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

Comparing AnyEvent-MP/MP.pm (file contents):
Revision 1.37 by root, Fri Aug 7 16:47:23 2009 UTC vs.
Revision 1.126 by root, Sat Mar 3 19:43:41 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 # contains this node's noderef	9	$NODE # contains this node's node ID
10	NODE # returns this node's noderef	10	NODE # returns this node's node ID
11	NODE $port # returns the noderef of the port
12		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
13	snd $port, type => data...;	20	snd $port, type => data...;
		21	snd $port, @msg;
		22	snd @msg_with_first_element_being_a_port;
14		23
15	$SELF # receiving/own port id in rcv callbacks	24	# creating/using ports, the simple way
		25	my $simple_port = port { my @msg = @_ };
16		26
17	rcv $port, smartmatch => $cb->($port, @msg);	27	# creating/using ports, tagged message matching
18		28	my $port = port;
19	# examples:
20	rcv $port2, ping => sub { snd $_[0], "pong"; 0 };	29	rcv $port, ping => sub { snd $_[0], "pong" };
21	rcv $port1, pong => sub { warn "pong received\n" };	30	rcv $port, pong => sub { warn "pong received\n" };
22	snd $port2, ping => $port1;
23		31
24	# more, smarter, matches (_any_ is exported by this module)	32	# create a port on another node
25	rcv $port, [child_died => $pid] => sub { ...	33	my $port = spawn $node, $initfunc, @initdata;
26	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
27		38
28	# monitoring	39	# monitoring
29	mon $port, $cb->(@msg) # callback is invoked on death	40	mon $localport, $cb->(@msg) # callback is invoked on death
30	mon $port, $otherport # kill otherport on abnormal death	41	mon $localport, $otherport # kill otherport on abnormal death
31	mon $port, $otherport, @msg # send message on 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.
32		59
33	=head1 DESCRIPTION	60	=head1 DESCRIPTION
34		61
35	This module (-family) implements a simple message passing framework.	62	This module (-family) implements a simple message passing framework.
36		63
37	Despite its simplicity, you can securely message other processes running	64	Despite its simplicity, you can securely message other processes running
38	on the same or other hosts.	65	on the same or other hosts, and you can supervise entities remotely.
39		66
40	For an introduction to this module family, see the L<AnyEvent::MP::Intro>	67	For an introduction to this module family, see the L<AnyEvent::MP::Intro>
41	manual page.	68	manual page and the examples under F<eg/>.
42
43	At the moment, this module family is severly broken and underdocumented,
44	so do not use. This was uploaded mainly to reserve the CPAN namespace -
45	stay tuned! The basic API should be finished, however.
46		69
47	=head1 CONCEPTS	70	=head1 CONCEPTS
48		71
49	=over 4	72	=over 4
50		73
51	=item port	74	=item port
52		75
53	A port is something you can send messages to (with the C<snd> function).	76	Not to be confused with a TCP port, a "port" is something you can send
		77	messages to (with the C<snd> function).
54		78
55	Some ports allow you to register C<rcv> handlers that can match specific	79	Ports allow you to register C<rcv> handlers that can match all or just
56	messages. All C<rcv> handlers will receive messages they match, messages	80	some messages. Messages send to ports will not be queued, regardless of
57	will not be queued.	81	anything was listening for them or not.
58		82
		83	Ports are represented by (printable) strings called "port IDs".
		84
59	=item port id - C<noderef#portname>	85	=item port ID - C<nodeid#portname>
60		86
61	A port id is normaly the concatenation of a noderef, a hash-mark (C<#>) as	87	A port ID is the concatenation of a node ID, a hash-mark (C<#>)
62	separator, and a port name (a printable string of unspecified format). An	88	as separator, and a port name (a printable string of unspecified
63	exception is the the node port, whose ID is identical to its node	89	format created by AnyEvent::MP).
64	reference.
65		90
66	=item node	91	=item node
67		92
68	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,
69	port. You can send messages to node ports to find existing ports or to	94	which enables nodes to manage each other remotely, and to create new
70	create new ports, among other things.	95	ports.
71		96
72	Nodes are either private (single-process only), slaves (connected to a	97	Nodes are either public (have one or more listening ports) or private
73	master node only) or public nodes (connectable from unrelated nodes).	98	(no listening ports). Private nodes cannot talk to other private nodes
		99	currently, but all nodes can talk to public nodes.
74		100
75	=item noderef - C<host:port,host:port...>, C<id@noderef>, C<id>	101	Nodes is represented by (printable) strings called "node IDs".
76		102
77	A node reference is a string that either simply identifies the node (for	103	=item node ID - C<[A-Za-z0-9_\-.:]*>
78	private and slave nodes), or contains a recipe on how to reach a given
79	node (for public nodes).
80		104
81	This recipe is simply a comma-separated list of C<address:port> pairs (for	105	A node ID is a string that uniquely identifies the node within a
82	TCP/IP, other protocols might look different).	106	network. Depending on the configuration used, node IDs can look like a
		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.
83		109
84	Node references come in two flavours: resolved (containing only numerical	110	=item binds - C<ip:port>
85	addresses) or unresolved (where hostnames are used instead of addresses).
86		111
87	Before using an unresolved node reference in a message you first have to	112	Nodes can only talk to each other by creating some kind of connection to
88	resolve it.	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).
89		170
90	=back	171	=back
91		172
92	=head1 VARIABLES/FUNCTIONS	173	=head1 VARIABLES/FUNCTIONS
93		174
95		176
96	=cut	177	=cut
97		178
98	package AnyEvent::MP;	179	package AnyEvent::MP;
99		180
		181	use AnyEvent::MP::Config ();
100	use AnyEvent::MP::Base;	182	use AnyEvent::MP::Kernel;
		183	use AnyEvent::MP::Kernel qw(%NODE %PORT %PORT_DATA $UNIQ $RUNIQ $ID);
101		184
102	use common::sense;	185	use common::sense;
103		186
104	use Carp ();	187	use Carp ();
105		188
106	use AE ();	189	use AE ();
		190	use Guard ();
107		191
108	use base "Exporter";	192	use base "Exporter";
109		193
110	our $VERSION = '0.1';	194	our $VERSION = $AnyEvent::MP::Config::VERSION;
		195
111	our @EXPORT = qw(	196	our @EXPORT = qw(
112	NODE $NODE *SELF node_of _any_	197	NODE $NODE *SELF node_of after
113	resolve_node initialise_node	198	configure
114	snd rcv mon kil reg psub	199	snd rcv mon mon_guard kil psub peval spawn cal
115	port	200	port
		201	db_set db_del db_reg
116	);	202	);
117		203
118	our $SELF;	204	our $SELF;
119		205
120	sub _self_die() {	206	sub _self_die() {
…		…
123	kil $SELF, die => $msg;	209	kil $SELF, die => $msg;
124	}	210	}
125		211
126	=item $thisnode = NODE / $NODE	212	=item $thisnode = NODE / $NODE
127		213
128	The C<NODE> function returns, and the C<$NODE> variable contains	214	The C<NODE> function returns, and the C<$NODE> variable contains, the node
129	the noderef of the local node. The value is initialised by a call	215	ID of the node running in the current process. This value is initialised by
130	to C<become_public> or C<become_slave>, after which all local port	216	a call to C<configure>.
