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

Comparing AnyEvent-MP/MP.pm (file contents):
Revision 1.5 by root, Sat Aug 1 07:44:02 2009 UTC vs.
Revision 1.75 by root, Mon Aug 31 13:18:06 2009 UTC

…		…
4		4
5	=head1 SYNOPSIS	5	=head1 SYNOPSIS
6		6
7	use AnyEvent::MP;	7	use AnyEvent::MP;
8		8
9	NODE # returns this node identifier
10	$NODE # contains this node identifier	9	$NODE # contains this node's node ID
		10	NODE # returns this node's node ID
11		11
		12	$SELF # receiving/own port id in rcv callbacks
		13
		14	# initialise the node so it can send/receive messages
		15	configure;
		16
		17	# ports are message destinations
		18
		19	# sending messages
12	snd $port, type => data...;	20	snd $port, type => data...;
		21	snd $port, @msg;
		22	snd @msg_with_first_element_being_a_port;
13		23
14	rcv $port, smartmatch => $cb->($port, @msg);	24	# creating/using ports, the simple way
		25	my $simple_port = port { my @msg = @_ };
15		26
16	# examples:	27	# creating/using ports, tagged message matching
		28	my $port = port;
17	rcv $port2, ping => sub { snd $_[0], "pong"; 0 };	29	rcv $port, ping => sub { snd $_[0], "pong" };
18	rcv $port1, pong => sub { warn "pong received\n" };	30	rcv $port, pong => sub { warn "pong received\n" };
19	snd $port2, ping => $port1;
20		31
21	# more, smarter, matches (_any_ is exported by this module)	32	# create a port on another node
22	rcv $port, [child_died => $pid] => sub { ...	33	my $port = spawn $node, $initfunc, @initdata;
23	rcv $port, [_any_, _any_, 3] => sub { .. $_[2] is 3	34
		35	# monitoring
		36	mon $port, $cb->(@msg) # callback is invoked on death
		37	mon $port, $otherport # kill otherport on abnormal death
		38	mon $port, $otherport, @msg # send message on death
		39
		40	=head1 CURRENT STATUS
		41
		42	bin/aemp - stable.
		43	AnyEvent::MP - stable API, should work.
		44	AnyEvent::MP::Intro - uptodate, but incomplete.
		45	AnyEvent::MP::Kernel - mostly stable.
		46	AnyEvent::MP::Global - stable API, protocol not yet final.
		47
		48	stay tuned.
24		49
25	=head1 DESCRIPTION	50	=head1 DESCRIPTION
26		51
27	This module (-family) implements a simple message passing framework.	52	This module (-family) implements a simple message passing framework.
28		53
29	Despite its simplicity, you can securely message other processes running	54	Despite its simplicity, you can securely message other processes running
30	on the same or other hosts.	55	on the same or other hosts, and you can supervise entities remotely.
		56
		57	For an introduction to this module family, see the L<AnyEvent::MP::Intro>
		58	manual page and the examples under F<eg/>.
		59
		60	At the moment, this module family is a bit underdocumented.
31		61
32	=head1 CONCEPTS	62	=head1 CONCEPTS
33		63
34	=over 4	64	=over 4
35		65
36	=item port	66	=item port
37		67
38	A port is something you can send messages to with the C<snd> function, and	68	A port is something you can send messages to (with the C<snd> function).
39	you can register C<rcv> handlers with. All C<rcv> handlers will receive
40	messages they match, messages will not be queued.
41		69
		70	Ports allow you to register C<rcv> handlers that can match all or just
		71	some messages. Messages send to ports will not be queued, regardless of
		72	anything was listening for them or not.
		73
42	=item port id - C<noderef#portname>	74	=item port ID - C<nodeid#portname>
43		75
44	A port id is always the noderef, a hash-mark (C<#>) as separator, followed	76	A port ID is the concatenation of a node ID, a hash-mark (C<#>) as
45	by a port name (a printable string of unspecified format).	77	separator, and a port name (a printable string of unspecified format).
46		78
47	=item node	79	=item node
48		80
49	A node is a single process containing at least one port - the node	81	A node is a single process containing at least one port - the node port,
50	port. You can send messages to node ports to let them create new ports,	82	which enables nodes to manage each other remotely, and to create new
51	among other things.	83	ports.
52		84
53	Initially, nodes are either private (single-process only) or hidden	85	Nodes are either public (have one or more listening ports) or private
54	(connected to a master node only). Only when they epxlicitly "become	86	(no listening ports). Private nodes cannot talk to other private nodes
55	public" can you send them messages from unrelated other nodes.	87	currently.
