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

Comparing AnyEvent-MP/MP.pm (file contents):
Revision 1.63 by root, Thu Aug 27 21:29:37 2009 UTC vs.
Revision 1.142 by root, Fri Mar 23 13:44:01 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
13	$SELF # receiving/own port id in rcv callbacks	12	$SELF # receiving/own port id in rcv callbacks
14		13
15	# initialise the node so it can send/receive messages	14	# initialise the node so it can send/receive messages
16	initialise_node;	15	configure;
17		16
18	# ports are message endpoints	17	# ports are message destinations
19		18
20	# sending messages	19	# sending messages
21	snd $port, type => data...;	20	snd $port, type => data...;
22	snd $port, @msg;	21	snd $port, @msg;
23	snd @msg_with_first_element_being_a_port;	22	snd @msg_with_first_element_being_a_port;
24		23
25	# creating/using ports, the simple way	24	# creating/using ports, the simple way
26	my $simple_port = port { my @msg = @_; 0 };	25	my $simple_port = port { my @msg = @_ };
27		26
28	# creating/using ports, tagged message matching	27	# creating/using ports, tagged message matching
29	my $port = port;	28	my $port = port;
30	rcv $port, ping => sub { snd $_[0], "pong"; 0 };	29	rcv $port, ping => sub { snd $_[0], "pong" };
31	rcv $port, pong => sub { warn "pong received\n"; 0 };	30	rcv $port, pong => sub { warn "pong received\n" };
32		31
33	# create a port on another node	32	# create a port on another node
34	my $port = spawn $node, $initfunc, @initdata;	33	my $port = spawn $node, $initfunc, @initdata;
35		34
		35	# destroy a port again
		36	kil $port; # "normal" kill
		37	kil $port, my_error => "everything is broken"; # error kill
		38
36	# monitoring	39	# monitoring
37	mon $port, $cb->(@msg) # callback is invoked on death	40	mon $port, $cb->(@msg) # callback is invoked on death
38	mon $port, $otherport # kill otherport on abnormal death	41	mon $port, $localport # kill localport on abnormal death
39	mon $port, $otherport, @msg # send message on death	42	mon $port, $localport, @msg # send message on death
40		43
41	=head1 CURRENT STATUS	44	# temporarily execute code in port context
		45	peval $port, sub { die "kill the port!" };
42		46
43	AnyEvent::MP - stable API, should work	47	# execute callbacks in $SELF port context
44	AnyEvent::MP::Intro - outdated	48	my $timer = AE::timer 1, 0, psub {
45	AnyEvent::MP::Kernel - WIP	49	die "kill the port, delayed";
46	AnyEvent::MP::Transport - mostly stable	50	};
47
48	stay tuned.
49		51
50	=head1 DESCRIPTION	52	=head1 DESCRIPTION
51		53
52	This module (-family) implements a simple message passing framework.	54	This module (-family) implements a simple message passing framework.
53		55
54	Despite its simplicity, you can securely message other processes running	56	Despite its simplicity, you can securely message other processes running
55	on the same or other hosts.	57	on the same or other hosts, and you can supervise entities remotely.
56		58
57	For an introduction to this module family, see the L<AnyEvent::MP::Intro>	59	For an introduction to this module family, see the L<AnyEvent::MP::Intro>
58	manual page.	60	manual page and the examples under F<eg/>.
59
60	At the moment, this module family is severly broken and underdocumented,
61	so do not use. This was uploaded mainly to reserve the CPAN namespace -
62	stay tuned!
63		61
64	=head1 CONCEPTS	62	=head1 CONCEPTS
65		63
66	=over 4	64	=over 4
67		65
68	=item port	66	=item port
69		67
70	A port is something you can send messages to (with the C<snd> function).	68	Not to be confused with a TCP port, a "port" is something you can send
		69	messages to (with the C<snd> function).
71		70
72	Ports allow you to register C<rcv> handlers that can match all or just	71	Ports allow you to register C<rcv> handlers that can match all or just
73	some messages. Messages will not be queued.	72	some messages. Messages send to ports will not be queued, regardless of
		73	anything was listening for them or not.
74		74
		75	Ports are represented by (printable) strings called "port IDs".
		76
75	=item port ID - C<noderef#portname>	77	=item port ID - C<nodeid#portname>
76		78
77	A port ID is the concatenation of a noderef, a hash-mark (C<#>) as	79	A port ID is the concatenation of a node ID, a hash-mark (C<#>)
78	separator, and a port name (a printable string of unspecified format). An	80	as separator, and a port name (a printable string of unspecified
79	exception is the the node port, whose ID is identical to its node	81	format created by AnyEvent::MP).
80	reference.
81		82
82	=item node	83	=item node
83		84
84	A node is a single process containing at least one port - the node port,	85	A node is a single process containing at least one port - the node port,
85	which provides nodes to manage each other remotely, and to create new	86	which enables nodes to manage each other remotely, and to create new
86	ports.	87	ports.
87		88
88	Nodes are either private (single-process only), slaves (can only talk to	89	Nodes are either public (have one or more listening ports) or private
89	public nodes, but do not need an open port) or public nodes (connectable	90	(no listening ports). Private nodes cannot talk to other private nodes
90	from any other node).	91	currently, but all nodes can talk to public nodes.
91		92
		93	Nodes is represented by (printable) strings called "node IDs".
		94
92	=item node ID - C<[a-za-Z0-9_\-.:]+>	95	=item node ID - C<[A-Za-z0-9_\-.:]*>
93		96
94	A node ID is a string that either simply identifies the node (for	97	A node ID is a string that uniquely identifies the node within a
95	private and slave nodes), or contains a recipe on how to reach a given	98	network. Depending on the configuration used, node IDs can look like a
96	node (for public nodes).	99	hostname, a hostname and a port, or a random string. AnyEvent::MP itself
		100	doesn't interpret node IDs in any way except to uniquely identify a node.
97		101
98	This recipe is simply a comma-separated list of C<address:port> pairs (for	102	=item binds - C<ip:port>
99	TCP/IP, other protocols might look different).
100		103
101	Node references come in two flavours: resolved (containing only numerical	104	Nodes can only talk to each other by creating some kind of connection to
102	addresses) or unresolved (where hostnames are used instead of addresses).	105	each other. To do this, nodes should listen on one or more local transport
		106	endpoints - binds.
103		107
104	Before using an unresolved node reference in a message you first have to	108	Currently, only standard C<ip:port> specifications can be used, which
105	resolve it.	109	specify TCP ports to listen on. So a bind is basically just a tcp socket
		110	in listening mode thta accepts conenctions form other nodes.
		111
		112	=item seed nodes
		113
		114	When a node starts, it knows nothing about the network it is in - it
		115	needs to connect to at least one other node that is already in the
		116	network. These other nodes are called "seed nodes".
		117
		118	Seed nodes themselves are not special - they are seed nodes only because
		119	some other node I<uses> them as such, but any node can be used as seed
		120	node for other nodes, and eahc node cna use a different set of seed nodes.
		121
		122	In addition to discovering the network, seed nodes are also used to
		123	maintain the network - all nodes using the same seed node form are part of
		124	the same network. If a network is split into multiple subnets because e.g.
		125	the network link between the parts goes down, then using the same seed
		126	nodes for all nodes ensures that eventually the subnets get merged again.
		127
		128	Seed nodes are expected to be long-running, and at least one seed node
		129	should always be available. They should also be relatively responsive - a
		130	seed node that blocks for long periods will slow down everybody else.
		131
		132	For small networks, it's best if every node uses the same set of seed
		133	nodes. For large networks, it can be useful to specify "regional" seed
		134	nodes for most nodes in an area, and use all seed nodes as seed nodes for
		135	each other. What's important is that all seed nodes connections form a
		136	complete graph, so that the network cannot split into separate subnets
		137	forever.
		138
		139	Seed nodes are represented by seed IDs.
		140
		141	=item seed IDs - C<host:port>
		142
		143	Seed IDs are transport endpoint(s) (usually a hostname/IP address and a
		144	TCP port) of nodes that should be used as seed nodes.
		145
		146	=item global nodes
		147
		148	An AEMP network needs a discovery service - nodes need to know how to
		149	connect to other nodes they only know by name. In addition, AEMP offers a
		150	distributed "group database", which maps group names to a list of strings
		151	- for example, to register worker ports.
		152
		153	A network needs at least one global node to work, and allows every node to
		154	be a global node.
