{""};	531	a port.
348		532
349	my ($host, $port) = AnyEvent::Socket::parse_hostport $t;	533	Example: create a port and run some initialisation code in it's context.
350		534
351	$LISTENER{$t} = AnyEvent::MP::Transport::mp_server $host, $port,	535	my $port = port { ... };
352	@args,
353	on_error => sub {
354	die "on_error<@_>\n";#d#
355	},
356	on_connect => sub {
357	my ($tp) = @_;
358		536
359	$NODE{$tp->{remote_id}} = $_[0];	537	peval $port, sub {
360	},	538	init
361	sub {	539	or die "unable to init";
362	my ($tp) = @_;	540	};
363		541
364	$NODE{"$tp->{peerhost}:$tp->{peerport}"} = $tp;	542	=cut
365	},	543
366	;	544	sub peval($$) {
		545	local $SELF = shift;
		546	my $cb = shift;
		547
		548	if (wantarray) {
		549	my @res = eval { &$cb };
		550	_self_die if $@;
		551	@res
		552	} else {
		553	my $res = eval { &$cb };
		554	_self_die if $@;
		555	$res
367	}	556	}
368
369	$PUBLIC = 1;
370	}	557	}
371		558
372	=back	559	=item $closure = psub { BLOCK }
373		560
374	=head1 NODE MESSAGES	561	Remembers C<$SELF> and creates a closure out of the BLOCK. When the
		562	closure is executed, sets up the environment in the same way as in C<rcv>
		563	callbacks, i.e. runtime errors will cause the port to get C<kil>ed.
375		564
376	Nodes understand the following messages sent to them. Many of them take	565	The effect is basically as if it returned C<< sub { peval $SELF, sub {
377	arguments called C<@reply>, which will simply be used to compose a reply	566	BLOCK }, @_ } >>.
378	message - C<$reply[0]> is the port to reply to, C<$reply[1]> the type and
379	the remaining arguments are simply the message data.
380		567
381	=over 4	568	This is useful when you register callbacks from C<rcv> callbacks:
382		569
383	=cut	570	rcv delayed_reply => sub {
		571	my ($delay, @reply) = @_;
		572	my $timer = AE::timer $delay, 0, psub {
		573	snd @reply, $SELF;
		574	};
		575	};
384		576
385	#############################################################################	577	=cut
386	# self node code
387		578
388	sub _new_port($) {	579	sub psub(&) {
389	my ($name) = @_;	580	my $cb = shift;
390		581
		582	my $port = $SELF
		583	or Carp::croak "psub can only be called from within rcv or psub callbacks, not";
		584
		585	sub {
		586	local $SELF = $port;
		587
		588	if (wantarray) {
		589	my @res = eval { &$cb };
		590	_self_die if $@;
		591	@res
		592	} else {
		593	my $res = eval { &$cb };
		594	_self_die if $@;
		595	$res
		596	}
		597	}
		598	}
		599
		600	=item $guard = mon $port, $cb->(@reason) # call $cb when $port dies
		601
		602	=item $guard = mon $port, $rcvport # kill $rcvport when $port dies
		603
		604	=item $guard = mon $port # kill $SELF when $port dies
		605
		606	=item $guard = mon $port, $rcvport, @msg # send a message when $port dies
		607
		608	Monitor the given port and do something when the port is killed or
		609	messages to it were lost, and optionally return a guard that can be used
		610	to stop monitoring again.
		611
		612	In the first form (callback), the callback is simply called with any
		613	number of C<@reason> elements (no @reason means that the port was deleted
		614	"normally"). Note also that I<< the callback B<must> never die >>, so use
		615	C<eval> if unsure.
		616
		617	In the second form (another port given), the other port (C<$rcvport>)
		618	will be C<kil>'ed with C<@reason>, if a @reason was specified, i.e. on
		619	"normal" kils nothing happens, while under all other conditions, the other
		620	port is killed with the same reason.
		621
		622	The third form (kill self) is the same as the second form, except that
		623	C<$rvport> defaults to C<$SELF>.
		624
		625	In the last form (message), a message of the form C<@msg, @reason> will be
		626	C<snd>.
		627
		628	Monitoring-actions are one-shot: once messages are lost (and a monitoring
		629	alert was raised), they are removed and will not trigger again.
		630
		631	As a rule of thumb, monitoring requests should always monitor a port from
		632	a local port (or callback). The reason is that kill messages might get
		633	lost, just like any other message. Another less obvious reason is that
		634	even monitoring requests can get lost (for example, when the connection
		635	to the other node goes down permanently). When monitoring a port locally
		636	these problems do not exist.