131	identifiers become invalid.
132		217
133	=item $noderef = node_of $port	218	=item $nodeid = node_of $port
134		219
135	Extracts and returns the noderef from a portid or a noderef.	220	Extracts and returns the node ID from a port ID or a node ID.
136		221
137	=item initialise_node $noderef, $seednode, $seednode...	222	=item configure $profile, key => value...
138		223
139	=item initialise_node "slave/", $master, $master...	224	=item configure key => value...
140		225
141	Before a node can talk to other nodes on the network it has to initialise	226	Before a node can talk to other nodes on the network (i.e. enter
142	itself - the minimum a node needs to know is it's own name, and optionally	227	"distributed mode") it has to configure itself - the minimum a node needs
143	it should know the noderefs of some other nodes in the network.	228	to know is its own name, and optionally it should know the addresses of
		229	some other nodes in the network to discover other nodes.
144		230
145	This function initialises a node - it must be called exactly once (or	231	This function configures a node - it must be called exactly once (or
146	never) before calling other AnyEvent::MP functions.	232	never) before calling other AnyEvent::MP functions.
147		233
148	All arguments are noderefs, which can be either resolved or unresolved.	234	The key/value pairs are basically the same ones as documented for the
149		235	F<aemp> command line utility (sans the set/del prefix), with two additions:
150	There are two types of networked nodes, public nodes and slave nodes:
151		236
152	=over 4	237	=over 4
153		238
154	=item public nodes	239	=item norc => $boolean (default false)
155		240
156	For public nodes, C<$noderef> must either be a (possibly unresolved)	241	If true, then the rc file (e.g. F<~/.perl-anyevent-mp>) will I<not>
157	noderef, in which case it will be resolved, or C<undef> (or missing), in	242	be consulted - all configuraiton options must be specified in the
158	which case the noderef will be guessed.	243	C<configure> call.
159		244
160	Afterwards, the node will bind itself on all endpoints and try to connect	245	=item force => $boolean (default false)
161	to all additional C<$seednodes> that are specified. Seednodes are optional
162	and can be used to quickly bootstrap the node into an existing network.
163		246
164	=item slave nodes	247	IF true, then the values specified in the C<configure> will take
165		248	precedence over any values configured via the rc file. The default is for
166	When the C<$noderef> is the special string C<slave/>, then the node will	249	the rc file to override any options specified in the program.
167	become a slave node. Slave nodes cannot be contacted from outside and will
168	route most of their traffic to the master node that they attach to.
169
170	At least one additional noderef is required: The node will try to connect
171	to all of them and will become a slave attached to the first node it can
172	successfully connect to.
173		250
174	=back	251	=back
175		252
176	This function will block until all nodes have been resolved and, for slave
177	nodes, until it has successfully established a connection to a master
178	server.
179
180	Example: become a public node listening on the default node.
181
182	initialise_node;
183
184	Example: become a public node, and try to contact some well-known master
185	servers to become part of the network.
186
187	initialise_node undef, "master1", "master2";
188
189	Example: become a public node listening on port C<4041>.
190
191	initialise_node 4041;
192
193	Example: become a public node, only visible on localhost port 4044.
194
195	initialise_node "locahost:4044";
196
197	Example: become a slave node to any of the specified master servers.
198
199	initialise_node "slave/", "master1", "192.168.13.17", "mp.example.net";
200
201	=item $cv = resolve_node $noderef
202
203	Takes an unresolved node reference that may contain hostnames and
204	abbreviated IDs, resolves all of them and returns a resolved node
205	reference.
206
207	In addition to C<address:port> pairs allowed in resolved noderefs, the
208	following forms are supported:
209
210	=over 4	253	=over 4
211		254
212	=item the empty string	255	=item step 1, gathering configuration from profiles
213		256
214	An empty-string component gets resolved as if the default port (4040) was	257	The function first looks up a profile in the aemp configuration (see the
215	specified.	258	L<aemp> commandline utility). The profile name can be specified via the
		259	named C<profile> parameter or can simply be the first parameter). If it is
		260	missing, then the nodename (F<uname -n>) will be used as profile name.
216		261
217	=item naked port numbers (e.g. C<1234>)	262	The profile data is then gathered as follows:
218		263
219	These are resolved by prepending the local nodename and a colon, to be	264	First, all remaining key => value pairs (all of which are conveniently
220	further resolved.	265	undocumented at the moment) will be interpreted as configuration
		266	data. Then they will be overwritten by any values specified in the global
		267	default configuration (see the F<aemp> utility), then the chain of
		268	profiles chosen by the profile name (and any C<parent> attributes).
221		269
222	=item hostnames (e.g. C<localhost:1234>, C<localhost>)	270	That means that the values specified in the profile have highest priority
		271	and the values specified directly via C<configure> have lowest priority,
		272	and can only be used to specify defaults.
223		273
224	These are resolved by using AnyEvent::DNS to resolve them, optionally	274	If the profile specifies a node ID, then this will become the node ID of
225	looking up SRV records for the C<aemp=4040> port, if no port was	275	this process. If not, then the profile name will be used as node ID, with
226	specified.	276	a unique randoms tring (C</%u>) appended.
		277
		278	The node ID can contain some C<%> sequences that are expanded: C<%n>
		279	is expanded to the local nodename, C<%u> is replaced by a random
		280	strign to make the node unique. For example, the F<aemp> commandline
		281	utility uses C<aemp/%n/%u> as nodename, which might expand to
		282	C<aemp/cerebro/ZQDGSIkRhEZQDGSIkRhE>.
		283
		284	=item step 2, bind listener sockets
		285
		286	The next step is to look up the binds in the profile, followed by binding
		287	aemp protocol listeners on all binds specified (it is possible and valid
		288	to have no binds, meaning that the node cannot be contacted form the
		289	outside. This means the node cannot talk to other nodes that also have no
		290	binds, but it can still talk to all "normal" nodes).
		291
		292	If the profile does not specify a binds list, then a default of C<*> is
		293	used, meaning the node will bind on a dynamically-assigned port on every
		294	local IP address it finds.
		295
		296	=item step 3, connect to seed nodes
		297
		298	As the last step, the seed ID list from the profile is passed to the
		299	L<AnyEvent::MP::Global> module, which will then use it to keep
		300	connectivity with at least one node at any point in time.
227		301
228	=back	302	=back
		303
		304	Example: become a distributed node using the local node name as profile.
		305	This should be the most common form of invocation for "daemon"-type nodes.
		306
		307	configure
		308
		309	Example: become a semi-anonymous node. This form is often used for
		310	commandline clients.
		311
		312	configure nodeid => "myscript/%n/%u";
		313
		314	Example: configure a node using a profile called seed, which is suitable
		315	for a seed node as it binds on all local addresses on a fixed port (4040,
		316	customary for aemp).
		317
		318	# use the aemp commandline utility
		319	# aemp profile seed binds '*:4040'
		320
		321	# then use it
		322	configure profile => "seed";
		323
		324	# or simply use aemp from the shell again:
		325	# aemp run profile seed
		326
		327	# or provide a nicer-to-remember nodeid
		328	# aemp run profile seed nodeid "$(hostname)"
229		329
230	=item $SELF	330	=item $SELF
231		331
232	Contains the current port id while executing C<rcv> callbacks or C<psub>	332	Contains the current port id while executing C<rcv> callbacks or C<psub>
233	blocks.	333	blocks.