		155
		156	Any node that loads the L<AnyEvent::MP::Global> module becomes a global
		157	node and tries to keep connections to all other nodes. So while it can
		158	make sense to make every node "global" in small networks, it usually makes
		159	sense to only make seed nodes into global nodes in large networks (nodes
		160	keep connections to seed nodes and global nodes, so makign them the same
		161	reduces overhead).
106		162
107	=back	163	=back
108		164
109	=head1 VARIABLES/FUNCTIONS	165	=head1 VARIABLES/FUNCTIONS
110		166
112		168
113	=cut	169	=cut
114		170
115	package AnyEvent::MP;	171	package AnyEvent::MP;
116		172
		173	use AnyEvent::MP::Config ();
117	use AnyEvent::MP::Kernel;	174	use AnyEvent::MP::Kernel;
		175	use AnyEvent::MP::Kernel qw(%NODE %PORT %PORT_DATA $UNIQ $RUNIQ $ID);
118		176
119	use common::sense;	177	use common::sense;
120		178
121	use Carp ();	179	use Carp ();
122		180
123	use AE ();	181	use AnyEvent ();
		182	use Guard ();
124		183
125	use base "Exporter";	184	use base "Exporter";
126		185
127	our $VERSION = $AnyEvent::MP::Kernel::VERSION;	186	our $VERSION = $AnyEvent::MP::Config::VERSION;
128		187
129	our @EXPORT = qw(	188	our @EXPORT = qw(
130	NODE $NODE *SELF node_of after	189	NODE $NODE *SELF node_of after
131	resolve_node initialise_node	190	configure
132	snd rcv mon mon_guard kil reg psub spawn	191	snd rcv mon mon_guard kil psub peval spawn cal
133	port	192	port
		193	db_set db_del db_reg
		194	db_mon db_family db_keys db_values
134	);	195	);
135		196
136	our $SELF;	197	our $SELF;
137		198
138	sub _self_die() {	199	sub _self_die() {
…		…
141	kil $SELF, die => $msg;	202	kil $SELF, die => $msg;
142	}	203	}
143		204
144	=item $thisnode = NODE / $NODE	205	=item $thisnode = NODE / $NODE
145		206
146	The C<NODE> function returns, and the C<$NODE> variable contains the	207	The C<NODE> function returns, and the C<$NODE> variable contains, the node
147	node id of the local node. The value is initialised by a call to	208	ID of the node running in the current process. This value is initialised by
148	C<initialise_node>.	209	a call to C<configure>.
149		210
150	=item $nodeid = node_of $port	211	=item $nodeid = node_of $port
151		212
152	Extracts and returns the noderef from a port ID or a node ID.	213	Extracts and returns the node ID from a port ID or a node ID.
153		214
154	=item initialise_node $profile_name	215	=item configure $profile, key => value...
155		216
		217	=item configure key => value...
		218
156	Before a node can talk to other nodes on the network it has to initialise	219	Before a node can talk to other nodes on the network (i.e. enter
157	itself - the minimum a node needs to know is it's own name, and optionally	220	"distributed mode") it has to configure itself - the minimum a node needs
158	it should know the noderefs of some other nodes in the network.	221	to know is its own name, and optionally it should know the addresses of
		222	some other nodes in the network to discover other nodes.
159		223
160	This function initialises a node - it must be called exactly once (or	224	This function configures a node - it must be called exactly once (or
161	never) before calling other AnyEvent::MP functions.	225	never) before calling other AnyEvent::MP functions.
162		226
163	All arguments (optionally except for the first) are noderefs, which can be	227	The key/value pairs are basically the same ones as documented for the
164	either resolved or unresolved.	228	F<aemp> command line utility (sans the set/del prefix), with these additions:
165
166	The first argument will be looked up in the configuration database first
167	(if it is C<undef> then the current nodename will be used instead) to find
168	the relevant configuration profile (see L<aemp>). If none is found then
169	the default configuration is used. The configuration supplies additional
170	seed/master nodes and can override the actual noderef.
171
172	There are two types of networked nodes, public nodes and slave nodes:
173		229
174	=over 4	230	=over 4
175		231
176	=item public nodes	232	=item norc => $boolean (default false)
177		233
178	For public nodes, C<$noderef> (supplied either directly to	234	If true, then the rc file (e.g. F<~/.perl-anyevent-mp>) will I<not>
179	C<initialise_node> or indirectly via a profile or the nodename) must be a	235	be consulted - all configuraiton options must be specified in the
180	noderef (possibly unresolved, in which case it will be resolved).	236	C<configure> call.
181		237
182	After resolving, the node will bind itself on all endpoints.	238	=item force => $boolean (default false)
183		239
184	=item slave nodes	240	IF true, then the values specified in the C<configure> will take
185		241	precedence over any values configured via the rc file. The default is for
186	When the C<$noderef> (either as given or overriden by the config file)	242	the rc file to override any options specified in the program.
187	is the special string C<slave/>, then the node will become a slave
188	node. Slave nodes cannot be contacted from outside, and cannot talk to
189	each other (at least in this version of AnyEvent::MP).
190
191	Slave nodes work by creating connections to all public nodes, using the
192	L<AnyEvent::MP::Global> service.
193		243
194	=back	244	=back
195		245
196	After initialising itself, the node will connect to all additional
197	C<$seednodes> that are specified diretcly or via a profile. Seednodes are
198	optional and can be used to quickly bootstrap the node into an existing
199	network.
200
201	All the seednodes will also be specially marked to automatically retry
202	connecting to them indefinitely, so make sure that seednodes are really
203	reliable and up (this might also change in the future).
204
205	Example: become a public node listening on the guessed noderef, or the one
206	specified via C<aemp> for the current node. This should be the most common
207	form of invocation for "daemon"-type nodes.
208
209	initialise_node;
210
211	Example: become a slave node to any of the the seednodes specified via
212	C<aemp>. This form is often used for commandline clients.
213
214	initialise_node "slave/";
215
216	Example: become a public node, and try to contact some well-known master
217	servers to become part of the network.
218
219	initialise_node undef, "master1", "master2";
220
221	Example: become a public node listening on port C<4041>.
222
223	initialise_node 4041;
224
225	Example: become a public node, only visible on localhost port 4044.
226
227	initialise_node "localhost:4044";
228
229	=item $cv = resolve_node $noderef
230
231	Takes an unresolved node reference that may contain hostnames and
232	abbreviated IDs, resolves all of them and returns a resolved node
233	reference.
234
235	In addition to C<address:port> pairs allowed in resolved noderefs, the
236	following forms are supported:
237
238	=over 4	246	=over 4
239		247
240	=item the empty string	248	=item step 1, gathering configuration from profiles
241		249
242	An empty-string component gets resolved as if the default port (4040) was	250	The function first looks up a profile in the aemp configuration (see the
243	specified.	251	L<aemp> commandline utility). The profile name can be specified via the
		252	named C<profile> parameter or can simply be the first parameter). If it is
		253	missing, then the nodename (F<uname -n>) will be used as profile name.
244		254
245	=item naked port numbers (e.g. C<1234>)	255	The profile data is then gathered as follows:
246		256
247	These are resolved by prepending the local nodename and a colon, to be	257	First, all remaining key => value pairs (all of which are conveniently
248	further resolved.	258	undocumented at the moment) will be interpreted as configuration
		259	data. Then they will be overwritten by any values specified in the global
		260	default configuration (see the F<aemp> utility), then the chain of
		261	profiles chosen by the profile name (and any C<parent> attributes).
249		262
250	=item hostnames (e.g. C<localhost:1234>, C<localhost>)	263	That means that the values specified in the profile have highest priority
		264	and the values specified directly via C<configure> have lowest priority,
		265	and can only be used to specify defaults.
251		266
252	These are resolved by using AnyEvent::DNS to resolve them, optionally	267	If the profile specifies a node ID, then this will become the node ID of
253	looking up SRV records for the C<aemp=4040> port, if no port was	268	this process. If not, then the profile name will be used as node ID, with
254	specified.	269	a unique randoms tring (C</%u>) appended.
		270
		271	The node ID can contain some C<%> sequences that are expanded: C<%n>
		272	is expanded to the local nodename, C<%u> is replaced by a random
		273	strign to make the node unique. For example, the F<aemp> commandline
		274	utility uses C<aemp/%n/%u> as nodename, which might expand to
		275	C<aemp/cerebro/ZQDGSIkRhEZQDGSIkRhE>.
		276
		277	=item step 2, bind listener sockets
		278
		279	The next step is to look up the binds in the profile, followed by binding
		280	aemp protocol listeners on all binds specified (it is possible and valid
		281	to have no binds, meaning that the node cannot be contacted form the
		282	outside. This means the node cannot talk to other nodes that also have no
		283	binds, but it can still talk to all "normal" nodes).