		637
		638	C<mon> effectively guarantees that, in the absence of hardware failures,
		639	after starting the monitor, either all messages sent to the port will
		640	arrive, or the monitoring action will be invoked after possible message
		641	loss has been detected. No messages will be lost "in between" (after
		642	the first lost message no further messages will be received by the
		643	port). After the monitoring action was invoked, further messages might get
		644	delivered again.
		645
		646	Inter-host-connection timeouts and monitoring depend on the transport
		647	used. The only transport currently implemented is TCP, and AnyEvent::MP
		648	relies on TCP to detect node-downs (this can take 10-15 minutes on a
		649	non-idle connection, and usually around two hours for idle connections).
		650
		651	This means that monitoring is good for program errors and cleaning up
		652	stuff eventually, but they are no replacement for a timeout when you need
		653	to ensure some maximum latency.
		654
		655	Example: call a given callback when C<$port> is killed.
		656
		657	mon $port, sub { warn "port died because of <@_>\n" };
		658
		659	Example: kill ourselves when C<$port> is killed abnormally.
		660
		661	mon $port;
		662
		663	Example: send us a restart message when another C<$port> is killed.
		664
		665	mon $port, $self => "restart";
		666
		667	=cut
		668
		669	sub mon {
391	my ($noderef, $portname) = split /#/, $name;	670	my ($nodeid, $port) = split /#/, shift, 2;
392		671
393	$PORT{$name} =	672	my $node = $NODE{$nodeid} || add_node $nodeid;
394	$PORT{$portname} = {	673
395	names => [$name, $portname],	674	my $cb = @_ ? shift : $SELF || Carp::croak 'mon: called with one argument only, but $SELF not set,';
		675
		676	unless (ref $cb) {
		677	if (@_) {
		678	# send a kill info message
		679	my (@msg) = ($cb, @_);
		680	$cb = sub { snd @msg, @_ };
		681	} else {
		682	# simply kill other port
		683	my $port = $cb;
		684	$cb = sub { kil $port, @_ if @_ };
		685	}
		686	}
		687
		688	$node->monitor ($port, $cb);
		689
		690	defined wantarray
		691	and ($cb += 0, Guard::guard { $node->unmonitor ($port, $cb) })
		692	}
		693
		694	=item $guard = mon_guard $port, $ref, $ref...
		695
		696	Monitors the given C<$port> and keeps the passed references. When the port
		697	is killed, the references will be freed.
		698
		699	Optionally returns a guard that will stop the monitoring.
		700
		701	This function is useful when you create e.g. timers or other watchers and
		702	want to free them when the port gets killed (note the use of C<psub>):
		703
		704	$port->rcv (start => sub {
		705	my $timer; $timer = mon_guard $port, AE::timer 1, 1, psub {
		706	undef $timer if 0.9 < rand;
		707	});
		708	});
		709
		710	=cut
		711
		712	sub mon_guard {
		713	my ($port, @refs) = @_;
		714
		715	#TODO: mon-less form?
		716
		717	mon $port, sub { 0 && @refs }
		718	}
		719
		720	=item kil $port[, @reason]
		721
		722	Kill the specified port with the given C<@reason>.
		723
		724	If no C<@reason> is specified, then the port is killed "normally" -
		725	monitor callback will be invoked, but the kil will not cause linked ports
		726	(C<mon $mport, $lport> form) to get killed.
		727
		728	If a C<@reason> is specified, then linked ports (C<mon $mport, $lport>
		729	form) get killed with the same reason.
		730
		731	Runtime errors while evaluating C<rcv> callbacks or inside C<psub> blocks
		732	will be reported as reason C<< die => $@ >>.
		733
		734	Transport/communication errors are reported as C<< transport_error =>
		735	$message >>.
		736
		737	=cut
		738
		739	=item $port = spawn $node, $initfunc[, @initdata]
		740
		741	Creates a port on the node C<$node> (which can also be a port ID, in which
		742	case it's the node where that port resides).
		743
		744	The port ID of the newly created port is returned immediately, and it is
		745	possible to immediately start sending messages or to monitor the port.
		746
		747	After the port has been created, the init function is called on the remote
		748	node, in the same context as a C<rcv> callback. This function must be a
		749	fully-qualified function name (e.g. C<MyApp::Chat::Server::init>). To
		750	specify a function in the main program, use C<::name>.