56		88
57	=item noderef - C<host:port,host:port...>, C<id@noderef>, C<id>	89	=item node ID - C<[a-za-Z0-9_\-.:]+>
58		90
59	A noderef is a string that either uniquely identifies a given node (for	91	A node ID is a string that uniquely identifies the node within a
60	private and hidden nodes), or contains a recipe on how to reach a given	92	network. Depending on the configuration used, node IDs can look like a
61	node (for public nodes).	93	hostname, a hostname and a port, or a random string. AnyEvent::MP itself
		94	doesn't interpret node IDs in any way.
		95
		96	=item binds - C<ip:port>
		97
		98	Nodes can only talk to each other by creating some kind of connection to
		99	each other. To do this, nodes should listen on one or more local transport
		100	endpoints - binds. Currently, only standard C<ip:port> specifications can
		101	be used, which specify TCP ports to listen on.
		102
		103	=item seeds - C<host:port>
		104
		105	When a node starts, it knows nothing about the network. To teach the node
		106	about the network it first has to contact some other node within the
		107	network. This node is called a seed.
		108
		109	Seeds are transport endpoint(s) of as many nodes as one wants. Those nodes
		110	are expected to be long-running, and at least one of those should always
		111	be available. When nodes run out of connections (e.g. due to a network
		112	error), they try to re-establish connections to some seednodes again to
		113	join the network.
		114
		115	Apart from being sued for seeding, seednodes are not special in any way -
		116	every public node can be a seednode.
62		117
63	=back	118	=back
64		119
65	=head1 VARIABLES/FUNCTIONS	120	=head1 VARIABLES/FUNCTIONS
66		121
…		…
68		123
69	=cut	124	=cut
70		125
71	package AnyEvent::MP;	126	package AnyEvent::MP;
72		127
73	use AnyEvent::MP::Util ();
74	use AnyEvent::MP::Node;	128	use AnyEvent::MP::Kernel;
75	use AnyEvent::MP::Transport;
76		129
77	use utf8;
78	use common::sense;	130	use common::sense;
79		131
80	use Carp ();	132	use Carp ();
81		133
82	use AE ();	134	use AE ();
83		135
84	use base "Exporter";	136	use base "Exporter";
85		137
86	our $VERSION = '0.0';	138	our $VERSION = $AnyEvent::MP::Kernel::VERSION;
87	our @EXPORT = qw(NODE $NODE $PORT snd rcv _any_);
88		139
89	our $DEFAULT_SECRET;	140	our @EXPORT = qw(
90	our $DEFAULT_PORT = "4040";	141	NODE $NODE *SELF node_of after
		142	configure
		143	snd rcv mon mon_guard kil reg psub spawn
		144	port
		145	);
91		146
92	our $CONNECT_INTERVAL = 5; # new connect every 5s, at least	147	our $SELF;
93	our $CONNECT_TIMEOUT = 30; # includes handshake
94		148
95	sub default_secret {	149	sub _self_die() {
96	unless (defined $DEFAULT_SECRET) {	150	my $msg = $@;
97	if (open my $fh, "<$ENV{HOME}/.aemp-secret") {	151	$msg =~ s/\n+$// unless ref $msg;
98	sysread $fh, $DEFAULT_SECRET, -s $fh;	152	kil $SELF, die => $msg;
99	} else {
100	$DEFAULT_SECRET = AnyEvent::MP::Util::nonce 32;
101	}
102	}
103
104	$DEFAULT_SECRET
105	}	153	}
106		154
107	=item NODE / $NODE	155	=item $thisnode = NODE / $NODE
108		156
109	The C<NODE ()> function and the C<$NODE> variable contain the noderef of	157	The C<NODE> function returns, and the C<$NODE> variable contains, the node
110	the local node. The value is initialised by a call to C<become_public> or	158	ID of the node running in the current process. This value is initialised by
111	C<become_slave>, after which all local port identifiers become invalid.	159	a call to C<configure>.
112		160
113	=cut	161	=item $nodeid = node_of $port
114		162
115	our $UNIQ = sprintf "%x.%x", $$, time; # per-process/node unique cookie	163	Extracts and returns the node ID from a port ID or a node ID.
116	our $PUBLIC = 0;
117	our $NODE;
118	our $PORT;
119		164
120	our %NODE; # node id to transport mapping, or "undef", for local node	165	=item configure key => value...
121	our %PORT; # local ports
122	our %LISTENER; # local transports
123		166
124	sub NODE() { $NODE }	167	Before a node can talk to other nodes on the network (i.e. enter
		168	"distributed mode") it has to configure itself - the minimum a node needs
		169	to know is its own name, and optionally it should know the addresses of
		170	some other nodes in the network to discover other nodes.
125		171
126	{	172	This function configures a node - it must be called exactly once (or
127	use POSIX ();	173	never) before calling other AnyEvent::MP functions.