234		334
235	=item SELF, %SELF, @SELF...	335	=item *SELF, SELF, %SELF, @SELF...
236		336
237	Due to some quirks in how perl exports variables, it is impossible to	337	Due to some quirks in how perl exports variables, it is impossible to
238	just export C<$SELF>, all the symbols called C<SELF> are exported by this	338	just export C<$SELF>, all the symbols named C<SELF> are exported by this
239	module, but only C<$SELF> is currently used.	339	module, but only C<$SELF> is currently used.
240		340
241	=item snd $port, type => @data	341	=item snd $port, type => @data
242		342
243	=item snd $port, @msg	343	=item snd $port, @msg
244		344
245	Send the given message to the given port ID, which can identify either	345	Send the given message to the given port, which can identify either a
246	a local or a remote port, and can be either a string or soemthignt hat	346	local or a remote port, and must be a port ID.
247	stringifies a sa port ID (such as a port object :).
248		347
249	While the message can be about anything, it is highly recommended to use a	348	While the message can be almost anything, it is highly recommended to
250	string as first element (a portid, or some word that indicates a request	349	use a string as first element (a port ID, or some word that indicates a
251	type etc.).	350	request type etc.) and to consist if only simple perl values (scalars,
		351	arrays, hashes) - if you think you need to pass an object, think again.
252		352
253	The message data effectively becomes read-only after a call to this	353	The message data logically becomes read-only after a call to this
254	function: modifying any argument is not allowed and can cause many	354	function: modifying any argument (or values referenced by them) is
255	problems.	355	forbidden, as there can be considerable time between the call to C<snd>
		356	and the time the message is actually being serialised - in fact, it might
		357	never be copied as within the same process it is simply handed to the
		358	receiving port.
256		359
257	The type of data you can transfer depends on the transport protocol: when	360	The type of data you can transfer depends on the transport protocol: when
258	JSON is used, then only strings, numbers and arrays and hashes consisting	361	JSON is used, then only strings, numbers and arrays and hashes consisting
259	of those are allowed (no objects). When Storable is used, then anything	362	of those are allowed (no objects). When Storable is used, then anything
260	that Storable can serialise and deserialise is allowed, and for the local	363	that Storable can serialise and deserialise is allowed, and for the local
261	node, anything can be passed.	364	node, anything can be passed. Best rely only on the common denominator of
		365	these.
262		366
263	=item $local_port = port	367	=item $local_port = port
264		368
265	Create a new local port object that can be used either as a pattern	369	Create a new local port object and returns its port ID. Initially it has
266	matching port ("full port") or a single-callback port ("miniport"),	370	no callbacks set and will throw an error when it receives messages.
267	depending on how C<rcv> callbacks are bound to the object.
268		371
269	=item $port = port { my @msg = @_; $finished }	372	=item $local_port = port { my @msg = @_ }
270		373
271	Creates a "miniport", that is, a very lightweight port without any pattern	374	Creates a new local port, and returns its ID. Semantically the same as
272	matching behind it, and returns its ID. Semantically the same as creating
273	a port and calling C<rcv $port, $callback> on it.	375	creating a port and calling C<rcv $port, $callback> on it.
274		376
275	The block will be called for every message received on the port. When the	377	The block will be called for every message received on the port, with the
276	callback returns a true value its job is considered "done" and the port	378	global variable C<$SELF> set to the port ID. Runtime errors will cause the
277	will be destroyed. Otherwise it will stay alive.	379	port to be C<kil>ed. The message will be passed as-is, no extra argument
		380	(i.e. no port ID) will be passed to the callback.
278		381
279	The message will be passed as-is, no extra argument (i.e. no port id) will	382	If you want to stop/destroy the port, simply C<kil> it:
280	be passed to the callback.
281		383
282	If you need the local port id in the callback, this works nicely:	384	my $port = port {
283		385	my @msg = @_;
284	my $port; $port = port {	386	...
285	snd $otherport, reply => $port;	387	kil $SELF;
286	};	388	};
287		389
288	=cut	390	=cut
289		391
290	sub rcv($@);	392	sub rcv($@);
291		393
		394	sub _kilme {
		395	die "received message on port without callback";
		396	}
		397
292	sub port(;&) {	398	sub port(;&) {
293	my $id = "$UNIQ." . $ID++;	399	my $id = $UNIQ . ++$ID;
294	my $port = "$NODE#$id";	400	my $port = "$NODE#$id";
295		401
296	if (@_) {	402	rcv $port, shift \|\| \&_kilme;
297	rcv $port, shift;
298	} else {
299	$PORT{$id} = sub { }; # nop
300	}
301		403
302	$port	404	$port
303	}	405	}
304		406
305	=item reg $port, $name
306
307	=item reg $name
308
309	Registers the given port (or C<$SELF><<< if missing) under the name
310	C<$name>. If the name already exists it is replaced.
311
312	A port can only be registered under one well known name.
313
314	A port automatically becomes unregistered when it is killed.
315
316	=cut
317
318	sub reg(@) {
319	my $port = @_ > 1 ? shift : $SELF \|\| Carp::croak 'reg: called with one argument only, but $SELF not set,';
320
321	$REG{$_[0]} = $port;
322	}
323
324	=item rcv $port, $callback->(@msg)	407	=item rcv $local_port, $callback->(@msg)
325		408
326	Replaces the callback on the specified miniport (after converting it to	409	Replaces the default callback on the specified port. There is no way to
327	one if required).	410	remove the default callback: use C<sub { }> to disable it, or better
328		411	C<kil> the port when it is no longer needed.
329	=item rcv $port, tagstring => $callback->(@msg), ...
330
331	=item rcv $port, $smartmatch => $callback->(@msg), ...
332
333	=item rcv $port, [$smartmatch...] => $callback->(@msg), ...
334
335	Register callbacks to be called on matching messages on the given full
336	port (after converting it to one if required) and return the port.
337
338	The callback has to return a true value when its work is done, after
339	which is will be removed, or a false value in which case it will stay
340	registered.
341		412
342	The global C<$SELF> (exported by this module) contains C<$port> while	413	The global C<$SELF> (exported by this module) contains C<$port> while
343	executing the callback.	414	executing the callback. Runtime errors during callback execution will
		415	result in the port being C<kil>ed.
344		416
345	Runtime errors wdurign callback execution will result in the port being	417	The default callback received all messages not matched by a more specific
346	C<kil>ed.	418	C<tag> match.
347		419
348	If the match is an array reference, then it will be matched against the	420	=item rcv $local_port, tag => $callback->(@msg_without_tag), ...
349	first elements of the message, otherwise only the first element is being
350	matched.
351		421
352	Any element in the match that is specified as C<_any_> (a function	422	Register (or replace) callbacks to be called on messages starting with the
353	exported by this module) matches any single element of the message.	423	given tag on the given port (and return the port), or unregister it (when
		424	C<$callback> is C<$undef> or missing). There can only be one callback
		425	registered for each tag.
354		426
355	While not required, it is highly recommended that the first matching	427	The original message will be passed to the callback, after the first
356	element is a string identifying the message. The one-string-only match is	428	element (the tag) has been removed. The callback will use the same
357	also the most efficient match (by far).	429	environment as the default callback (see above).