		751
		752	If the function doesn't exist, then the node tries to C<require>
		753	the package, then the package above the package and so on (e.g.
		754	C<MyApp::Chat::Server>, C<MyApp::Chat>, C<MyApp>) until the function
		755	exists or it runs out of package names.
		756
		757	The init function is then called with the newly-created port as context
		758	object (C<$SELF>) and the C<@initdata> values as arguments. It I<must>
		759	call one of the C<rcv> functions to set callbacks on C<$SELF>, otherwise
		760	the port might not get created.
		761
		762	A common idiom is to pass a local port, immediately monitor the spawned
		763	port, and in the remote init function, immediately monitor the passed
		764	local port. This two-way monitoring ensures that both ports get cleaned up
		765	when there is a problem.
		766
		767	C<spawn> guarantees that the C<$initfunc> has no visible effects on the
		768	caller before C<spawn> returns (by delaying invocation when spawn is
		769	called for the local node).
		770
		771	Example: spawn a chat server port on C<$othernode>.
		772
		773	# this node, executed from within a port context:
		774	my $server = spawn $othernode, "MyApp::Chat::Server::connect", $SELF;
		775	mon $server;
		776
		777	# init function on C<$othernode>
		778	sub connect {
		779	my ($srcport) = @_;
		780
		781	mon $srcport;
		782
		783	rcv $SELF, sub {
		784	...
		785	};
		786	}
		787
		788	=cut
		789
		790	sub _spawn {
		791	my $port = shift;
		792	my $init = shift;
		793
		794	# rcv will create the actual port
		795	local $SELF = "$NODE#$port";
		796	eval {
		797	&{ load_func $init }
		798	};
		799	_self_die if $@;
		800	}
		801
		802	sub spawn(@) {
		803	my ($nodeid, undef) = split /#/, shift, 2;
		804
		805	my $id = $RUNIQ . ++$ID;
		806
		807	$_[0] =~ /::/
		808	or Carp::croak "spawn init function must be a fully-qualified name, caught";
		809
		810	snd_to_func $nodeid, "AnyEvent::MP::_spawn" => $id, @_;
		811
		812	"$nodeid#$id"
		813	}
		814
		815
		816	=item after $timeout, @msg
		817
		818	=item after $timeout, $callback
		819
		820	Either sends the given message, or call the given callback, after the
		821	specified number of seconds.
		822
		823	This is simply a utility function that comes in handy at times - the
		824	AnyEvent::MP author is not convinced of the wisdom of having it, though,
		825	so it may go away in the future.
		826
		827	=cut
		828
		829	sub after($@) {
		830	my ($timeout, @action) = @_;
		831
		832	my $t; $t = AE::timer $timeout, 0, sub {
		833	undef $t;
		834	ref $action[0]
		835	? $action[0]()
		836	: snd @action;
396	};	837	};
397	}	838	}
398		839
399	$NODE{""} = new AnyEvent::MP::Node::Self noderef => $NODE;	840	=item cal $port, @msg, $callback[, $timeout]
400	_new_port "";
401		841
402	=item devnull => ...	842	A simple form of RPC - sends a message to the given C<$port> with the
		843	given contents (C<@msg>), but adds a reply port to the message.
403		844
404	Generic data sink/CPU heat conversion.	845	The reply port is created temporarily just for the purpose of receiving
		846	the reply, and will be C<kil>ed when no longer needed.
405		847
406	=cut	848	A reply message sent to the port is passed to the C<$callback> as-is.
407		849
408	rcv "", devnull => sub { () };	850	If an optional time-out (in seconds) is given and it is not C<undef>,
		851	then the callback will be called without any arguments after the time-out
		852	elapsed and the port is C<kil>ed.
409		853
410	=item relay => $port, @msg	854	If no time-out is given (or it is C<undef>), then the local port will
		855	monitor the remote port instead, so it eventually gets cleaned-up.
411		856
412	Simply forwards the message to the given port.	857	Currently this function returns the temporary port, but this "feature"
		858	might go in future versions unless you can make a convincing case that
		859	this is indeed useful for something.