128	my $nodename = (POSIX::uname)[1];
129	$NODE = "$$\@$nodename";
130	}
131		174
132	sub _ANY_() { 1 }	175	=over 4
133	sub _any_() { \&_ANY_ }
134		176
135	sub add_node {	177	=item step 1, gathering configuration from profiles
136	my ($noderef) = @_;
137		178
138	return $NODE{$noderef}	179	The function first looks up a profile in the aemp configuration (see the
139	if exists $NODE{$noderef};	180	L<aemp> commandline utility). The profile name can be specified via the
		181	named C<profile> parameter. If it is missing, then the nodename (F<uname
		182	-n>) will be used as profile name.
140		183
141	for (split /,/, $noderef) {	184	The profile data is then gathered as follows:
142	return $NODE{$noderef} = $NODE{$_}
143	if exists $NODE{$_};
144	}
145		185
146	# for indirect sends, use a different class	186	First, all remaining key => value pairs (all of which are conviniently
147	my $node = new AnyEvent::MP::Node::Direct $noderef;	187	undocumented at the moment) will be interpreted as configuration
		188	data. Then they will be overwritten by any values specified in the global
		189	default configuration (see the F<aemp> utility), then the chain of
		190	profiles chosen by the profile name (and any C<parent> attributes).
148		191
149	$NODE{$_} = $node	192	That means that the values specified in the profile have highest priority
150	for $noderef, split /,/, $noderef;	193	and the values specified directly via C<configure> have lowest priority,
		194	and can only be used to specify defaults.
151		195
152	$node	196	If the profile specifies a node ID, then this will become the node ID of
153	}	197	this process. If not, then the profile name will be used as node ID. The
		198	special node ID of C<anon/> will be replaced by a random node ID.
154		199
		200	=item step 2, bind listener sockets
		201
		202	The next step is to look up the binds in the profile, followed by binding
		203	aemp protocol listeners on all binds specified (it is possible and valid
		204	to have no binds, meaning that the node cannot be contacted form the
		205	outside. This means the node cannot talk to other nodes that also have no
		206	binds, but it can still talk to all "normal" nodes).
		207
		208	If the profile does not specify a binds list, then a default of C<*> is
		209	used, meaning the node will bind on a dynamically-assigned port on every
		210	local IP address it finds.
		211
		212	=item step 3, connect to seed nodes
		213
		214	As the last step, the seeds list from the profile is passed to the
		215	L<AnyEvent::MP::Global> module, which will then use it to keep
		216	connectivity with at least one node at any point in time.
		217
		218	=back
		219
		220	Example: become a distributed node using the locla node name as profile.
		221	This should be the most common form of invocation for "daemon"-type nodes.
		222
		223	configure
		224
		225	Example: become an anonymous node. This form is often used for commandline
		226	clients.
		227
		228	configure nodeid => "anon/";
		229
		230	Example: configure a node using a profile called seed, which si suitable
		231	for a seed node as it binds on all local addresses on a fixed port (4040,
		232	customary for aemp).
		233
		234	# use the aemp commandline utility
		235	# aemp profile seed nodeid anon/ binds '*:4040'
		236
		237	# then use it
		238	configure profile => "seed";
		239
		240	# or simply use aemp from the shell again:
		241	# aemp run profile seed
		242
		243	# or provide a nicer-to-remember nodeid
		244	# aemp run profile seed nodeid "$(hostname)"
		245
		246	=item $SELF
		247
		248	Contains the current port id while executing C<rcv> callbacks or C<psub>
		249	blocks.
		250
		251	=item *SELF, SELF, %SELF, @SELF...
		252
		253	Due to some quirks in how perl exports variables, it is impossible to
		254	just export C<$SELF>, all the symbols named C<SELF> are exported by this
		255	module, but only C<$SELF> is currently used.
		256
155	=item snd $portid, type => @data	257	=item snd $port, type => @data
156		258
157	=item snd $portid, @msg	259	=item snd $port, @msg
158		260
159	Send the given message to the given port ID, which can identify either a	261	Send the given message to the given port, which can identify either a
160	local or a remote port.	262	local or a remote port, and must be a port ID.
161		263
162	While the message can be about anything, it is highly recommended to use	264	While the message can be almost anything, it is highly recommended to
163	a constant string as first element.	265	use a string as first element (a port ID, or some word that indicates a
		266	request type etc.) and to consist if only simple perl values (scalars,
		267	arrays, hashes) - if you think you need to pass an object, think again.
164		268
165	The message data effectively becomes read-only after a call to this	269	The message data logically becomes read-only after a call to this
166	function: modifying any argument is not allowed and can cause many	270	function: modifying any argument (or values referenced by them) is
167	problems.	271	forbidden, as there can be considerable time between the call to C<snd>
		272	and the time the message is actually being serialised - in fact, it might
		273	never be copied as within the same process it is simply handed to the
		274	receiving port.