		284
		285	If the profile does not specify a binds list, then a default of C<*> is
		286	used, meaning the node will bind on a dynamically-assigned port on every
		287	local IP address it finds.
		288
		289	=item step 3, connect to seed nodes
		290
		291	As the last step, the seed ID list from the profile is passed to the
		292	L<AnyEvent::MP::Global> module, which will then use it to keep
		293	connectivity with at least one node at any point in time.
255		294
256	=back	295	=back
		296
		297	Example: become a distributed node using the local node name as profile.
		298	This should be the most common form of invocation for "daemon"-type nodes.
		299
		300	configure
		301
		302	Example: become a semi-anonymous node. This form is often used for
		303	commandline clients.
		304
		305	configure nodeid => "myscript/%n/%u";
		306
		307	Example: configure a node using a profile called seed, which is suitable
		308	for a seed node as it binds on all local addresses on a fixed port (4040,
		309	customary for aemp).
		310
		311	# use the aemp commandline utility
		312	# aemp profile seed binds '*:4040'
		313
		314	# then use it
		315	configure profile => "seed";
		316
		317	# or simply use aemp from the shell again:
		318	# aemp run profile seed
		319
		320	# or provide a nicer-to-remember nodeid
		321	# aemp run profile seed nodeid "$(hostname)"
257		322
258	=item $SELF	323	=item $SELF
259		324
260	Contains the current port id while executing C<rcv> callbacks or C<psub>	325	Contains the current port id while executing C<rcv> callbacks or C<psub>
261	blocks.	326	blocks.
262		327
263	=item SELF, %SELF, @SELF...	328	=item *SELF, SELF, %SELF, @SELF...
264		329
265	Due to some quirks in how perl exports variables, it is impossible to	330	Due to some quirks in how perl exports variables, it is impossible to
266	just export C<$SELF>, all the symbols called C<SELF> are exported by this	331	just export C<$SELF>, all the symbols named C<SELF> are exported by this
267	module, but only C<$SELF> is currently used.	332	module, but only C<$SELF> is currently used.
268		333
269	=item snd $port, type => @data	334	=item snd $port, type => @data
270		335
271	=item snd $port, @msg	336	=item snd $port, @msg
272		337
273	Send the given message to the given port ID, which can identify either	338	Send the given message to the given port, which can identify either a
274	a local or a remote port, and must be a port ID.	339	local or a remote port, and must be a port ID.
275		340
276	While the message can be about anything, it is highly recommended to use a	341	While the message can be almost anything, it is highly recommended to
277	string as first element (a port ID, or some word that indicates a request	342	use a string as first element (a port ID, or some word that indicates a
278	type etc.).	343	request type etc.) and to consist if only simple perl values (scalars,
		344	arrays, hashes) - if you think you need to pass an object, think again.
279		345
280	The message data effectively becomes read-only after a call to this	346	The message data logically becomes read-only after a call to this
281	function: modifying any argument is not allowed and can cause many	347	function: modifying any argument (or values referenced by them) is
282	problems.	348	forbidden, as there can be considerable time between the call to C<snd>
		349	and the time the message is actually being serialised - in fact, it might
		350	never be copied as within the same process it is simply handed to the
		351	receiving port.
283		352
284	The type of data you can transfer depends on the transport protocol: when	353	The type of data you can transfer depends on the transport protocol: when
285	JSON is used, then only strings, numbers and arrays and hashes consisting	354	JSON is used, then only strings, numbers and arrays and hashes consisting
286	of those are allowed (no objects). When Storable is used, then anything	355	of those are allowed (no objects). When Storable is used, then anything
287	that Storable can serialise and deserialise is allowed, and for the local	356	that Storable can serialise and deserialise is allowed, and for the local
288	node, anything can be passed.	357	node, anything can be passed. Best rely only on the common denominator of
		358	these.
289		359
290	=item $local_port = port	360	=item $local_port = port
291		361
292	Create a new local port object and returns its port ID. Initially it has	362	Create a new local port object and returns its port ID. Initially it has
293	no callbacks set and will throw an error when it receives messages.	363	no callbacks set and will throw an error when it receives messages.
…		…
312		382
313	=cut	383	=cut
314		384
315	sub rcv($@);	385	sub rcv($@);
316		386
317	sub _kilme {	387	my $KILME = sub {
318	die "received message on port without callback";	388	(my $tag = substr $_[0], 0, 30) =~ s/([\x20-\x7e])/./g;
319	}	389	kil $SELF, unhandled_message => "no callback found for message '$tag'";
		390	};
320		391
321	sub port(;&) {	392	sub port(;&) {
322	my $id = "$UNIQ." . $ID++;	393	my $id = $UNIQ . ++$ID;
323	my $port = "$NODE#$id";	394	my $port = "$NODE#$id";
324		395
325	rcv $port, shift \|\| \&_kilme;	396	rcv $port, shift \|\| $KILME;
326		397
327	$port	398	$port
328	}	399	}
329		400
330	=item rcv $local_port, $callback->(@msg)	401	=item rcv $local_port, $callback->(@msg)
…		…
335		406
336	The global C<$SELF> (exported by this module) contains C<$port> while	407	The global C<$SELF> (exported by this module) contains C<$port> while
337	executing the callback. Runtime errors during callback execution will	408	executing the callback. Runtime errors during callback execution will
338	result in the port being C<kil>ed.	409	result in the port being C<kil>ed.
339		410
340	The default callback received all messages not matched by a more specific	411	The default callback receives all messages not matched by a more specific
341	C<tag> match.	412	C<tag> match.
342		413
343	=item rcv $local_port, tag => $callback->(@msg_without_tag), ...	414	=item rcv $local_port, tag => $callback->(@msg_without_tag), ...
344		415
345	Register (or replace) callbacks to be called on messages starting with the	416	Register (or replace) callbacks to be called on messages starting with the
…		…
366	msg1 => sub { ... },	437	msg1 => sub { ... },
367	...	438	...
368	;	439	;
369		440
370	Example: temporarily register a rcv callback for a tag matching some port	441	Example: temporarily register a rcv callback for a tag matching some port
371	(e.g. for a rpc reply) and unregister it after a message was received.	442	(e.g. for an rpc reply) and unregister it after a message was received.
372		443
373	rcv $port, $otherport => sub {	444	rcv $port, $otherport => sub {
374	my @reply = @_;	445	my @reply = @_;
375		446
376	rcv $SELF, $otherport;	447	rcv $SELF, $otherport;
…		…
378		449
379	=cut	450	=cut
380		451
381	sub rcv($@) {	452	sub rcv($@) {
382	my $port = shift;	453	my $port = shift;
383	my ($noderef, $portid) = split /#/, $port, 2;	454	my ($nodeid, $portid) = split /#/, $port, 2;
384		455
385	$NODE{$noderef} == $NODE{""}	456	$NODE{$nodeid} == $NODE{""}
386	or Carp::croak "$port: rcv can only be called on local ports, caught";	457	or Carp::croak "$port: rcv can only be called on local ports, caught";
387		458
388	while (@_) {	459	while (@_) {
389	if (ref $_[0]) {	460	if (ref $_[0]) {
390	if (my $self = $PORT_DATA{$portid}) {	461	if (my $self = $PORT_DATA{$portid}) {
391	"AnyEvent::MP::Port" eq ref $self	462	"AnyEvent::MP::Port" eq ref $self
392	or Carp::croak "$port: rcv can only be called on message matching ports, caught";	463	or Carp::croak "$port: rcv can only be called on message matching ports, caught";
393		464
394	$self->[2] = shift;	465	$self->[0] = shift;
395	} else {	466	} else {
396	my $cb = shift;	467	my $cb = shift;
397	$PORT{$portid} = sub {	468	$PORT{$portid} = sub {
398	local $SELF = $port;	469	local $SELF = $port;
399	eval { &$cb }; _self_die if $@;	470	eval { &$cb }; _self_die if $@;
400	};	471	};
401	}	472	}
402	} elsif (defined $_[0]) {	473	} elsif (defined $_[0]) {
403	my $self = $PORT_DATA{$portid} \|\|= do {	474	my $self = $PORT_DATA{$portid} \|\|= do {
404	my $self = bless [$PORT{$port} \|\| sub { }, { }, $port], "AnyEvent::MP::Port";	475	my $self = bless [$PORT{$portid} \|\| sub { }, { }, $port], "AnyEvent::MP::Port";
405		476
406	$PORT{$portid} = sub {	477	$PORT{$portid} = sub {
407	local $SELF = $port;	478	local $SELF = $port;
408		479
409	if (my $cb = $self->[1]{$_[0]}) {	480	if (my $cb = $self->[1]{$_[0]}) {
…		…
431	}	502	}
432		503
433	$port	504	$port
434	}	505	}
435		506
		507	=item peval $port, $coderef[, @args]
		508
		509	Evaluates the given C<$codref> within the contetx of C<$port>, that is,
		510	when the code throews an exception the C<$port> will be killed.