358		430
359	Example: create a port and bind receivers on it in one go.	431	Example: create a port and bind receivers on it in one go.
360		432
361	my $port = rcv port,	433	my $port = rcv port,
362	msg1 => sub { ...; 0 },	434	msg1 => sub { ... },
363	msg2 => sub { ...; 0 },	435	msg2 => sub { ... },
364	;	436	;
365		437
366	Example: create a port, bind receivers and send it in a message elsewhere	438	Example: create a port, bind receivers and send it in a message elsewhere
367	in one go:	439	in one go:
368		440
369	snd $otherport, reply =>	441	snd $otherport, reply =>
370	rcv port,	442	rcv port,
371	msg1 => sub { ...; 0 },	443	msg1 => sub { ... },
372	...	444	...
373	;	445	;
374		446
		447	Example: temporarily register a rcv callback for a tag matching some port
		448	(e.g. for an rpc reply) and unregister it after a message was received.
		449
		450	rcv $port, $otherport => sub {
		451	my @reply = @_;
		452
		453	rcv $SELF, $otherport;
		454	};
		455
375	=cut	456	=cut
376		457
377	sub rcv($@) {	458	sub rcv($@) {
378	my $port = shift;	459	my $port = shift;
379	my ($noderef, $portid) = split /#/, $port, 2;	460	my ($nodeid, $portid) = split /#/, $port, 2;
380		461
381	($NODE{$noderef} \|\| add_node $noderef) == $NODE{""}	462	$NODE{$nodeid} == $NODE{""}
382	or Carp::croak "$port: rcv can only be called on local ports, caught";	463	or Carp::croak "$port: rcv can only be called on local ports, caught";
383		464
384	if (@_ == 1) {	465	while (@_) {
		466	if (ref $_[0]) {
		467	if (my $self = $PORT_DATA{$portid}) {
		468	"AnyEvent::MP::Port" eq ref $self
		469	or Carp::croak "$port: rcv can only be called on message matching ports, caught";
		470
		471	$self->[0] = shift;
		472	} else {
385	my $cb = shift;	473	my $cb = shift;
386	delete $PORT_DATA{$portid};
387	$PORT{$portid} = sub {	474	$PORT{$portid} = sub {
388	local $SELF = $port;	475	local $SELF = $port;
389	eval {	476	eval { &$cb }; _self_die if $@;
390	&$cb	477	};
391	and kil $port;
392	};	478	}
393	_self_die if $@;	479	} elsif (defined $_[0]) {
394	};
395	} else {
396	my $self = $PORT_DATA{$portid} \|\|= do {	480	my $self = $PORT_DATA{$portid} \|\|= do {
397	my $self = bless {	481	my $self = bless [$PORT{$portid} \|\| sub { }, { }, $port], "AnyEvent::MP::Port";
398	id => $port,
399	}, "AnyEvent::MP::Port";
400		482
401	$PORT{$portid} = sub {	483	$PORT{$portid} = sub {
402	local $SELF = $port;	484	local $SELF = $port;
403		485
404	eval {
405	for (@{ $self->{rc0}{$_[0]} }) {	486	if (my $cb = $self->[1]{$_[0]}) {
406	$_ && &{$_->[0]}	487	shift;
407	&& undef $_;	488	eval { &$cb }; _self_die if $@;
408	}	489	} else {
409
410	for (@{ $self->{rcv}{$_[0]} }) {
411	$_ && [@_[1 .. @{$_->[1]}]] ~~ $_->[1]
412	&& &{$_->[0]}	490	&{ $self->[0] };
413	&& undef $_;
414	}
415
416	for (@{ $self->{any} }) {
417	$_ && [@_[0 .. $#{$_->[1]}]] ~~ $_->[1]
418	&& &{$_->[0]}
419	&& undef $_;
420	}	491	}
421	};	492	};
422	_self_die if $@;	493
		494	$self
423	};	495	};
424		496
425	$self
426	};
427
428	"AnyEvent::MP::Port" eq ref $self	497	"AnyEvent::MP::Port" eq ref $self
429	or Carp::croak "$port: rcv can only be called on message matching ports, caught";	498	or Carp::croak "$port: rcv can only be called on message matching ports, caught";
430		499
431	while (@_) {
432	my ($match, $cb) = splice @_, 0, 2;	500	my ($tag, $cb) = splice @_, 0, 2;
433		501
434	if (!ref $match) {	502	if (defined $cb) {
435	push @{ $self->{rc0}{$match} }, [$cb];	503	$self->[1]{$tag} = $cb;
436	} elsif (("ARRAY" eq ref $match && !ref $match->[0])) {
437	my ($type, @match) = @$match;
438	@match
439	? push @{ $self->{rcv}{$match->[0]} }, [$cb, \@match]
440	: push @{ $self->{rc0}{$match->[0]} }, [$cb];
441	} else {	504	} else {
442	push @{ $self->{any} }, [$cb, $match];	505	delete $self->[1]{$tag};
443	}	506	}
444	}	507	}
445	}	508	}
446		509
447	$port	510	$port
448	}	511	}
449		512
		513	=item peval $port, $coderef[, @args]
		514
		515	Evaluates the given C<$codref> within the contetx of C<$port>, that is,
		516	when the code throews an exception the C<$port> will be killed.
		517
		518	Any remaining args will be passed to the callback. Any return values will
		519	be returned to the caller.
		520
		521	This is useful when you temporarily want to execute code in the context of
		522	a port.
		523
		524	Example: create a port and run some initialisation code in it's context.
		525
		526	my $port = port { ... };
		527
		528	peval $port, sub {
		529	init
		530	or die "unable to init";
		531	};
		532
		533	=cut
		534
		535	sub peval($$) {
		536	local $SELF = shift;
		537	my $cb = shift;
		538
		539	if (wantarray) {
		540	my @res = eval { &$cb };
		541	_self_die if $@;
		542	@res
		543	} else {
		544	my $res = eval { &$cb };
		545	_self_die if $@;
		546	$res
		547	}
		548	}
		549
450	=item $closure = psub { BLOCK }	550	=item $closure = psub { BLOCK }
451		551
452	Remembers C<$SELF> and creates a closure out of the BLOCK. When the	552	Remembers C<$SELF> and creates a closure out of the BLOCK. When the
453	closure is executed, sets up the environment in the same way as in C<rcv>	553	closure is executed, sets up the environment in the same way as in C<rcv>
454	callbacks, i.e. runtime errors will cause the port to get C<kil>ed.	554	callbacks, i.e. runtime errors will cause the port to get C<kil>ed.