168		275
169	The type of data you can transfer depends on the transport protocol: when	276	The type of data you can transfer depends on the transport protocol: when
170	JSON is used, then only strings, numbers and arrays and hashes consisting	277	JSON is used, then only strings, numbers and arrays and hashes consisting
171	of those are allowed (no objects). When Storable is used, then anything	278	of those are allowed (no objects). When Storable is used, then anything
172	that Storable can serialise and deserialise is allowed, and for the local	279	that Storable can serialise and deserialise is allowed, and for the local
173	node, anything can be passed.	280	node, anything can be passed. Best rely only on the common denominator of
		281	these.
174		282
175	=cut	283	=item $local_port = port
176		284
177	sub snd(@) {	285	Create a new local port object and returns its port ID. Initially it has
178	my ($noderef, $port) = split /#/, shift, 2;	286	no callbacks set and will throw an error when it receives messages.
179		287
180	add_node $noderef	288	=item $local_port = port { my @msg = @_ }
181	unless exists $NODE{$noderef};
182		289
183	$NODE{$noderef}->send (["$port", [@_]]);	290	Creates a new local port, and returns its ID. Semantically the same as
184	}	291	creating a port and calling C<rcv $port, $callback> on it.
185		292
		293	The block will be called for every message received on the port, with the
		294	global variable C<$SELF> set to the port ID. Runtime errors will cause the
		295	port to be C<kil>ed. The message will be passed as-is, no extra argument
		296	(i.e. no port ID) will be passed to the callback.
		297
		298	If you want to stop/destroy the port, simply C<kil> it:
		299
		300	my $port = port {
		301	my @msg = @_;
		302	...
		303	kil $SELF;
		304	};
		305
		306	=cut
		307
		308	sub rcv($@);
		309
		310	sub _kilme {
		311	die "received message on port without callback";
		312	}
		313
		314	sub port(;&) {
		315	my $id = "$UNIQ." . $ID++;
		316	my $port = "$NODE#$id";
		317
		318	rcv $port, shift \|\| \&_kilme;
		319
		320	$port
		321	}
		322
186	=item rcv $portid, type => $callback->(@msg)	323	=item rcv $local_port, $callback->(@msg)
187		324
188	=item rcv $portid, $smartmatch => $callback->(@msg)	325	Replaces the default callback on the specified port. There is no way to
		326	remove the default callback: use C<sub { }> to disable it, or better
		327	C<kil> the port when it is no longer needed.
189		328
190	=item rcv $portid, [$smartmatch...] => $callback->(@msg)	329	The global C<$SELF> (exported by this module) contains C<$port> while
		330	executing the callback. Runtime errors during callback execution will
		331	result in the port being C<kil>ed.
191		332
192	Register a callback on the port identified by C<$portid>, which I<must> be	333	The default callback received all messages not matched by a more specific
193	a local port.	334	C<tag> match.
194		335
195	The callback has to return a true value when its work is done, after	336	=item rcv $local_port, tag => $callback->(@msg_without_tag), ...
196	which is will be removed, or a false value in which case it will stay
197	registered.
198		337
199	If the match is an array reference, then it will be matched against the	338	Register (or replace) callbacks to be called on messages starting with the
200	first elements of the message, otherwise only the first element is being	339	given tag on the given port (and return the port), or unregister it (when
201	matched.	340	C<$callback> is C<$undef> or missing). There can only be one callback
		341	registered for each tag.
202		342
203	Any element in the match that is specified as C<_any_> (a function	343	The original message will be passed to the callback, after the first
204	exported by this module) matches any single element of the message.	344	element (the tag) has been removed. The callback will use the same
		345	environment as the default callback (see above).
205		346
206	While not required, it is highly recommended that the first matching	347	Example: create a port and bind receivers on it in one go.
207	element is a string identifying the message. The one-string-only match is	348
208	also the most efficient match (by far).	349	my $port = rcv port,
		350	msg1 => sub { ... },
		351	msg2 => sub { ... },
		352	;
		353
		354	Example: create a port, bind receivers and send it in a message elsewhere
		355	in one go:
		356
		357	snd $otherport, reply =>
		358	rcv port,
		359	msg1 => sub { ... },
		360	...
		361	;
		362
		363	Example: temporarily register a rcv callback for a tag matching some port
		364	(e.g. for a rpc reply) and unregister it after a message was received.