		511
		512	Any remaining args will be passed to the callback. Any return values will
		513	be returned to the caller.
		514
		515	This is useful when you temporarily want to execute code in the context of
		516	a port.
		517
		518	Example: create a port and run some initialisation code in it's context.
		519
		520	my $port = port { ... };
		521
		522	peval $port, sub {
		523	init
		524	or die "unable to init";
		525	};
		526
		527	=cut
		528
		529	sub peval($$) {
		530	local $SELF = shift;
		531	my $cb = shift;
		532
		533	if (wantarray) {
		534	my @res = eval { &$cb };
		535	_self_die if $@;
		536	@res
		537	} else {
		538	my $res = eval { &$cb };
		539	_self_die if $@;
		540	$res
		541	}
		542	}
		543
436	=item $closure = psub { BLOCK }	544	=item $closure = psub { BLOCK }
437		545
438	Remembers C<$SELF> and creates a closure out of the BLOCK. When the	546	Remembers C<$SELF> and creates a closure out of the BLOCK. When the
439	closure is executed, sets up the environment in the same way as in C<rcv>	547	closure is executed, sets up the environment in the same way as in C<rcv>
440	callbacks, i.e. runtime errors will cause the port to get C<kil>ed.	548	callbacks, i.e. runtime errors will cause the port to get C<kil>ed.
		549
		550	The effect is basically as if it returned C<< sub { peval $SELF, sub {
		551	BLOCK }, @_ } >>.
441		552
442	This is useful when you register callbacks from C<rcv> callbacks:	553	This is useful when you register callbacks from C<rcv> callbacks:
443		554
444	rcv delayed_reply => sub {	555	rcv delayed_reply => sub {
445	my ($delay, @reply) = @_;	556	my ($delay, @reply) = @_;
…		…
469	$res	580	$res
470	}	581	}
471	}	582	}
472	}	583	}
473		584
474	=item $guard = mon $port, $cb->(@reason)	585	=item $guard = mon $port, $rcvport # kill $rcvport when $port dies
475		586
476	=item $guard = mon $port, $rcvport	587	=item $guard = mon $port # kill $SELF when $port dies
477		588
478	=item $guard = mon $port	589	=item $guard = mon $port, $cb->(@reason) # call $cb when $port dies
479		590
480	=item $guard = mon $port, $rcvport, @msg	591	=item $guard = mon $port, $rcvport, @msg # send a message when $port dies
481		592
482	Monitor the given port and do something when the port is killed or	593	Monitor the given port and do something when the port is killed or
483	messages to it were lost, and optionally return a guard that can be used	594	messages to it were lost, and optionally return a guard that can be used
484	to stop monitoring again.	595	to stop monitoring again.
485		596
		597	The first two forms distinguish between "normal" and "abnormal" kil's:
		598
		599	In the first form (another port given), if the C<$port> is C<kil>'ed with
		600	a non-empty reason, the other port (C<$rcvport>) will be kil'ed with the
		601	same reason. That is, on "normal" kil's nothing happens, while under all
		602	other conditions, the other port is killed with the same reason.
		603
		604	The second form (kill self) is the same as the first form, except that
		605	C<$rvport> defaults to C<$SELF>.
		606
		607	The remaining forms don't distinguish between "normal" and "abnormal" kil's
		608	- it's up to the callback or receiver to check whether the C<@reason> is
		609	empty and act accordingly.
		610
		611	In the third form (callback), the callback is simply called with any
		612	number of C<@reason> elements (empty @reason means that the port was deleted
		613	"normally"). Note also that I<< the callback B<must> never die >>, so use
		614	C<eval> if unsure.
		615
		616	In the last form (message), a message of the form C<$rcvport, @msg,
		617	@reason> will be 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, even if it
		621	turns out that the port is still alive.
		622
		623	As a rule of thumb, monitoring requests should always monitor a remote
		624	port locally (using a local C<$rcvport> or a callback). The reason is that
		625	kill messages might get lost, just like any other message. Another less
		626	obvious reason is that even monitoring requests can get lost (for example,
		627	when the connection to the other node goes down permanently). When
		628	monitoring a port locally these problems do not exist.
		629
486	C<mon> effectively guarantees that, in the absence of hardware failures,	630	C<mon> effectively guarantees that, in the absence of hardware failures,
487	that after starting the monitor, either all messages sent to the port	631	after starting the monitor, either all messages sent to the port will
488	will arrive, or the monitoring action will be invoked after possible	632	arrive, or the monitoring action will be invoked after possible message
489	message loss has been detected. No messages will be lost "in between"	633	loss has been detected. No messages will be lost "in between" (after
490	(after the first lost message no further messages will be received by the	634	the first lost message no further messages will be received by the
491	port). After the monitoring action was invoked, further messages might get	635	port). After the monitoring action was invoked, further messages might get
492	delivered again.	636	delivered again.
493		637
494	Note that monitoring-actions are one-shot: once released, they are removed	638	Inter-host-connection timeouts and monitoring depend on the transport
495	and will not trigger again.	639	used. The only transport currently implemented is TCP, and AnyEvent::MP
		640	relies on TCP to detect node-downs (this can take 10-15 minutes on a
		641	non-idle connection, and usually around two hours for idle connections).
496		642
497	In the first form (callback), the callback is simply called with any	643	This means that monitoring is good for program errors and cleaning up
498	number of C<@reason> elements (no @reason means that the port was deleted	644	stuff eventually, but they are no replacement for a timeout when you need
499	"normally"). Note also that I<< the callback B<must> never die >>, so use	645	to ensure some maximum latency.
500	C<eval> if unsure.
501
502	In the second form (another port given), the other port (C<$rcvport>)
503	will be C<kil>'ed with C<@reason>, iff a @reason was specified, i.e. on
504	"normal" kils nothing happens, while under all other conditions, the other
505	port is killed with the same reason.
506
507	The third form (kill self) is the same as the second form, except that
508	C<$rvport> defaults to C<$SELF>.
509
510	In the last form (message), a message of the form C<@msg, @reason> will be
511	C<snd>.
512
513	As a rule of thumb, monitoring requests should always monitor a port from
514	a local port (or callback). The reason is that kill messages might get
515	lost, just like any other message. Another less obvious reason is that
516	even monitoring requests can get lost (for exmaple, when the connection
517	to the other node goes down permanently). When monitoring a port locally
518	these problems do not exist.
519		646
520	Example: call a given callback when C<$port> is killed.	647	Example: call a given callback when C<$port> is killed.
521		648
522	mon $port, sub { warn "port died because of <@_>\n" };	649	mon $port, sub { warn "port died because of <@_>\n" };
523		650
…		…
530	mon $port, $self => "restart";	657	mon $port, $self => "restart";
531		658
532	=cut	659	=cut
533		660
534	sub mon {	661	sub mon {
535	my ($noderef, $port) = split /#/, shift, 2;	662	my ($nodeid, $port) = split /#/, shift, 2;
536		663
537	my $node = $NODE{$noderef} \|\| add_node $noderef;	664	my $node = $NODE{$nodeid} \|\| add_node $nodeid;
538		665
539	my $cb = @_ ? shift : $SELF \|\| Carp::croak 'mon: called with one argument only, but $SELF not set,';	666	my $cb = @_ ? shift : $SELF \|\| Carp::croak 'mon: called with one argument only, but $SELF not set,';
540		667
541	unless (ref $cb) {	668	unless (ref $cb) {
542	if (@_) {	669	if (@_) {
…		…
551	}	678	}
552		679
553	$node->monitor ($port, $cb);	680	$node->monitor ($port, $cb);
554		681
555	defined wantarray	682	defined wantarray
556	and AnyEvent::Util::guard { $node->unmonitor ($port, $cb) }	683	and ($cb += 0, Guard::guard { $node->unmonitor ($port, $cb) })
557	}	684	}
558		685
559	=item $guard = mon_guard $port, $ref, $ref...	686	=item $guard = mon_guard $port, $ref, $ref...
560		687
561	Monitors the given C<$port> and keeps the passed references. When the port	688	Monitors the given C<$port> and keeps the passed references. When the port
562	is killed, the references will be freed.	689	is killed, the references will be freed.