		555
		556	The effect is basically as if it returned C<< sub { peval $SELF, sub {
		557	BLOCK }, @_ } >>.
455		558
456	This is useful when you register callbacks from C<rcv> callbacks:	559	This is useful when you register callbacks from C<rcv> callbacks:
457		560
458	rcv delayed_reply => sub {	561	rcv delayed_reply => sub {
459	my ($delay, @reply) = @_;	562	my ($delay, @reply) = @_;
…		…
483	$res	586	$res
484	}	587	}
485	}	588	}
486	}	589	}
487		590
488	=item $guard = mon $port, $cb->(@reason)	591	=item $guard = mon $port, $cb->(@reason) # call $cb when $port dies
489		592
490	=item $guard = mon $port, $rcvport	593	=item $guard = mon $port, $rcvport # kill $rcvport when $port dies
491		594
492	=item $guard = mon $port	595	=item $guard = mon $port # kill $SELF when $port dies
493		596
494	=item $guard = mon $port, $rcvport, @msg	597	=item $guard = mon $port, $rcvport, @msg # send a message when $port dies
495		598
496	Monitor the given port and do something when the port is killed, and	599	Monitor the given port and do something when the port is killed or
497	optionally return a guard that can be used to stop monitoring again.	600	messages to it were lost, and optionally return a guard that can be used
		601	to stop monitoring again.
498		602
499	In the first form (callback), the callback is simply called with any	603	In the first form (callback), the callback is simply called with any
500	number of C<@reason> elements (no @reason means that the port was deleted	604	number of C<@reason> elements (no @reason means that the port was deleted
501	"normally"). Note also that I<< the callback B<must> never die >>, so use	605	"normally"). Note also that I<< the callback B<must> never die >>, so use
502	C<eval> if unsure.	606	C<eval> if unsure.
503		607
504	In the second form (another port given), the other port (C<$rcvport)	608	In the second form (another port given), the other port (C<$rcvport>)
505	will be C<kil>'ed with C<@reason>, iff a @reason was specified, i.e. on	609	will be C<kil>'ed with C<@reason>, if a @reason was specified, i.e. on
506	"normal" kils nothing happens, while under all other conditions, the other	610	"normal" kils nothing happens, while under all other conditions, the other
507	port is killed with the same reason.	611	port is killed with the same reason.
508		612
509	The third form (kill self) is the same as the second form, except that	613	The third form (kill self) is the same as the second form, except that
510	C<$rvport> defaults to C<$SELF>.	614	C<$rvport> defaults to C<$SELF>.
511		615
512	In the last form (message), a message of the form C<@msg, @reason> will be	616	In the last form (message), a message of the form C<@msg, @reason> will be
513	C<snd>.	617	C<snd>.
		618
		619	Monitoring-actions are one-shot: once messages are lost (and a monitoring
		620	alert was raised), they are removed and will not trigger again.
514		621
515	As a rule of thumb, monitoring requests should always monitor a port from	622	As a rule of thumb, monitoring requests should always monitor a port from
516	a local port (or callback). The reason is that kill messages might get	623	a local port (or callback). The reason is that kill messages might get
517	lost, just like any other message. Another less obvious reason is that	624	lost, just like any other message. Another less obvious reason is that
518	even monitoring requests can get lost (for exmaple, when the connection	625	even monitoring requests can get lost (for example, when the connection
519	to the other node goes down permanently). When monitoring a port locally	626	to the other node goes down permanently). When monitoring a port locally
520	these problems do not exist.	627	these problems do not exist.
521		628
		629	C<mon> effectively guarantees that, in the absence of hardware failures,
		630	after starting the monitor, either all messages sent to the port will
		631	arrive, or the monitoring action will be invoked after possible message
		632	loss has been detected. No messages will be lost "in between" (after
		633	the first lost message no further messages will be received by the
		634	port). After the monitoring action was invoked, further messages might get
		635	delivered again.
		636
		637	Inter-host-connection timeouts and monitoring depend on the transport
		638	used. The only transport currently implemented is TCP, and AnyEvent::MP
		639	relies on TCP to detect node-downs (this can take 10-15 minutes on a
		640	non-idle connection, and usually around two hours for idle connections).
		641
		642	This means that monitoring is good for program errors and cleaning up
		643	stuff eventually, but they are no replacement for a timeout when you need
		644	to ensure some maximum latency.
		645
522	Example: call a given callback when C<$port> is killed.	646	Example: call a given callback when C<$port> is killed.
523		647
524	mon $port, sub { warn "port died because of <@_>\n" };	648	mon $port, sub { warn "port died because of <@_>\n" };
525		649
526	Example: kill ourselves when C<$port> is killed abnormally.	650	Example: kill ourselves when C<$port> is killed abnormally.
…		…
532	mon $port, $self => "restart";	656	mon $port, $self => "restart";
533		657
534	=cut	658	=cut
535		659
536	sub mon {	660	sub mon {
537	my ($noderef, $port) = split /#/, shift, 2;	661	my ($nodeid, $port) = split /#/, shift, 2;
538		662
539	my $node = $NODE{$noderef} \|\| add_node $noderef;	663	my $node = $NODE{$nodeid} \|\| add_node $nodeid;
540		664
541	my $cb = @_ ? $_[0] : $SELF \|\| Carp::croak 'mon: called with one argument only, but $SELF not set,';	665	my $cb = @_ ? shift : $SELF \|\| Carp::croak 'mon: called with one argument only, but $SELF not set,';
542		666
543	unless (ref $cb) {	667	unless (ref $cb) {
544	if (@_) {	668	if (@_) {
545	# send a kill info message	669	# send a kill info message
546	my (@msg) = @_;	670	my (@msg) = ($cb, @_);
547	$cb = sub { snd @msg, @_ };	671	$cb = sub { snd @msg, @_ };
548	} else {	672	} else {
549	# simply kill other port	673	# simply kill other port
550	my $port = $cb;	674	my $port = $cb;
551	$cb = sub { kil $port, @_ if @_ };	675	$cb = sub { kil $port, @_ if @_ };
…		…
553	}	677	}
554		678
555	$node->monitor ($port, $cb);	679	$node->monitor ($port, $cb);
556		680
557	defined wantarray	681	defined wantarray
558	and AnyEvent::Util::guard { $node->unmonitor ($port, $cb) }	682	and ($cb += 0, Guard::guard { $node->unmonitor ($port, $cb) })
559	}	683	}
560		684
561	=item $guard = mon_guard $port, $ref, $ref...	685	=item $guard = mon_guard $port, $ref, $ref...
562		686
563	Monitors the given C<$port> and keeps the passed references. When the port	687	Monitors the given C<$port> and keeps the passed references. When the port
564	is killed, the references will be freed.	688	is killed, the references will be freed.
565		689
566	Optionally returns a guard that will stop the monitoring.	690	Optionally returns a guard that will stop the monitoring.
567		691
568	This function is useful when you create e.g. timers or other watchers and	692	This function is useful when you create e.g. timers or other watchers and
569	want to free them when the port gets killed:	693	want to free them when the port gets killed (note the use of C<psub>):
570		694
571	$port->rcv (start => sub {	695	$port->rcv (start => sub {
572	my $timer; $timer = mon_guard $port, AE::timer 1, 1, sub {	696	my $timer; $timer = mon_guard $port, AE::timer 1, 1, psub {
573	undef $timer if 0.9 < rand;	697	undef $timer if 0.9 < rand;
574	});	698	});
575	});	699	});
576		700
577	=cut	701	=cut
…		…
586		710
587	=item kil $port[, @reason]	711	=item kil $port[, @reason]
588		712
589	Kill the specified port with the given C<@reason>.	713	Kill the specified port with the given C<@reason>.
590		714
591	If no C<@reason> is specified, then the port is killed "normally" (linked	715	If no C<@reason> is specified, then the port is killed "normally" -
592	ports will not be kileld, or even notified).	716	monitor callback will be invoked, but the kil will not cause linked ports
		717	(C<mon $mport, $lport> form) to get killed.