		365
		366	rcv $port, $otherport => sub {
		367	my @reply = @_;
		368
		369	rcv $SELF, $otherport;
		370	};
209		371
210	=cut	372	=cut
211		373
212	sub rcv($@) {	374	sub rcv($@) {
213	my ($port, $match, $cb) = @_;	375	my $port = shift;
214
215	my $port = $PORT{$port}
216	or do {
217	my ($noderef, $lport) = split /#/, $port;	376	my ($nodeid, $portid) = split /#/, $port, 2;
218	"AnyEvent::MP::Node::Self" eq ref $NODE{$noderef}	377
		378	$NODE{$nodeid} == $NODE{""}
219	or Carp::croak "$port: can only rcv on local ports";	379	or Carp::croak "$port: rcv can only be called on local ports, caught";
220		380
221	$PORT{$lport}	381	while (@_) {
222	or Carp::croak "$port: port does not exist";	382	if (ref $_[0]) {
223		383	if (my $self = $PORT_DATA{$portid}) {
224	$PORT{$port} = $PORT{$lport} # also return	384	"AnyEvent::MP::Port" eq ref $self
225	};	385	or Carp::croak "$port: rcv can only be called on message matching ports, caught";
226		386
227	if (!ref $match) {	387	$self->[2] = shift;
228	push @{ $port->{rc0}{$match} }, [$cb];
229	} elsif (("ARRAY" eq ref $match && !ref $match->[0])) {
230	my ($type, @match) = @$match;
231	@match
232	? push @{ $port->{rcv}{$match->[0]} }, [$cb, \@match]
233	: push @{ $port->{rc0}{$match->[0]} }, [$cb];
234	} else {	388	} else {
235	push @{ $port->{any} }, [$cb, $match];	389	my $cb = shift;
236	}	390	$PORT{$portid} = sub {
237	}	391	local $SELF = $port;
238		392	eval { &$cb }; _self_die if $@;
239	sub _inject {
240	my ($port, $msg) = @{+shift};
241
242	$port = $PORT{$port}
243	or return;
244
245	@_ = @$msg;
246
247	for (@{ $port->{rc0}{$msg->[0]} }) {
248	$_ && &{$_->[0]}
249	&& undef $_;
250	}
251
252	for (@{ $port->{rcv}{$msg->[0]} }) {
253	$_ && [@_[1..$#{$_->[1]}]] ~~ $_->[1]
254	&& &{$_->[0]}
255	&& undef $_;
256	}
257
258	for (@{ $port->{any} }) {
259	$_ && [@_[0..$#{$_->[1]}]] ~~ $_->[1]
260	&& &{$_->[0]}
261	&& undef $_;
262	}
263	}
264
265	sub normalise_noderef($) {
266	my ($noderef) = @_;
267
268	my $cv = AE::cv;
269	my @res;
270
271	$cv->begin (sub {
272	my %seen;
273	my @refs;
274	for (sort { $a->[0] <=> $b->[0] } @res) {
275	push @refs, $_->[1] unless $seen{$_->[1]}++
276	}
277	shift->send (join ",", @refs);
278	});
279
280	$noderef = $DEFAULT_PORT unless length $noderef;
281
282	my $idx;
283	for my $t (split /,/, $noderef) {
284	my $pri = ++$idx;
285
286	#TODO: this should be outside normalise_noderef and in become_public
287	if ($t =~ /^\d*$/) {
288	my $nodename = (POSIX::uname)[1];
289
290	$cv->begin;
291	AnyEvent::Socket::resolve_sockaddr $nodename, $t \|\| "aemp=$DEFAULT_PORT", "tcp", 0, undef, sub {
292	for (@_) {
293	my ($service, $host) = AnyEvent::Socket::unpack_sockaddr $_->[3];
294	push @res, [
295	$pri += 1e-5,
296	AnyEvent::Socket::format_hostport AnyEvent::Socket::format_address $host, $service
297	];
298	}	393	};
299	$cv->end;	394	}
		395	} elsif (defined $_[0]) {
		396	my $self = $PORT_DATA{$portid} \|\|= do {
		397	my $self = bless [$PORT{$port} \|\| sub { }, { }, $port], "AnyEvent::MP::Port";
		398
		399	$PORT{$portid} = sub {
		400	local $SELF = $port;
		401
		402	if (my $cb = $self->[1]{$_[0]}) {
		403	shift;
		404	eval { &$cb }; _self_die if $@;
		405	} else {
		406	&{ $self->[0] };
		407	}
		408	};
		409
		410	$self
300	};	411	};
301		412
302	# my (undef, undef, undef, undef, @ipv4) = gethostbyname $nodename;	413	"AnyEvent::MP::Port" eq ref $self
303	#	414	or Carp::croak "$port: rcv can only be called on message matching ports, caught";
304	# for (@ipv4) {	415
305	# push @res, [	416	my ($tag, $cb) = splice @_, 0, 2;
306	# $pri,	417
307	# AnyEvent::Socket::format_hostport AnyEvent::Socket::format_address $_, $t \|\| $DEFAULT_PORT,	418	if (defined $cb) {
308	# ];	419	$self->[1]{$tag} = $cb;
309	# }
310	} else {	420	} else {
311	my ($host, $port) = AnyEvent::Socket::parse_hostport $t, "aemp=$DEFAULT_PORT"	421	delete $self->[1]{$tag};
312	or Carp::croak "$t: unparsable transport descriptor";
313
314	$cv->begin;
315	AnyEvent::Socket::resolve_sockaddr $host, $port, "tcp", 0, undef, sub {
316	for (@_) {
317	my ($service, $host) = AnyEvent::Socket::unpack_sockaddr $_->[3];
318	push @res, [
319	$pri += 1e-5,
320	AnyEvent::Socket::format_hostport AnyEvent::Socket::format_address $host, $service
321	];
322	}
323	$cv->end;
324	}	422	}
325	}	423	}
326	}	424	}
327		425
328	$cv->end;	426	$port
329
330	$cv
331	}	427	}
332		428
333	sub become_public {	429	=item $closure = psub { BLOCK }
334	return if $PUBLIC;
335		430
336	my $noderef = join ",", ref $_[0] ? @{+shift} : shift;	431	Remembers C<$SELF> and creates a closure out of the BLOCK. When the
337	my @args = @_;	432	closure is executed, sets up the environment in the same way as in C<rcv>
		433	callbacks, i.e. runtime errors will cause the port to get C<kil>ed.