413		860
414	=cut	861	=cut
415		862
416	rcv "", relay => sub { \&snd; () };	863	sub cal(@) {
		864	my $timeout = ref $_[-1] ? undef : pop;
		865	my $cb = pop;
417		866
418	=item eval => $string[ @reply]	867	my $port = port {
		868	undef $timeout;
		869	kil $SELF;
		870	&$cb;
		871	};
419		872
420	Evaluates the given string. If C<@reply> is given, then a message of the	873	if (defined $timeout) {
421	form C<@reply, $@, @evalres> is sent.	874	$timeout = AE::timer $timeout, 0, sub {
		875	undef $timeout;
		876	kil $port;
		877	$cb->();
		878	};
		879	} else {
		880	mon $_[0], sub {
		881	kil $port;
		882	$cb->();
		883	};
		884	}
422		885
423	Example: crash another node.	886	push @_, $port;
		887	&snd;
424		888
425	snd $othernode, eval => "exit";	889	$port
426		890	}
427	=cut
428
429	rcv "", eval => sub {
430	my (undef, $string, @reply) = @_;
431	my @res = eval $string;
432	snd @reply, "$@", @res if @reply;
433	()
434	};
435
436	=item time => @reply
437
438	Replies the the current node time to C<@reply>.
439
440	Example: tell the current node to send the current time to C<$myport> in a
441	C<timereply> message.
442
443	snd $NODE, time => $myport, timereply => 1, 2;
444	# => snd $myport, timereply => 1, 2, <time>
445
446	=cut
447
448	rcv "", time => sub { shift; snd @_, AE::time; () };
449		891
450	=back	892	=back
451		893
		894	=head1 DISTRIBUTED DATABASE
		895
		896	AnyEvent::MP comes with a simple distributed database. The database will
		897	be mirrored asynchronously at all global nodes. Other nodes bind to one of
		898	the global nodes for their needs.
		899
		900	The database consists of a two-level hash - a hash contains a hash which
		901	contains values.
		902
		903	The top level hash key is called "family", and the second-level hash key
		904	is called "subkey" or simply "key".
		905
		906	The family must be alphanumeric, i.e. start with a letter and consist
		907	of letters, digits, underscores and colons (C<[A-Za-z][A-Za-z0-9_:]*>,
		908	pretty much like Perl module names.
		909
		910	As the family namespace is global, it is recommended to prefix family names
		911	with the name of the application or module using it.
		912
		913	The subkeys must be non-empty strings, with no further restrictions.
		914
		915	The values should preferably be strings, but other perl scalars should
		916	work as well (such as undef, arrays and hashes).
		917
		918	Every database entry is owned by one node - adding the same family/subkey
		919	combination on multiple nodes will not cause discomfort for AnyEvent::MP,
		920	but the result might be nondeterministic, i.e. the key might have
		921	different values on different nodes.
		922
		923	Different subkeys in the same family can be owned by different nodes
		924	without problems, and in fact, this is the common method to create worker
		925	pools. For example, a worker port for image scaling might do this:
		926
		927	db_set my_image_scalers => $port;
		928
		929	And clients looking for an image scaler will want to get the
		930	C<my_image_scalers> keys:
		931
		932	db_keys "my_image_scalers" => 60 => sub {
		933	#d##TODO#
		934
		935	=over
		936
		937	=item db_set $family => $subkey [=> $value]
		938
		939	Sets (or replaces) a key to the database - if C<$value> is omitted,
		940	C<undef> is used instead.
		941
		942	=item db_del $family => $subkey
		943
		944	Deletes a key from the database.
		945
		946	=item $guard = db_reg $family => $subkey [=> $value]
		947
		948	Sets the key on the database and returns a guard. When the guard is
		949	destroyed, the key is deleted from the database. If C<$value> is missing,
		950	then C<undef> is used.
		951
		952	=item $guard = db_mon $family => $cb->($familyhash, \@subkeys...)
		953
		954	Creates a monitor on the given database family. Each time a key is set or
		955	or is deleted the callback is called with a hash containing the database
		956	family and an arrayref with subkeys that have changed.
		957
		958	Specifically, if one of the passed subkeys exists in the $familyhash, then
		959	it is currently set to the value in the $familyhash. Otherwise, it has
		960	been deleted.
		961
		962	The first call will be with the current contents of the family and all
		963	keys, as if they were just added.
		964
		965	It is possible that the callback is called with a change event even though
		966	the subkey is already present and the value has not changed.