593		718
594	Otherwise, linked ports get killed with the same reason (second form of	719	If a C<@reason> is specified, then linked ports (C<mon $mport, $lport>
595	C<mon>, see below).	720	form) get killed with the same reason.
596		721
597	Runtime errors while evaluating C<rcv> callbacks or inside C<psub> blocks	722	Runtime errors while evaluating C<rcv> callbacks or inside C<psub> blocks
598	will be reported as reason C<< die => $@ >>.	723	will be reported as reason C<< die => $@ >>.
599		724
600	Transport/communication errors are reported as C<< transport_error =>	725	Transport/communication errors are reported as C<< transport_error =>
601	$message >>.	726	$message >>.
602		727
		728	=cut
		729
		730	=item $port = spawn $node, $initfunc[, @initdata]
		731
		732	Creates a port on the node C<$node> (which can also be a port ID, in which
		733	case it's the node where that port resides).
		734
		735	The port ID of the newly created port is returned immediately, and it is
		736	possible to immediately start sending messages or to monitor the port.
		737
		738	After the port has been created, the init function is called on the remote
		739	node, in the same context as a C<rcv> callback. This function must be a
		740	fully-qualified function name (e.g. C<MyApp::Chat::Server::init>). To
		741	specify a function in the main program, use C<::name>.
		742
		743	If the function doesn't exist, then the node tries to C<require>
		744	the package, then the package above the package and so on (e.g.
		745	C<MyApp::Chat::Server>, C<MyApp::Chat>, C<MyApp>) until the function
		746	exists or it runs out of package names.
		747
		748	The init function is then called with the newly-created port as context
		749	object (C<$SELF>) and the C<@initdata> values as arguments. It I<must>
		750	call one of the C<rcv> functions to set callbacks on C<$SELF>, otherwise
		751	the port might not get created.
		752
		753	A common idiom is to pass a local port, immediately monitor the spawned
		754	port, and in the remote init function, immediately monitor the passed
		755	local port. This two-way monitoring ensures that both ports get cleaned up
		756	when there is a problem.
		757
		758	C<spawn> guarantees that the C<$initfunc> has no visible effects on the
		759	caller before C<spawn> returns (by delaying invocation when spawn is
		760	called for the local node).
		761
		762	Example: spawn a chat server port on C<$othernode>.
		763
		764	# this node, executed from within a port context:
		765	my $server = spawn $othernode, "MyApp::Chat::Server::connect", $SELF;
		766	mon $server;
		767
		768	# init function on C<$othernode>
		769	sub connect {
		770	my ($srcport) = @_;
		771
		772	mon $srcport;
		773
		774	rcv $SELF, sub {
		775	...
		776	};
		777	}
		778
		779	=cut
		780
		781	sub _spawn {
		782	my $port = shift;
		783	my $init = shift;
		784
		785	# rcv will create the actual port
		786	local $SELF = "$NODE#$port";
		787	eval {
		788	&{ load_func $init }
		789	};
		790	_self_die if $@;
		791	}
		792
		793	sub spawn(@) {
		794	my ($nodeid, undef) = split /#/, shift, 2;
		795
		796	my $id = $RUNIQ . ++$ID;
		797
		798	$_[0] =~ /::/
		799	or Carp::croak "spawn init function must be a fully-qualified name, caught";
		800
		801	snd_to_func $nodeid, "AnyEvent::MP::_spawn" => $id, @_;
		802
		803	"$nodeid#$id"
		804	}
		805
		806
		807	=item after $timeout, @msg
		808
		809	=item after $timeout, $callback
		810
		811	Either sends the given message, or call the given callback, after the
		812	specified number of seconds.
		813
		814	This is simply a utility function that comes in handy at times - the
		815	AnyEvent::MP author is not convinced of the wisdom of having it, though,
		816	so it may go away in the future.
		817
		818	=cut
		819
		820	sub after($@) {
		821	my ($timeout, @action) = @_;
		822
		823	my $t; $t = AE::timer $timeout, 0, sub {
		824	undef $t;
		825	ref $action[0]
		826	? $action[0]()
		827	: snd @action;
		828	};
		829	}
		830
		831	=item cal $port, @msg, $callback[, $timeout]
		832
		833	A simple form of RPC - sends a message to the given C<$port> with the
		834	given contents (C<@msg>), but adds a reply port to the message.
		835
		836	The reply port is created temporarily just for the purpose of receiving
		837	the reply, and will be C<kil>ed when no longer needed.
		838
		839	A reply message sent to the port is passed to the C<$callback> as-is.
		840
		841	If an optional time-out (in seconds) is given and it is not C<undef>,
		842	then the callback will be called without any arguments after the time-out
		843	elapsed and the port is C<kil>ed.
		844
		845	If no time-out is given (or it is C<undef>), then the local port will
		846	monitor the remote port instead, so it eventually gets cleaned-up.
		847
		848	Currently this function returns the temporary port, but this "feature"
		849	might go in future versions unless you can make a convincing case that
		850	this is indeed useful for something.
		851
		852	=cut
		853
		854	sub cal(@) {
		855	my $timeout = ref $_[-1] ? undef : pop;
		856	my $cb = pop;
		857
		858	my $port = port {
		859	undef $timeout;
		860	kil $SELF;
		861	&$cb;
		862	};
		863
		864	if (defined $timeout) {
		865	$timeout = AE::timer $timeout, 0, sub {
		866	undef $timeout;
		867	kil $port;
		868	$cb->();
		869	};
		870	} else {
		871	mon $_[0], sub {
		872	kil $port;
		873	$cb->();
		874	};
		875	}
		876
		877	push @_, $port;
		878	&snd;
		879
		880	$port
		881	}
		882
603	=back	883	=back
604		884
605	=head1 NODE MESSAGES	885	=head1 DISTRIBUTED DATABASE
606		886
607	Nodes understand the following messages sent to them. Many of them take	887	AnyEvent::MP comes with a simple distributed database. The database will
608	arguments called C<@reply>, which will simply be used to compose a reply	888	be mirrored asynchronously at all global nodes. Other nodes bind to one of
609	message - C<$reply[0]> is the port to reply to, C<$reply[1]> the type and	889	the global nodes for their needs.
610	the remaining arguments are simply the message data.
611		890
612	While other messages exist, they are not public and subject to change.	891	The database consists of a two-level hash - a hash contains a hash which
		892	contains values.
613		893
		894	The top level hash key is called "family", and the second-level hash key
		895	is called "subkey" or simply "key".
		896
		897	The family must be alphanumeric, i.e. start with a letter and consist
		898	of letters, digits, underscores and colons (C<[A-Za-z][A-Za-z0-9_:]*>,
		899	pretty much like Perl module names.
		900
		901	As the family namespace is global, it is recommended to prefix family names
		902	with the name of the application or module using it.
		903
		904	The subkeys must be non-empty strings, with no further restrictions.
		905
		906	The values should preferably be strings, but other perl scalars should
		907	work as well (such as undef, arrays and hashes).
		908
		909	Every database entry is owned by one node - adding the same family/subkey
		910	combination on multiple nodes will not cause discomfort for AnyEvent::MP,
		911	but the result might be nondeterministic, i.e. the key might have
		912	different values on different nodes.