338		434
339	$NODE = (normalise_noderef $noderef)->recv;	435	This is useful when you register callbacks from C<rcv> callbacks:
340		436
341	for my $t (split /,/, $NODE) {	437	rcv delayed_reply => sub {
342	$NODE{$t} = $NODE{""};	438	my ($delay, @reply) = @_;
343		439	my $timer = AE::timer $delay, 0, psub {
344	my ($host, $port) = AnyEvent::Socket::parse_hostport $t;	440	snd @reply, $SELF;
345
346	$LISTENER{$t} = AnyEvent::MP::Transport::mp_server $host, $port,
347	@args,
348	on_error => sub {
349	die "on_error<@_>\n";#d#
350	},
351	on_connect => sub {
352	my ($tp) = @_;
353
354	$NODE{$tp->{remote_id}} = $_[0];
355	},
356	sub {
357	my ($tp) = @_;
358
359	$NODE{"$tp->{peerhost}:$tp->{peerport}"} = $tp;
360	},
361	;	441	};
		442	};
		443
		444	=cut
		445
		446	sub psub(&) {
		447	my $cb = shift;
		448
		449	my $port = $SELF
		450	or Carp::croak "psub can only be called from within rcv or psub callbacks, not";
		451
		452	sub {
		453	local $SELF = $port;
		454
		455	if (wantarray) {
		456	my @res = eval { &$cb };
		457	_self_die if $@;
		458	@res
		459	} else {
		460	my $res = eval { &$cb };
		461	_self_die if $@;
		462	$res
		463	}
362	}	464	}
		465	}
363		466
364	$PUBLIC = 1;	467	=item $guard = mon $port, $cb->(@reason) # call $cb when $port dies
		468
		469	=item $guard = mon $port, $rcvport # kill $rcvport when $port dies
		470
		471	=item $guard = mon $port # kill $SELF when $port dies
		472
		473	=item $guard = mon $port, $rcvport, @msg # send a message when $port dies
		474
		475	Monitor the given port and do something when the port is killed or
		476	messages to it were lost, and optionally return a guard that can be used
		477	to stop monitoring again.
		478
		479	C<mon> effectively guarantees that, in the absence of hardware failures,
		480	after starting the monitor, either all messages sent to the port will
		481	arrive, or the monitoring action will be invoked after possible message
		482	loss has been detected. No messages will be lost "in between" (after
		483	the first lost message no further messages will be received by the
		484	port). After the monitoring action was invoked, further messages might get
		485	delivered again.
		486
		487	Note that monitoring-actions are one-shot: once messages are lost (and a
		488	monitoring alert was raised), they are removed and will not trigger again.
		489
		490	In the first form (callback), the callback is simply called with any
		491	number of C<@reason> elements (no @reason means that the port was deleted
		492	"normally"). Note also that I<< the callback B<must> never die >>, so use
		493	C<eval> if unsure.
		494
		495	In the second form (another port given), the other port (C<$rcvport>)
		496	will be C<kil>'ed with C<@reason>, iff a @reason was specified, i.e. on
		497	"normal" kils nothing happens, while under all other conditions, the other
		498	port is killed with the same reason.
		499
		500	The third form (kill self) is the same as the second form, except that
		501	C<$rvport> defaults to C<$SELF>.
		502
		503	In the last form (message), a message of the form C<@msg, @reason> will be
		504	C<snd>.
		505
		506	As a rule of thumb, monitoring requests should always monitor a port from
		507	a local port (or callback). The reason is that kill messages might get
		508	lost, just like any other message. Another less obvious reason is that
		509	even monitoring requests can get lost (for exmaple, when the connection
		510	to the other node goes down permanently). When monitoring a port locally
		511	these problems do not exist.