563		690
564	Optionally returns a guard that will stop the monitoring.	691	Optionally returns a guard that will stop the monitoring.
565		692
566	This function is useful when you create e.g. timers or other watchers and	693	This function is useful when you create e.g. timers or other watchers and
567	want to free them when the port gets killed:	694	want to free them when the port gets killed (note the use of C<psub>):
568		695
569	$port->rcv (start => sub {	696	$port->rcv (start => sub {
570	my $timer; $timer = mon_guard $port, AE::timer 1, 1, sub {	697	my $timer; $timer = mon_guard $port, AE::timer 1, 1, psub {
571	undef $timer if 0.9 < rand;	698	undef $timer if 0.9 < rand;
572	});	699	});
573	});	700	});
574		701
575	=cut	702	=cut
…		…
584		711
585	=item kil $port[, @reason]	712	=item kil $port[, @reason]
586		713
587	Kill the specified port with the given C<@reason>.	714	Kill the specified port with the given C<@reason>.
588		715
589	If no C<@reason> is specified, then the port is killed "normally" (linked	716	If no C<@reason> is specified, then the port is killed "normally" -
590	ports will not be kileld, or even notified).	717	monitor callback will be invoked, but the kil will not cause linked ports
		718	(C<mon $mport, $lport> form) to get killed.
591		719
592	Otherwise, linked ports get killed with the same reason (second form of	720	If a C<@reason> is specified, then linked ports (C<mon $mport, $lport>
593	C<mon>, see below).	721	form) get killed with the same reason.
594		722
595	Runtime errors while evaluating C<rcv> callbacks or inside C<psub> blocks	723	Runtime errors while evaluating C<rcv> callbacks or inside C<psub> blocks
596	will be reported as reason C<< die => $@ >>.	724	will be reported as reason C<< die => $@ >>.
597		725
598	Transport/communication errors are reported as C<< transport_error =>	726	Transport/communication errors are reported as C<< transport_error =>
599	$message >>.	727	$message >>.
600		728
601	=cut	729	Common idioms:
		730
		731	# silently remove yourself, do not kill linked ports
		732	kil $SELF;
		733
		734	# report a failure in some detail
		735	kil $SELF, failure_mode_1 => "it failed with too high temperature";
		736
		737	# do not waste much time with killing, just die when something goes wrong
		738	open my $fh, "<file"
		739	or die "file: $!";
602		740
603	=item $port = spawn $node, $initfunc[, @initdata]	741	=item $port = spawn $node, $initfunc[, @initdata]
604		742
605	Creates a port on the node C<$node> (which can also be a port ID, in which	743	Creates a port on the node C<$node> (which can also be a port ID, in which
606	case it's the node where that port resides).	744	case it's the node where that port resides).
607		745
608	The port ID of the newly created port is return immediately, and it is	746	The port ID of the newly created port is returned immediately, and it is
609	permissible to immediately start sending messages or monitor the port.	747	possible to immediately start sending messages or to monitor the port.
610		748
611	After the port has been created, the init function is	749	After the port has been created, the init function is called on the remote
612	called. This function must be a fully-qualified function name	750	node, in the same context as a C<rcv> callback. This function must be a
613	(e.g. C<MyApp::Chat::Server::init>). To specify a function in the main	751	fully-qualified function name (e.g. C<MyApp::Chat::Server::init>). To
614	program, use C<::name>.	752	specify a function in the main program, use C<::name>.
615		753
616	If the function doesn't exist, then the node tries to C<require>	754	If the function doesn't exist, then the node tries to C<require>
617	the package, then the package above the package and so on (e.g.	755	the package, then the package above the package and so on (e.g.
618	C<MyApp::Chat::Server>, C<MyApp::Chat>, C<MyApp>) until the function	756	C<MyApp::Chat::Server>, C<MyApp::Chat>, C<MyApp>) until the function
619	exists or it runs out of package names.	757	exists or it runs out of package names.
620		758
621	The init function is then called with the newly-created port as context	759	The init function is then called with the newly-created port as context
622	object (C<$SELF>) and the C<@initdata> values as arguments.	760	object (C<$SELF>) and the C<@initdata> values as arguments. It I<must>
		761	call one of the C<rcv> functions to set callbacks on C<$SELF>, otherwise
		762	the port might not get created.
623		763
624	A common idiom is to pass your own port, monitor the spawned port, and	764	A common idiom is to pass a local port, immediately monitor the spawned
625	in the init function, monitor the original port. This two-way monitoring	765	port, and in the remote init function, immediately monitor the passed
626	ensures that both ports get cleaned up when there is a problem.	766	local port. This two-way monitoring ensures that both ports get cleaned up
		767	when there is a problem.
		768
		769	C<spawn> guarantees that the C<$initfunc> has no visible effects on the
		770	caller before C<spawn> returns (by delaying invocation when spawn is
		771	called for the local node).
627		772
628	Example: spawn a chat server port on C<$othernode>.	773	Example: spawn a chat server port on C<$othernode>.
629		774
630	# this node, executed from within a port context:	775	# this node, executed from within a port context:
631	my $server = spawn $othernode, "MyApp::Chat::Server::connect", $SELF;	776	my $server = spawn $othernode, "MyApp::Chat::Server::connect", $SELF;
…		…
646		791
647	sub _spawn {	792	sub _spawn {
648	my $port = shift;	793	my $port = shift;
649	my $init = shift;	794	my $init = shift;
650		795
		796	# rcv will create the actual port
651	local $SELF = "$NODE#$port";	797	local $SELF = "$NODE#$port";
652	eval {	798	eval {
653	&{ load_func $init }	799	&{ load_func $init }
654	};	800	};
655	_self_die if $@;	801	_self_die if $@;
656	}	802	}
657		803
658	sub spawn(@) {	804	sub spawn(@) {
659	my ($noderef, undef) = split /#/, shift, 2;	805	my ($nodeid, undef) = split /#/, shift, 2;
660		806
661	my $id = "$RUNIQ." . $ID++;	807	my $id = $RUNIQ . ++$ID;
662		808
663	$_[0] =~ /::/	809	$_[0] =~ /::/
664	or Carp::croak "spawn init function must be a fully-qualified name, caught";	810	or Carp::croak "spawn init function must be a fully-qualified name, caught";
665		811
666	snd_to_func $noderef, "AnyEvent::MP::_spawn" => $id, @_;	812	snd_to_func $nodeid, "AnyEvent::MP::_spawn" => $id, @_;
667		813
668	"$noderef#$id"	814	"$nodeid#$id"
669	}	815	}
		816
670		817
671	=item after $timeout, @msg	818	=item after $timeout, @msg
672		819
673	=item after $timeout, $callback	820	=item after $timeout, $callback
674		821
675	Either sends the given message, or call the given callback, after the	822	Either sends the given message, or call the given callback, after the
676	specified number of seconds.	823	specified number of seconds.
677		824
678	This is simply a utility function that come sin handy at times.	825	This is simply a utility function that comes in handy at times - the
		826	AnyEvent::MP author is not convinced of the wisdom of having it, though,
		827	so it may go away in the future.
679		828
680	=cut	829	=cut
681		830
682	sub after($@) {	831	sub after($@) {
683	my ($timeout, @action) = @_;	832	my ($timeout, @action) = @_;
…		…
688	? $action[0]()	837	? $action[0]()
689	: snd @action;	838	: snd @action;
690	};	839	};
691	}	840	}
692		841
		842	#=item $cb2 = timeout $seconds, $cb[, @args]
		843
		844	=item cal $port, @msg, $callback[, $timeout]
		845
		846	A simple form of RPC - sends a message to the given C<$port> with the
		847	given contents (C<@msg>), but adds a reply port to the message.
		848
		849	The reply port is created temporarily just for the purpose of receiving
		850	the reply, and will be C<kil>ed when no longer needed.
		851
		852	A reply message sent to the port is passed to the C<$callback> as-is.
		853
		854	If an optional time-out (in seconds) is given and it is not C<undef>,
		855	then the callback will be called without any arguments after the time-out
		856	elapsed and the port is C<kil>ed.
		857
		858	If no time-out is given (or it is C<undef>), then the local port will
		859	monitor the remote port instead, so it eventually gets cleaned-up.
		860
		861	Currently this function returns the temporary port, but this "feature"
		862	might go in future versions unless you can make a convincing case that
		863	this is indeed useful for something.