		967
		968	The monitoring stops when the guard object is destroyed.
		969
		970	Example: on every change to the family "mygroup", print out all keys.
		971
		972	my $guard = db_mon mygroup => sub {
		973	my ($family, $keys) = @_;
		974	print "mygroup members: ", (join " ", keys %$family), "\n";
		975	};
		976
		977	Exmaple: wait until the family "My::Module::workers" is non-empty.
		978
		979	my $guard; $guard = db_mon My::Module::workers => sub {
		980	my ($family, $keys) = @_;
		981	return unless %$family;
		982	undef $guard;
		983	print "My::Module::workers now nonempty\n";
		984	};
		985
		986	Example: print all changes to the family "AnyRvent::Fantasy::Module".
		987
		988	my $guard = db_mon AnyRvent::Fantasy::Module => sub {
		989	my ($family, $keys) = @_;
		990
		991	for (@$keys) {
		992	print "$_: ",
		993	(exists $family->{$_}
		994	? $family->{$_}
		995	: "(deleted)"),
		996	"\n";
		997	}
		998	};
		999
		1000	=cut
		1001
		1002	=back
		1003
		1004	=head1 AnyEvent::MP vs. Distributed Erlang
		1005
		1006	AnyEvent::MP got lots of its ideas from distributed Erlang (Erlang node
		1007	== aemp node, Erlang process == aemp port), so many of the documents and
		1008	programming techniques employed by Erlang apply to AnyEvent::MP. Here is a
		1009	sample:
		1010
		1011	http://www.erlang.se/doc/programming_rules.shtml
		1012	http://erlang.org/doc/getting_started/part_frame.html # chapters 3 and 4
		1013	http://erlang.org/download/erlang-book-part1.pdf # chapters 5 and 6
		1014	http://erlang.org/download/armstrong_thesis_2003.pdf # chapters 4 and 5
		1015
		1016	Despite the similarities, there are also some important differences:
		1017
		1018	=over 4
		1019
		1020	=item * Node IDs are arbitrary strings in AEMP.
		1021
		1022	Erlang relies on special naming and DNS to work everywhere in the same
		1023	way. AEMP relies on each node somehow knowing its own address(es) (e.g. by
		1024	configuration or DNS), and possibly the addresses of some seed nodes, but
		1025	will otherwise discover other nodes (and their IDs) itself.
		1026
		1027	=item * Erlang has a "remote ports are like local ports" philosophy, AEMP
		1028	uses "local ports are like remote ports".
		1029
		1030	The failure modes for local ports are quite different (runtime errors
		1031	only) then for remote ports - when a local port dies, you I<know> it dies,
		1032	when a connection to another node dies, you know nothing about the other
		1033	port.
		1034
		1035	Erlang pretends remote ports are as reliable as local ports, even when
		1036	they are not.
		1037
		1038	AEMP encourages a "treat remote ports differently" philosophy, with local
		1039	ports being the special case/exception, where transport errors cannot
		1040	occur.
		1041
		1042	=item * Erlang uses processes and a mailbox, AEMP does not queue.
		1043
		1044	Erlang uses processes that selectively receive messages out of order, and
		1045	therefore needs a queue. AEMP is event based, queuing messages would serve
		1046	no useful purpose. For the same reason the pattern-matching abilities
		1047	of AnyEvent::MP are more limited, as there is little need to be able to
		1048	filter messages without dequeuing them.
		1049
		1050	This is not a philosophical difference, but simply stems from AnyEvent::MP
		1051	being event-based, while Erlang is process-based.
		1052
		1053	You cna have a look at L<Coro::MP> for a more Erlang-like process model on
		1054	top of AEMP and Coro threads.
		1055
		1056	=item * Erlang sends are synchronous, AEMP sends are asynchronous.
		1057
		1058	Sending messages in Erlang is synchronous and blocks the process until
		1059	a conenction has been established and the message sent (and so does not
		1060	need a queue that can overflow). AEMP sends return immediately, connection
		1061	establishment is handled in the background.
		1062
		1063	=item * Erlang suffers from silent message loss, AEMP does not.
		1064
		1065	Erlang implements few guarantees on messages delivery - messages can get
		1066	lost without any of the processes realising it (i.e. you send messages a,
		1067	b, and c, and the other side only receives messages a and c).
		1068
		1069	AEMP guarantees (modulo hardware errors) correct ordering, and the
		1070	guarantee that after one message is lost, all following ones sent to the
		1071	same port are lost as well, until monitoring raises an error, so there are
		1072	no silent "holes" in the message sequence.
		1073
		1074	If you want your software to be very reliable, you have to cope with
		1075	corrupted and even out-of-order messages in both Erlang and AEMP. AEMP
		1076	simply tries to work better in common error cases, such as when a network
		1077	link goes down.
		1078
		1079	=item * Erlang can send messages to the wrong port, AEMP does not.