		512
		513	Example: call a given callback when C<$port> is killed.
		514
		515	mon $port, sub { warn "port died because of <@_>\n" };
		516
		517	Example: kill ourselves when C<$port> is killed abnormally.
		518
		519	mon $port;
		520
		521	Example: send us a restart message when another C<$port> is killed.
		522
		523	mon $port, $self => "restart";
		524
		525	=cut
		526
		527	sub mon {
		528	my ($nodeid, $port) = split /#/, shift, 2;
		529
		530	my $node = $NODE{$nodeid} \|\| add_node $nodeid;
		531
		532	my $cb = @_ ? shift : $SELF \|\| Carp::croak 'mon: called with one argument only, but $SELF not set,';
		533
		534	unless (ref $cb) {
		535	if (@_) {
		536	# send a kill info message
		537	my (@msg) = ($cb, @_);
		538	$cb = sub { snd @msg, @_ };
		539	} else {
		540	# simply kill other port
		541	my $port = $cb;
		542	$cb = sub { kil $port, @_ if @_ };
		543	}
		544	}
		545
		546	$node->monitor ($port, $cb);
		547
		548	defined wantarray
		549	and AnyEvent::Util::guard { $node->unmonitor ($port, $cb) }
		550	}
		551
		552	=item $guard = mon_guard $port, $ref, $ref...
		553
		554	Monitors the given C<$port> and keeps the passed references. When the port
		555	is killed, the references will be freed.
		556
		557	Optionally returns a guard that will stop the monitoring.
		558
		559	This function is useful when you create e.g. timers or other watchers and
		560	want to free them when the port gets killed (note the use of C<psub>):
		561
		562	$port->rcv (start => sub {
		563	my $timer; $timer = mon_guard $port, AE::timer 1, 1, psub {
		564	undef $timer if 0.9 < rand;
		565	});
		566	});
		567
		568	=cut
		569
		570	sub mon_guard {
		571	my ($port, @refs) = @_;
		572
		573	#TODO: mon-less form?
		574
		575	mon $port, sub { 0 && @refs }
		576	}
		577
		578	=item kil $port[, @reason]
		579
		580	Kill the specified port with the given C<@reason>.
		581
		582	If no C<@reason> is specified, then the port is killed "normally" (ports
		583	monitoring other ports will not necessarily die because a port dies
		584	"normally").
		585
		586	Otherwise, linked ports get killed with the same reason (second form of
		587	C<mon>, see above).
		588
		589	Runtime errors while evaluating C<rcv> callbacks or inside C<psub> blocks
		590	will be reported as reason C<< die => $@ >>.
		591
		592	Transport/communication errors are reported as C<< transport_error =>
		593	$message >>.
		594
		595	=cut
		596
		597	=item $port = spawn $node, $initfunc[, @initdata]
		598
		599	Creates a port on the node C<$node> (which can also be a port ID, in which
		600	case it's the node where that port resides).
		601
		602	The port ID of the newly created port is returned immediately, and it is
		603	possible to immediately start sending messages or to monitor the port.
		604
		605	After the port has been created, the init function is called on the remote
		606	node, in the same context as a C<rcv> callback. This function must be a
		607	fully-qualified function name (e.g. C<MyApp::Chat::Server::init>). To
		608	specify a function in the main program, use C<::name>.
		609
		610	If the function doesn't exist, then the node tries to C<require>
		611	the package, then the package above the package and so on (e.g.
		612	C<MyApp::Chat::Server>, C<MyApp::Chat>, C<MyApp>) until the function
		613	exists or it runs out of package names.
		614
		615	The init function is then called with the newly-created port as context
		616	object (C<$SELF>) and the C<@initdata> values as arguments.
		617
		618	A common idiom is to pass a local port, immediately monitor the spawned
		619	port, and in the remote init function, immediately monitor the passed
		620	local port. This two-way monitoring ensures that both ports get cleaned up
		621	when there is a problem.
		622
		623	Example: spawn a chat server port on C<$othernode>.
		624
		625	# this node, executed from within a port context:
		626	my $server = spawn $othernode, "MyApp::Chat::Server::connect", $SELF;
		627	mon $server;
		628
		629	# init function on C<$othernode>
		630	sub connect {
		631	my ($srcport) = @_;
		632
		633	mon $srcport;
		634
		635	rcv $SELF, sub {
		636	...