		864
		865	=cut
		866
		867	sub cal(@) {
		868	my $timeout = ref $_[-1] ? undef : pop;
		869	my $cb = pop;
		870
		871	my $port = port {
		872	undef $timeout;
		873	kil $SELF;
		874	&$cb;
		875	};
		876
		877	if (defined $timeout) {
		878	$timeout = AE::timer $timeout, 0, sub {
		879	undef $timeout;
		880	kil $port;
		881	$cb->();
		882	};
		883	} else {
		884	mon $_[0], sub {
		885	kil $port;
		886	$cb->();
		887	};
		888	}
		889
		890	push @_, $port;
		891	&snd;
		892
		893	$port
		894	}
		895
		896	=back
		897
		898	=head1 DISTRIBUTED DATABASE
		899
		900	AnyEvent::MP comes with a simple distributed database. The database will
		901	be mirrored asynchronously on all global nodes. Other nodes bind to one
		902	of the global nodes for their needs. Every node has a "local database"
		903	which contains all the values that are set locally. All local databases
		904	are merged together to form the global database, which can be queried.
		905
		906	The database structure is that of a two-level hash - the database hash
		907	contains hashes which contain values, similarly to a perl hash of hashes,
		908	i.e.:
		909
		910	$DATABASE{$family}{$subkey} = $value
		911
		912	The top level hash key is called "family", and the second-level hash key
		913	is called "subkey" or simply "key".
		914
		915	The family must be alphanumeric, i.e. start with a letter and consist
		916	of letters, digits, underscores and colons (C<[A-Za-z][A-Za-z0-9_:]*>,
		917	pretty much like Perl module names.
		918
		919	As the family namespace is global, it is recommended to prefix family names
		920	with the name of the application or module using it.
		921
		922	The subkeys must be non-empty strings, with no further restrictions.
		923
		924	The values should preferably be strings, but other perl scalars should
		925	work as well (such as C<undef>, arrays and hashes).
		926
		927	Every database entry is owned by one node - adding the same family/subkey
		928	combination on multiple nodes will not cause discomfort for AnyEvent::MP,
		929	but the result might be nondeterministic, i.e. the key might have
		930	different values on different nodes.
		931
		932	Different subkeys in the same family can be owned by different nodes
		933	without problems, and in fact, this is the common method to create worker
		934	pools. For example, a worker port for image scaling might do this:
		935
		936	db_set my_image_scalers => $port;
		937
		938	And clients looking for an image scaler will want to get the
		939	C<my_image_scalers> keys from time to time:
		940
		941	db_keys my_image_scalers => sub {
		942	@ports = @{ $_[0] };
		943	};
		944
		945	Or better yet, they want to monitor the database family, so they always
		946	have a reasonable up-to-date copy:
		947
		948	db_mon my_image_scalers => sub {
		949	@ports = keys %{ $_[0] };
		950	};
		951
		952	In general, you can set or delete single subkeys, but query and monitor
		953	whole families only.
		954
		955	If you feel the need to monitor or query a single subkey, try giving it
		956	it's own family.
		957
		958	=over
		959
		960	=item $guard = db_set $family => $subkey [=> $value]
		961
		962	Sets (or replaces) a key to the database - if C<$value> is omitted,
		963	C<undef> is used instead.
		964
		965	When called in non-void context, C<db_set> returns a guard that
		966	automatically calls C<db_del> when it is destroyed.
		967
		968	=item db_del $family => $subkey...
		969
		970	Deletes one or more subkeys from the database family.
		971
		972	=item $guard = db_reg $family => $port => $value
		973
		974	=item $guard = db_reg $family => $port
		975
		976	=item $guard = db_reg $family
		977
		978	Registers a port in the given family and optionally returns a guard to
		979	remove it.
		980
		981	This function basically does the same as:
		982
		983	db_set $family => $port => $value
		984
		985	Except that the port is monitored and automatically removed from the
		986	database family when it is kil'ed.
		987
		988	If C<$value> is missing, C<undef> is used. If C<$port> is missing, then
		989	C<$SELF> is used.
		990
		991	This function is most useful to register a port in some port group (which
		992	is just another name for a database family), and have it removed when the
		993	port is gone. This works best when the port is a local port.
		994
		995	=cut
		996
		997	sub db_reg($$;$) {
		998	my $family = shift;
		999	my $port = @_ ? shift : $SELF;
		1000
		1001	my $clr = sub { db_del $family => $port };
		1002	mon $port, $clr;
		1003
		1004	db_set $family => $port => $_[0];
		1005
		1006	defined wantarray
		1007	and &Guard::guard ($clr)
		1008	}
		1009
		1010	=item db_family $family => $cb->(\%familyhash)
		1011
		1012	Queries the named database C<$family> and call the callback with the
		1013	family represented as a hash. You can keep and freely modify the hash.
		1014
		1015	=item db_keys $family => $cb->(\@keys)
		1016
		1017	Same as C<db_family>, except it only queries the family I<subkeys> and passes
		1018	them as array reference to the callback.
		1019
		1020	=item db_values $family => $cb->(\@values)
		1021
		1022	Same as C<db_family>, except it only queries the family I<values> and passes them
		1023	as array reference to the callback.
		1024
		1025	=item $guard = db_mon $family => $cb->($familyhash, \@added, \@changed, \@deleted)
		1026
		1027	Creates a monitor on the given database family. Each time a key is set
		1028	or or is deleted the callback is called with a hash containing the
		1029	database family and three lists of added, changed and deleted subkeys,
		1030	respectively. If no keys have changed then the array reference might be
		1031	C<undef> or even missing.
		1032
		1033	If not called in void context, a guard object is returned that, when
		1034	destroyed, stops the monitor.
		1035
		1036	The family hash reference and the key arrays belong to AnyEvent::MP and
		1037	B<must not be modified or stored> by the callback. When in doubt, make a
		1038	copy.
		1039
		1040	As soon as possible after the monitoring starts, the callback will be
		1041	called with the intiial contents of the family, even if it is empty,
		1042	i.e. there will always be a timely call to the callback with the current
		1043	contents.
		1044
		1045	It is possible that the callback is called with a change event even though
		1046	the subkey is already present and the value has not changed.
		1047
		1048	The monitoring stops when the guard object is destroyed.
		1049
		1050	Example: on every change to the family "mygroup", print out all keys.
		1051
		1052	my $guard = db_mon mygroup => sub {
		1053	my ($family, $a, $c, $d) = @_;
		1054	print "mygroup members: ", (join " ", keys %$family), "\n";
		1055	};
		1056
		1057	Exmaple: wait until the family "My::Module::workers" is non-empty.
		1058
		1059	my $guard; $guard = db_mon My::Module::workers => sub {
		1060	my ($family, $a, $c, $d) = @_;
		1061	return unless %$family;
		1062	undef $guard;
		1063	print "My::Module::workers now nonempty\n";
		1064	};
		1065
		1066	Example: print all changes to the family "AnyRvent::Fantasy::Module".
		1067
		1068	my $guard = db_mon AnyRvent::Fantasy::Module => sub {
		1069	my ($family, $a, $c, $d) = @_;
		1070
		1071	print "+$_=$family->{$_}\n" for @$a;
		1072	print "*$_=$family->{$_}\n" for @$c;
		1073	print "-$_=$family->{$_}\n" for @$d;
		1074	};
		1075
		1076	=cut
		1077
693	=back	1078	=back
694		1079
695	=head1 AnyEvent::MP vs. Distributed Erlang	1080	=head1 AnyEvent::MP vs. Distributed Erlang
696		1081
697	AnyEvent::MP got lots of its ideas from distributed Erlang (Erlang node	1082	AnyEvent::MP got lots of its ideas from distributed Erlang (Erlang node
698	== aemp node, Erlang process == aemp port), so many of the documents and	1083	== aemp node, Erlang process == aemp port), so many of the documents and
699	programming techniques employed by Erlang apply to AnyEvent::MP. Here is a	1084	programming techniques employed by Erlang apply to AnyEvent::MP. Here is a
700	sample:	1085	sample:
701		1086
702	http://www.Erlang.se/doc/programming_rules.shtml	1087	http://www.erlang.se/doc/programming_rules.shtml
703	http://Erlang.org/doc/getting_started/part_frame.html # chapters 3 and 4	1088	http://erlang.org/doc/getting_started/part_frame.html # chapters 3 and 4
704	http://Erlang.org/download/Erlang-book-part1.pdf # chapters 5 and 6	1089	http://erlang.org/download/erlang-book-part1.pdf # chapters 5 and 6
705	http://Erlang.org/download/armstrong_thesis_2003.pdf # chapters 4 and 5	1090	http://erlang.org/download/armstrong_thesis_2003.pdf # chapters 4 and 5
706		1091
707	Despite the similarities, there are also some important differences:	1092	Despite the similarities, there are also some important differences:
708		1093
709	=over 4	1094	=over 4
710		1095
711	=item * Node references contain the recipe on how to contact them.	1096	=item * Node IDs are arbitrary strings in AEMP.