		1080
		1081	In Erlang it is quite likely that a node that restarts reuses an Erlang
		1082	process ID known to other nodes for a completely different process,
		1083	causing messages destined for that process to end up in an unrelated
		1084	process.
		1085
		1086	AEMP does not reuse port IDs, so old messages or old port IDs floating
		1087	around in the network will not be sent to an unrelated port.
		1088
		1089	=item * Erlang uses unprotected connections, AEMP uses secure
		1090	authentication and can use TLS.
		1091
		1092	AEMP can use a proven protocol - TLS - to protect connections and
		1093	securely authenticate nodes.
		1094
		1095	=item * The AEMP protocol is optimised for both text-based and binary
		1096	communications.
		1097
		1098	The AEMP protocol, unlike the Erlang protocol, supports both programming
		1099	language independent text-only protocols (good for debugging), and binary,
		1100	language-specific serialisers (e.g. Storable). By default, unless TLS is
		1101	used, the protocol is actually completely text-based.
		1102
		1103	It has also been carefully designed to be implementable in other languages
		1104	with a minimum of work while gracefully degrading functionality to make the
		1105	protocol simple.
		1106
		1107	=item * AEMP has more flexible monitoring options than Erlang.
		1108
		1109	In Erlang, you can chose to receive I<all> exit signals as messages or
		1110	I<none>, there is no in-between, so monitoring single Erlang processes is
		1111	difficult to implement.
		1112
		1113	Monitoring in AEMP is more flexible than in Erlang, as one can choose
		1114	between automatic kill, exit message or callback on a per-port basis.
		1115
		1116	=item * Erlang tries to hide remote/local connections, AEMP does not.
		1117
		1118	Monitoring in Erlang is not an indicator of process death/crashes, in the
		1119	same way as linking is (except linking is unreliable in Erlang).
		1120
		1121	In AEMP, you don't "look up" registered port names or send to named ports
		1122	that might or might not be persistent. Instead, you normally spawn a port
		1123	on the remote node. The init function monitors you, and you monitor the
		1124	remote port. Since both monitors are local to the node, they are much more
		1125	reliable (no need for C<spawn_link>).
		1126
		1127	This also saves round-trips and avoids sending messages to the wrong port
		1128	(hard to do in Erlang).
		1129
		1130	=back
		1131
		1132	=head1 RATIONALE
		1133
		1134	=over 4
		1135
		1136	=item Why strings for port and node IDs, why not objects?
		1137
		1138	We considered "objects", but found that the actual number of methods
		1139	that can be called are quite low. Since port and node IDs travel over
		1140	the network frequently, the serialising/deserialising would add lots of
		1141	overhead, as well as having to keep a proxy object everywhere.
		1142
		1143	Strings can easily be printed, easily serialised etc. and need no special
		1144	procedures to be "valid".
		1145
		1146	And as a result, a port with just a default receiver consists of a single
		1147	code reference stored in a global hash - it can't become much cheaper.
		1148
		1149	=item Why favour JSON, why not a real serialising format such as Storable?
		1150
		1151	In fact, any AnyEvent::MP node will happily accept Storable as framing
		1152	format, but currently there is no way to make a node use Storable by
		1153	default (although all nodes will accept it).
		1154
		1155	The default framing protocol is JSON because a) JSON::XS is many times
		1156	faster for small messages and b) most importantly, after years of
		1157	experience we found that object serialisation is causing more problems
		1158	than it solves: Just like function calls, objects simply do not travel
		1159	easily over the network, mostly because they will always be a copy, so you
		1160	always have to re-think your design.
		1161
		1162	Keeping your messages simple, concentrating on data structures rather than
		1163	objects, will keep your messages clean, tidy and efficient.
		1164
		1165	=back
		1166
452	=head1 SEE ALSO	1167	=head1 SEE ALSO
		1168
		1169	L<AnyEvent::MP::Intro> - a gentle introduction.
		1170
		1171	L<AnyEvent::MP::Kernel> - more, lower-level, stuff.
		1172
		1173	L<AnyEvent::MP::Global> - network maintenance and port groups, to find
		1174	your applications.
		1175
		1176	L<AnyEvent::MP::DataConn> - establish data connections between nodes.
		1177
		1178	L<AnyEvent::MP::LogCatcher> - simple service to display log messages from
		1179	all nodes.
453		1180
454	L<AnyEvent>.	1181	L<AnyEvent>.
455		1182
456	=head1 AUTHOR	1183	=head1 AUTHOR
457		1184

Diff Legend

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