		913
		914	Different subkeys in the same family can be owned by different nodes
		915	without problems, and in fact, this is the common method to create worker
		916	pools. For example, a worker port for image scaling might do this:
		917
		918	db_set my_image_scalers => $port;
		919
		920	And clients looking for an image scaler will want to get the
		921	C<my_image_scalers> keys:
		922
		923	db_keys "my_image_scalers" => 60 => sub {
		924	#d##TODO#
		925
614	=over 4	926	=over
615		927
616	=cut	928	=item db_set $family => $subkey [=> $value]
617		929
618	=item lookup => $name, @reply	930	Sets (or replaces) a key to the database - if C<$value> is omitted,
		931	C<undef> is used instead.
619		932
620	Replies with the port ID of the specified well-known port, or C<undef>.	933	=item db_del $family => $subkey
621		934
622	=item devnull => ...	935	Deletes a key from the database.
623		936
624	Generic data sink/CPU heat conversion.	937	=item $guard = db_reg $family => $subkey [=> $value]
625		938
626	=item relay => $port, @msg	939	Sets the key on the database and returns a guard. When the guard is
		940	destroyed, the key is deleted from the database. If C<$value> is missing,
		941	then C<undef> is used.
627		942
628	Simply forwards the message to the given port.	943	=cut
629
630	=item eval => $string[ @reply]
631
632	Evaluates the given string. If C<@reply> is given, then a message of the
633	form C<@reply, $@, @evalres> is sent.
634
635	Example: crash another node.
636
637	snd $othernode, eval => "exit";
638
639	=item time => @reply
640
641	Replies the the current node time to C<@reply>.
642
643	Example: tell the current node to send the current time to C<$myport> in a
644	C<timereply> message.
645
646	snd $NODE, time => $myport, timereply => 1, 2;
647	# => snd $myport, timereply => 1, 2, <time>
648		944
649	=back	945	=back
650		946
651	=head1 AnyEvent::MP vs. Distributed Erlang	947	=head1 AnyEvent::MP vs. Distributed Erlang
652		948
653	AnyEvent::MP got lots of its ideas from distributed Erlang (Erlang node	949	AnyEvent::MP got lots of its ideas from distributed Erlang (Erlang node
654	== aemp node, Erlang process == aemp port), so many of the documents and	950	== aemp node, Erlang process == aemp port), so many of the documents and
655	programming techniques employed by Erlang apply to AnyEvent::MP. Here is a	951	programming techniques employed by Erlang apply to AnyEvent::MP. Here is a
656	sample:	952	sample:
657		953
658	http://www.Erlang.se/doc/programming_rules.shtml	954	http://www.erlang.se/doc/programming_rules.shtml
659	http://Erlang.org/doc/getting_started/part_frame.html # chapters 3 and 4	955	http://erlang.org/doc/getting_started/part_frame.html # chapters 3 and 4
660	http://Erlang.org/download/Erlang-book-part1.pdf # chapters 5 and 6	956	http://erlang.org/download/erlang-book-part1.pdf # chapters 5 and 6
661	http://Erlang.org/download/armstrong_thesis_2003.pdf # chapters 4 and 5	957	http://erlang.org/download/armstrong_thesis_2003.pdf # chapters 4 and 5
662		958
663	Despite the similarities, there are also some important differences:	959	Despite the similarities, there are also some important differences:
664		960
665	=over 4	961	=over 4
666		962
667	=item * Node references contain the recipe on how to contact them.	963	=item * Node IDs are arbitrary strings in AEMP.
668		964
669	Erlang relies on special naming and DNS to work everywhere in the	965	Erlang relies on special naming and DNS to work everywhere in the same
670	same way. AEMP relies on each node knowing it's own address(es), with	966	way. AEMP relies on each node somehow knowing its own address(es) (e.g. by
671	convenience functionality.	967	configuration or DNS), and possibly the addresses of some seed nodes, but
		968	will otherwise discover other nodes (and their IDs) itself.
672		969
673	This means that AEMP requires a less tightly controlled environment at the	970	=item * Erlang has a "remote ports are like local ports" philosophy, AEMP
674	cost of longer node references and a slightly higher management overhead.	971	uses "local ports are like remote ports".
		972
		973	The failure modes for local ports are quite different (runtime errors
		974	only) then for remote ports - when a local port dies, you I<know> it dies,
		975	when a connection to another node dies, you know nothing about the other
		976	port.
		977
		978	Erlang pretends remote ports are as reliable as local ports, even when
		979	they are not.
		980
		981	AEMP encourages a "treat remote ports differently" philosophy, with local
		982	ports being the special case/exception, where transport errors cannot
		983	occur.
675		984
676	=item * Erlang uses processes and a mailbox, AEMP does not queue.	985	=item * Erlang uses processes and a mailbox, AEMP does not queue.
677		986
678	Erlang uses processes that selctively receive messages, and therefore	987	Erlang uses processes that selectively receive messages out of order, and
679	needs a queue. AEMP is event based, queuing messages would serve no useful	988	therefore needs a queue. AEMP is event based, queuing messages would serve
680	purpose.	989	no useful purpose. For the same reason the pattern-matching abilities
		990	of AnyEvent::MP are more limited, as there is little need to be able to
		991	filter messages without dequeuing them.
681		992
682	(But see L<Coro::MP> for a more Erlang-like process model on top of AEMP).	993	This is not a philosophical difference, but simply stems from AnyEvent::MP
		994	being event-based, while Erlang is process-based.
		995
		996	You cna have a look at L<Coro::MP> for a more Erlang-like process model on
		997	top of AEMP and Coro threads.
683		998
684	=item * Erlang sends are synchronous, AEMP sends are asynchronous.	999	=item * Erlang sends are synchronous, AEMP sends are asynchronous.
685		1000
686	Sending messages in Erlang is synchronous and blocks the process. AEMP	1001	Sending messages in Erlang is synchronous and blocks the process until
687	sends are immediate, connection establishment is handled in the	1002	a conenction has been established and the message sent (and so does not
688	background.	1003	need a queue that can overflow). AEMP sends return immediately, connection
		1004	establishment is handled in the background.
689		1005
690	=item * Erlang can silently lose messages, AEMP cannot.	1006	=item * Erlang suffers from silent message loss, AEMP does not.
691		1007
692	Erlang makes few guarantees on messages delivery - messages can get lost	1008	Erlang implements few guarantees on messages delivery - messages can get
693	without any of the processes realising it (i.e. you send messages a, b,	1009	lost without any of the processes realising it (i.e. you send messages a,
694	and c, and the other side only receives messages a and c).	1010	b, and c, and the other side only receives messages a and c).
695		1011
696	AEMP guarantees correct ordering, and the guarantee that there are no	1012	AEMP guarantees (modulo hardware errors) correct ordering, and the
		1013	guarantee that after one message is lost, all following ones sent to the
		1014	same port are lost as well, until monitoring raises an error, so there are
697	holes in the message sequence.	1015	no silent "holes" in the message sequence.
698		1016
699	=item * In Erlang, processes can be declared dead and later be found to be	1017	If you want your software to be very reliable, you have to cope with
700	alive.	1018	corrupted and even out-of-order messages in both Erlang and AEMP. AEMP
701		1019	simply tries to work better in common error cases, such as when a network
702	In Erlang it can happen that a monitored process is declared dead and	1020	link goes down.
703	linked processes get killed, but later it turns out that the process is
704	still alive - and can receive messages.