712		1097
713	Erlang relies on special naming and DNS to work everywhere in the	1098	Erlang relies on special naming and DNS to work everywhere in the same
714	same way. AEMP relies on each node knowing it's own address(es), with	1099	way. AEMP relies on each node somehow knowing its own address(es) (e.g. by
715	convenience functionality.	1100	configuration or DNS), and possibly the addresses of some seed nodes, but
716		1101	will otherwise discover other nodes (and their IDs) itself.
717	This means that AEMP requires a less tightly controlled environment at the
718	cost of longer node references and a slightly higher management overhead.
719		1102
720	=item * Erlang has a "remote ports are like local ports" philosophy, AEMP	1103	=item * Erlang has a "remote ports are like local ports" philosophy, AEMP
721	uses "local ports are like remote ports".	1104	uses "local ports are like remote ports".
722		1105
723	The failure modes for local ports are quite different (runtime errors	1106	The failure modes for local ports are quite different (runtime errors
…		…
732	ports being the special case/exception, where transport errors cannot	1115	ports being the special case/exception, where transport errors cannot
733	occur.	1116	occur.
734		1117
735	=item * Erlang uses processes and a mailbox, AEMP does not queue.	1118	=item * Erlang uses processes and a mailbox, AEMP does not queue.
736		1119
737	Erlang uses processes that selectively receive messages, and therefore	1120	Erlang uses processes that selectively receive messages out of order, and
738	needs a queue. AEMP is event based, queuing messages would serve no	1121	therefore needs a queue. AEMP is event based, queuing messages would serve
739	useful purpose. For the same reason the pattern-matching abilities of	1122	no useful purpose. For the same reason the pattern-matching abilities
740	AnyEvent::MP are more limited, as there is little need to be able to	1123	of AnyEvent::MP are more limited, as there is little need to be able to
741	filter messages without dequeing them.	1124	filter messages without dequeuing them.
742		1125
743	(But see L<Coro::MP> for a more Erlang-like process model on top of AEMP).	1126	This is not a philosophical difference, but simply stems from AnyEvent::MP
		1127	being event-based, while Erlang is process-based.
		1128
		1129	You cna have a look at L<Coro::MP> for a more Erlang-like process model on
		1130	top of AEMP and Coro threads.
744		1131
745	=item * Erlang sends are synchronous, AEMP sends are asynchronous.	1132	=item * Erlang sends are synchronous, AEMP sends are asynchronous.
746		1133
747	Sending messages in Erlang is synchronous and blocks the process (and	1134	Sending messages in Erlang is synchronous and blocks the process until
		1135	a conenction has been established and the message sent (and so does not
748	so does not need a queue that can overflow). AEMP sends are immediate,	1136	need a queue that can overflow). AEMP sends return immediately, connection
749	connection establishment is handled in the background.	1137	establishment is handled in the background.
750		1138
751	=item * Erlang suffers from silent message loss, AEMP does not.	1139	=item * Erlang suffers from silent message loss, AEMP does not.
752		1140
753	Erlang makes few guarantees on messages delivery - messages can get lost	1141	Erlang implements few guarantees on messages delivery - messages can get
754	without any of the processes realising it (i.e. you send messages a, b,	1142	lost without any of the processes realising it (i.e. you send messages a,
755	and c, and the other side only receives messages a and c).	1143	b, and c, and the other side only receives messages a and c).
756		1144
757	AEMP guarantees correct ordering, and the guarantee that there are no	1145	AEMP guarantees (modulo hardware errors) correct ordering, and the
		1146	guarantee that after one message is lost, all following ones sent to the
		1147	same port are lost as well, until monitoring raises an error, so there are
758	holes in the message sequence.	1148	no silent "holes" in the message sequence.
759		1149
760	=item * In Erlang, processes can be declared dead and later be found to be	1150	If you want your software to be very reliable, you have to cope with
761	alive.	1151	corrupted and even out-of-order messages in both Erlang and AEMP. AEMP
762		1152	simply tries to work better in common error cases, such as when a network
763	In Erlang it can happen that a monitored process is declared dead and	1153	link goes down.
764	linked processes get killed, but later it turns out that the process is
765	still alive - and can receive messages.
766
767	In AEMP, when port monitoring detects a port as dead, then that port will
768	eventually be killed - it cannot happen that a node detects a port as dead
769	and then later sends messages to it, finding it is still alive.
770		1154
771	=item * Erlang can send messages to the wrong port, AEMP does not.	1155	=item * Erlang can send messages to the wrong port, AEMP does not.
772		1156
773	In Erlang it is quite likely that a node that restarts reuses a process ID	1157	In Erlang it is quite likely that a node that restarts reuses an Erlang
774	known to other nodes for a completely different process, causing messages	1158	process ID known to other nodes for a completely different process,
775	destined for that process to end up in an unrelated process.	1159	causing messages destined for that process to end up in an unrelated
		1160	process.
776		1161
777	AEMP never reuses port IDs, so old messages or old port IDs floating	1162	AEMP does not reuse port IDs, so old messages or old port IDs floating
778	around in the network will not be sent to an unrelated port.	1163	around in the network will not be sent to an unrelated port.
779		1164
780	=item * Erlang uses unprotected connections, AEMP uses secure	1165	=item * Erlang uses unprotected connections, AEMP uses secure
781	authentication and can use TLS.	1166	authentication and can use TLS.
782		1167
783	AEMP can use a proven protocol - SSL/TLS - to protect connections and	1168	AEMP can use a proven protocol - TLS - to protect connections and
784	securely authenticate nodes.	1169	securely authenticate nodes.
785		1170
786	=item * The AEMP protocol is optimised for both text-based and binary	1171	=item * The AEMP protocol is optimised for both text-based and binary
787	communications.	1172	communications.
788		1173
789	The AEMP protocol, unlike the Erlang protocol, supports both	1174	The AEMP protocol, unlike the Erlang protocol, supports both programming
790	language-independent text-only protocols (good for debugging) and binary,	1175	language independent text-only protocols (good for debugging), and binary,
791	language-specific serialisers (e.g. Storable).	1176	language-specific serialisers (e.g. Storable). By default, unless TLS is
		1177	used, the protocol is actually completely text-based.
792		1178
793	It has also been carefully designed to be implementable in other languages	1179	It has also been carefully designed to be implementable in other languages
794	with a minimum of work while gracefully degrading fucntionality to make the	1180	with a minimum of work while gracefully degrading functionality to make the
795	protocol simple.	1181	protocol simple.
796		1182
797	=item * AEMP has more flexible monitoring options than Erlang.	1183	=item * AEMP has more flexible monitoring options than Erlang.
798		1184
799	In Erlang, you can chose to receive I<all> exit signals as messages	1185	In Erlang, you can chose to receive I<all> exit signals as messages or
800	or I<none>, there is no in-between, so monitoring single processes is	1186	I<none>, there is no in-between, so monitoring single Erlang processes is
801	difficult to implement. Monitoring in AEMP is more flexible than in	1187	difficult to implement.
802	Erlang, as one can choose between automatic kill, exit message or callback	1188
803	on a per-process basis.	1189	Monitoring in AEMP is more flexible than in Erlang, as one can choose
		1190	between automatic kill, exit message or callback on a per-port basis.
804		1191
805	=item * Erlang tries to hide remote/local connections, AEMP does not.	1192	=item * Erlang tries to hide remote/local connections, AEMP does not.
806		1193
807	Monitoring in Erlang is not an indicator of process death/crashes,	1194	Monitoring in Erlang is not an indicator of process death/crashes, in the
808	as linking is (except linking is unreliable in Erlang).	1195	same way as linking is (except linking is unreliable in Erlang).
809		1196
810	In AEMP, you don't "look up" registered port names or send to named ports	1197	In AEMP, you don't "look up" registered port names or send to named ports
811	that might or might not be persistent. Instead, you normally spawn a port	1198	that might or might not be persistent. Instead, you normally spawn a port
812	on the remote node. The init function monitors the you, and you monitor	1199	on the remote node. The init function monitors you, and you monitor the
813	the remote port. Since both monitors are local to the node, they are much	1200	remote port. Since both monitors are local to the node, they are much more
814	more reliable.	1201	reliable (no need for C<spawn_link>).
815		1202
816	This also saves round-trips and avoids sending messages to the wrong port	1203	This also saves round-trips and avoids sending messages to the wrong port
817	(hard to do in Erlang).	1204	(hard to do in Erlang).
818		1205
819	=back	1206	=back
820		1207
821	=head1 RATIONALE	1208	=head1 RATIONALE
822		1209
823	=over 4	1210	=over 4
824		1211
825	=item Why strings for ports and noderefs, why not objects?	1212	=item Why strings for port and node IDs, why not objects?
826		1213
827	We considered "objects", but found that the actual number of methods	1214	We considered "objects", but found that the actual number of methods
828	thatc an be called are very low. Since port IDs and noderefs travel over	1215	that can be called are quite low. Since port and node IDs travel over
829	the network frequently, the serialising/deserialising would add lots of	1216	the network frequently, the serialising/deserialising would add lots of
830	overhead, as well as having to keep a proxy object.	1217	overhead, as well as having to keep a proxy object everywhere.
831		1218
832	Strings can easily be printed, easily serialised etc. and need no special	1219	Strings can easily be printed, easily serialised etc. and need no special
833	procedures to be "valid".	1220	procedures to be "valid".
834		1221
835	And a a miniport consists of a single closure stored in a global hash - it	1222	And as a result, a port with just a default receiver consists of a single
836	can't become much cheaper.	1223	code reference stored in a global hash - it can't become much cheaper.
837		1224
838	=item Why favour JSON, why not real serialising format such as Storable?	1225	=item Why favour JSON, why not a real serialising format such as Storable?
839		1226
840	In fact, any AnyEvent::MP node will happily accept Storable as framing	1227	In fact, any AnyEvent::MP node will happily accept Storable as framing
841	format, but currently there is no way to make a node use Storable by	1228	format, but currently there is no way to make a node use Storable by
842	default.	1229	default (although all nodes will accept it).
843		1230
844	The default framing protocol is JSON because a) JSON::XS is many times	1231	The default framing protocol is JSON because a) JSON::XS is many times
845	faster for small messages and b) most importantly, after years of	1232	faster for small messages and b) most importantly, after years of
846	experience we found that object serialisation is causing more problems	1233	experience we found that object serialisation is causing more problems
847	than it gains: Just like function calls, objects simply do not travel	1234	than it solves: Just like function calls, objects simply do not travel
848	easily over the network, mostly because they will always be a copy, so you	1235	easily over the network, mostly because they will always be a copy, so you
849	always have to re-think your design.	1236	always have to re-think your design.
850		1237
851	Keeping your messages simple, concentrating on data structures rather than	1238	Keeping your messages simple, concentrating on data structures rather than
852	objects, will keep your messages clean, tidy and efficient.	1239	objects, will keep your messages clean, tidy and efficient.
853		1240
854	=back	1241	=back
855		1242
		1243	=head1 PORTING FROM AnyEvent::MP VERSION 1.X
		1244
		1245	AEMP version 2 has a few major incompatible changes compared to version 1:
		1246
		1247	=over 4
		1248
		1249	=item AnyEvent::MP::Global no longer has group management functions.
		1250
		1251	At least not officially - the grp_* functions are still exported and might
		1252	work, but they will be removed in some later release.
		1253
		1254	AnyEvent::MP now comes with a distributed database that is more
		1255	powerful. Its database families map closely to port groups, but the API
		1256	has changed (the functions are also now exported by AnyEvent::MP). Here is
		1257	a rough porting guide:
		1258
		1259	grp_reg $group, $port # old
		1260	db_reg $group, $port # new
		1261
		1262	$list = grp_get $group # old
		1263	db_keys $group, sub { my $list = shift } # new
		1264
		1265	grp_mon $group, $cb->(\@ports, $add, $del) # old
		1266	db_mon $group, $cb->(\%ports, $add, $change, $del) # new
		1267
		1268	C<grp_reg> is a no-brainer (just replace by C<db_reg>), but C<grp_get> is
		1269	no longer instant, because the local node might not have a copy of the
		1270	group. You can either modify your code to allow for a callback, or use
		1271	C<db_mon> to keep an updated copy of the group:
		1272
		1273	my $local_group_copy;
		1274	db_mon $group => sub { $local_group_copy = $_[0] };
		1275
		1276	# now "keys %$local_group_copy" always returns the most up-to-date
		1277	# list of ports in the group.
		1278
		1279	C<grp_mon> can be replaced by C<db_mon> with minor changes - C<db_mon>
		1280	passes a hash as first argument, and an extra C<$chg> argument that can be
		1281	ignored:
		1282
		1283	db_mon $group => sub {
		1284	my ($ports, $add, $chg, $lde) = @_;
		1285	$ports = [keys %$ports];
		1286
		1287	# now $ports, $add and $del are the same as
		1288	# were originally passed by grp_mon.
		1289	...
		1290	};
		1291
		1292	=item Nodes not longer connect to all other nodes.
		1293
		1294	In AEMP 1.x, every node automatically loads the L<AnyEvent::MP::Global>
		1295	module, which in turn would create connections to all other nodes in the
		1296	network (helped by the seed nodes).
		1297
		1298	In version 2.x, global nodes still connect to all other global nodes, but
		1299	other nodes don't - now every node either is a global node itself, or
		1300	attaches itself to another global node.
		1301
		1302	If a node isn't a global node itself, then it attaches itself to one
		1303	of its seed nodes. If that seed node isn't a global node yet, it will
		1304	automatically be upgraded to a global node.
		1305
		1306	So in many cases, nothing needs to be changed - one just has to make sure
		1307	that all seed nodes are meshed together with the other seed nodes (as with
		1308	AEMP 1.x), and other nodes specify them as seed nodes. This is most easily
		1309	achieved by specifying the same set of seed nodes for all nodes in the
		1310	network.
		1311
		1312	Not opening a connection to every other node is usually an advantage,
		1313	except when you need the lower latency of an already established
		1314	connection. To ensure a node establishes a connection to another node,
		1315	you can monitor the node port (C<mon $node, ...>), which will attempt to
		1316	create the connection (and notify you when the connection fails).
		1317
		1318	=item Listener-less nodes (nodes without binds) are gone.
		1319
		1320	And are not coming back, at least not in their old form. If no C<binds>
		1321	are specified for a node, AnyEvent::MP assumes a default of C<:>.
		1322
		1323	There are vague plans to implement some form of routing domains, which
		1324	might or might not bring back listener-less nodes, but don't count on it.
		1325
		1326	The fact that most connections are now optional somewhat mitigates this,
		1327	as a node can be effectively unreachable from the outside without any
		1328	problems, as long as it isn't a global node and only reaches out to other
		1329	nodes (as opposed to being contacted from other nodes).
		1330
		1331	=item $AnyEvent::MP::Kernel::WARN has gone.
		1332
		1333	AnyEvent has acquired a logging framework (L<AnyEvent::Log>), and AEMP now
		1334	uses this, and so should your programs.
		1335
		1336	Every module now documents what kinds of messages it generates, with
		1337	AnyEvent::MP acting as a catch all.
		1338
		1339	On the positive side, this means that instead of setting
		1340	C<PERL_ANYEVENT_MP_WARNLEVEL>, you can get away by setting C<AE_VERBOSE> -
		1341	much less to type.
		1342
		1343	=back
		1344
		1345	=head1 LOGGING
		1346
		1347	AnyEvent::MP does not normally log anything by itself, but sinc eit is the
		1348	root of the contetx hierarchy for AnyEvent::MP modules, it will receive
		1349	all log messages by submodules.
		1350
856	=head1 SEE ALSO	1351	=head1 SEE ALSO
		1352
		1353	L<AnyEvent::MP::Intro> - a gentle introduction.
		1354
		1355	L<AnyEvent::MP::Kernel> - more, lower-level, stuff.
		1356
		1357	L<AnyEvent::MP::Global> - network maintenance and port groups, to find
		1358	your applications.
		1359
		1360	L<AnyEvent::MP::DataConn> - establish data connections between nodes.
		1361
		1362	L<AnyEvent::MP::LogCatcher> - simple service to display log messages from
		1363	all nodes.
857		1364
858	L<AnyEvent>.	1365	L<AnyEvent>.
859		1366
860	=head1 AUTHOR	1367	=head1 AUTHOR
861		1368

Diff Legend

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

Comparing AnyEvent-MP/MP.pm (file contents): Revision 1.63 by root, Thu Aug 27 21:29:37 2009 UTC vs. Revision 1.142 by root, Fri Mar 23 13:44:01 2012 UTC

Diff Legend

Comparing AnyEvent-MP/MP.pm (file contents):
Revision 1.63 by root, Thu Aug 27 21:29:37 2009 UTC vs.
Revision 1.142 by root, Fri Mar 23 13:44:01 2012 UTC