705
706	In AEMP, when port monitoring detects a port as dead, then that port will
707	eventually be killed - it cannot happen that a node detects a port as dead
708	and then later sends messages to it, finding it is still alive.
709		1021
710	=item * Erlang can send messages to the wrong port, AEMP does not.	1022	=item * Erlang can send messages to the wrong port, AEMP does not.
711		1023
712	In Erlang it is quite possible that a node that restarts reuses a process	1024	In Erlang it is quite likely that a node that restarts reuses an Erlang
713	ID known to other nodes for a completely different process, causing	1025	process ID known to other nodes for a completely different process,
714	messages destined for that process to end up in an unrelated process.	1026	causing messages destined for that process to end up in an unrelated
		1027	process.
715		1028
716	AEMP never reuses port IDs, so old messages or old port IDs floating	1029	AEMP does not reuse port IDs, so old messages or old port IDs floating
717	around in the network will not be sent to an unrelated port.	1030	around in the network will not be sent to an unrelated port.
718		1031
719	=item * Erlang uses unprotected connections, AEMP uses secure	1032	=item * Erlang uses unprotected connections, AEMP uses secure
720	authentication and can use TLS.	1033	authentication and can use TLS.
721		1034
722	AEMP can use a proven protocol - SSL/TLS - to protect connections and	1035	AEMP can use a proven protocol - TLS - to protect connections and
723	securely authenticate nodes.	1036	securely authenticate nodes.
724		1037
725	=item * The AEMP protocol is optimised for both text-based and binary	1038	=item * The AEMP protocol is optimised for both text-based and binary
726	communications.	1039	communications.
727		1040
728	The AEMP protocol, unlike the Erlang protocol, supports both	1041	The AEMP protocol, unlike the Erlang protocol, supports both programming
729	language-independent text-only protocols (good for debugging) and binary,	1042	language independent text-only protocols (good for debugging), and binary,
730	language-specific serialisers (e.g. Storable).	1043	language-specific serialisers (e.g. Storable). By default, unless TLS is
		1044	used, the protocol is actually completely text-based.
731		1045
732	It has also been carefully designed to be implementable in other languages	1046	It has also been carefully designed to be implementable in other languages
733	with a minimum of work while gracefully degrading fucntionality to make the	1047	with a minimum of work while gracefully degrading functionality to make the
734	protocol simple.	1048	protocol simple.
735		1049
736	=item * AEMP has more flexible monitoring options than Erlang.	1050	=item * AEMP has more flexible monitoring options than Erlang.
737		1051
738	In Erlang, you can chose to receive I<all> exit signals as messages	1052	In Erlang, you can chose to receive I<all> exit signals as messages or
739	or I<none>, there is no in-between, so monitoring single processes is	1053	I<none>, there is no in-between, so monitoring single Erlang processes is
740	difficult to implement. Monitoring in AEMP is more flexible than in	1054	difficult to implement.
741	Erlang, as one can choose between automatic kill, exit message or callback	1055
742	on a per-process basis.	1056	Monitoring in AEMP is more flexible than in Erlang, as one can choose
		1057	between automatic kill, exit message or callback on a per-port basis.
743		1058
744	=item * Erlang tries to hide remote/local connections, AEMP does not.	1059	=item * Erlang tries to hide remote/local connections, AEMP does not.
745		1060
746	Monitoring in Erlang is not an indicator of process death/crashes,	1061	Monitoring in Erlang is not an indicator of process death/crashes, in the
747	as linking is (except linking is unreliable in Erlang).	1062	same way as linking is (except linking is unreliable in Erlang).
748		1063
749	In AEMP, you don't "look up" registered port names or send to named ports	1064	In AEMP, you don't "look up" registered port names or send to named ports
750	that might or might not be persistent. Instead, you normally spawn a port	1065	that might or might not be persistent. Instead, you normally spawn a port
751	on the remote node. The init function monitors the you, and you monitor	1066	on the remote node. The init function monitors you, and you monitor the
752	the remote port. Since both monitors are local to the node, they are much	1067	remote port. Since both monitors are local to the node, they are much more
753	more reliable.	1068	reliable (no need for C<spawn_link>).
754		1069
755	This also saves round-trips and avoids sending messages to the wrong port	1070	This also saves round-trips and avoids sending messages to the wrong port
756	(hard to do in Erlang).	1071	(hard to do in Erlang).
757		1072
758	=back	1073	=back
759		1074
		1075	=head1 RATIONALE
		1076
		1077	=over 4
		1078
		1079	=item Why strings for port and node IDs, why not objects?
		1080
		1081	We considered "objects", but found that the actual number of methods
		1082	that can be called are quite low. Since port and node IDs travel over
		1083	the network frequently, the serialising/deserialising would add lots of
		1084	overhead, as well as having to keep a proxy object everywhere.
		1085
		1086	Strings can easily be printed, easily serialised etc. and need no special
		1087	procedures to be "valid".
		1088
		1089	And as a result, a port with just a default receiver consists of a single
		1090	code reference stored in a global hash - it can't become much cheaper.
		1091
		1092	=item Why favour JSON, why not a real serialising format such as Storable?
		1093
		1094	In fact, any AnyEvent::MP node will happily accept Storable as framing
		1095	format, but currently there is no way to make a node use Storable by
		1096	default (although all nodes will accept it).
		1097
		1098	The default framing protocol is JSON because a) JSON::XS is many times
		1099	faster for small messages and b) most importantly, after years of
		1100	experience we found that object serialisation is causing more problems
		1101	than it solves: Just like function calls, objects simply do not travel
		1102	easily over the network, mostly because they will always be a copy, so you
		1103	always have to re-think your design.
		1104
		1105	Keeping your messages simple, concentrating on data structures rather than
		1106	objects, will keep your messages clean, tidy and efficient.
		1107
		1108	=back
		1109
760	=head1 SEE ALSO	1110	=head1 SEE ALSO
		1111
		1112	L<AnyEvent::MP::Intro> - a gentle introduction.
		1113
		1114	L<AnyEvent::MP::Kernel> - more, lower-level, stuff.
		1115
		1116	L<AnyEvent::MP::Global> - network maintenance and port groups, to find
		1117	your applications.
		1118
		1119	L<AnyEvent::MP::DataConn> - establish data connections between nodes.
		1120
		1121	L<AnyEvent::MP::LogCatcher> - simple service to display log messages from
		1122	all nodes.
761		1123
762	L<AnyEvent>.	1124	L<AnyEvent>.
763		1125
764	=head1 AUTHOR	1126	=head1 AUTHOR
765		1127

Diff Legend

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

Comparing AnyEvent-MP/MP.pm (file contents): Revision 1.37 by root, Fri Aug 7 16:47:23 2009 UTC vs. Revision 1.126 by root, Sat Mar 3 19:43:41 2012 UTC

Diff Legend

Comparing AnyEvent-MP/MP.pm (file contents):
Revision 1.37 by root, Fri Aug 7 16:47:23 2009 UTC vs.
Revision 1.126 by root, Sat Mar 3 19:43:41 2012 UTC