		637	};
		638	}
		639
		640	=cut
		641
		642	sub _spawn {
		643	my $port = shift;
		644	my $init = shift;
		645
		646	local $SELF = "$NODE#$port";
		647	eval {
		648	&{ load_func $init }
		649	};
		650	_self_die if $@;
		651	}
		652
		653	sub spawn(@) {
		654	my ($nodeid, undef) = split /#/, shift, 2;
		655
		656	my $id = "$RUNIQ." . $ID++;
		657
		658	$_[0] =~ /::/
		659	or Carp::croak "spawn init function must be a fully-qualified name, caught";
		660
		661	snd_to_func $nodeid, "AnyEvent::MP::_spawn" => $id, @_;
		662
		663	"$nodeid#$id"
		664	}
		665
		666	=item after $timeout, @msg
		667
		668	=item after $timeout, $callback
		669
		670	Either sends the given message, or call the given callback, after the
		671	specified number of seconds.
		672
		673	This is simply a utility function that comes in handy at times - the
		674	AnyEvent::MP author is not convinced of the wisdom of having it, though,
		675	so it may go away in the future.
		676
		677	=cut
		678
		679	sub after($@) {
		680	my ($timeout, @action) = @_;
		681
		682	my $t; $t = AE::timer $timeout, 0, sub {
		683	undef $t;
		684	ref $action[0]
		685	? $action[0]()
		686	: snd @action;
		687	};
365	}	688	}
366		689
367	=back	690	=back
368		691
369	=head1 NODE MESSAGES	692	=head1 AnyEvent::MP vs. Distributed Erlang
370		693
371	Nodes understand the following messages sent to them. Many of them take	694	AnyEvent::MP got lots of its ideas from distributed Erlang (Erlang node
372	arguments called C<@reply>, which will simply be used to compose a reply	695	== aemp node, Erlang process == aemp port), so many of the documents and
373	message - C<$reply[0]> is the port to reply to, C<$reply[1]> the type and	696	programming techniques employed by Erlang apply to AnyEvent::MP. Here is a
374	the remaining arguments are simply the message data.	697	sample:
		698
		699	http://www.Erlang.se/doc/programming_rules.shtml
		700	http://Erlang.org/doc/getting_started/part_frame.html # chapters 3 and 4
		701	http://Erlang.org/download/Erlang-book-part1.pdf # chapters 5 and 6
		702	http://Erlang.org/download/armstrong_thesis_2003.pdf # chapters 4 and 5
		703
		704	Despite the similarities, there are also some important differences:
375		705
376	=over 4	706	=over 4
377		707
378	=cut	708	=item * Node IDs are arbitrary strings in AEMP.
379		709
380	#############################################################################	710	Erlang relies on special naming and DNS to work everywhere in the same
381	# self node code	711	way. AEMP relies on each node somehow knowing its own address(es) (e.g. by
		712	configuraiton or DNS), but will otherwise discover other odes itself.
382		713
383	sub _new_port($) {	714	=item * Erlang has a "remote ports are like local ports" philosophy, AEMP
384	my ($name) = @_;	715	uses "local ports are like remote ports".
385		716
386	my ($noderef, $portname) = split /#/, $name;	717	The failure modes for local ports are quite different (runtime errors
		718	only) then for remote ports - when a local port dies, you I<know> it dies,
		719	when a connection to another node dies, you know nothing about the other
		720	port.
387		721
388	$PORT{$name} =	722	Erlang pretends remote ports are as reliable as local ports, even when
389	$PORT{$portname} = {	723	they are not.
390	names => [$name, $portname],
391	};
392	}
393		724
394	$NODE{""} = new AnyEvent::MP::Node::Self noderef => $NODE;	725	AEMP encourages a "treat remote ports differently" philosophy, with local
395	_new_port "";	726	ports being the special case/exception, where transport errors cannot
		727	occur.
396		728
397	=item relay => $port, @msg	729	=item * Erlang uses processes and a mailbox, AEMP does not queue.
398		730
399	Simply forwards the message to the given port.	731	Erlang uses processes that selectively receive messages, and therefore
		732	needs a queue. AEMP is event based, queuing messages would serve no
		733	useful purpose. For the same reason the pattern-matching abilities of
		734	AnyEvent::MP are more limited, as there is little need to be able to
		735	filter messages without dequeing them.
400		736
401	=cut	737	(But see L<Coro::MP> for a more Erlang-like process model on top of AEMP).
402		738
403	rcv "", relay => \&snd;	739	=item * Erlang sends are synchronous, AEMP sends are asynchronous.
404		740
405	=item eval => $string[ @reply]	741	Sending messages in Erlang is synchronous and blocks the process (and
		742	so does not need a queue that can overflow). AEMP sends are immediate,
		743	connection establishment is handled in the background.
406		744
407	Evaluates the given string. If C<@reply> is given, then a message of the	745	=item * Erlang suffers from silent message loss, AEMP does not.
408	form C<@reply, $@, @evalres> is sent.
409		746
410	Example: crash another node.	747	Erlang makes few guarantees on messages delivery - messages can get lost
		748	without any of the processes realising it (i.e. you send messages a, b,
		749	and c, and the other side only receives messages a and c).
411		750
412	snd $othernode, eval => "exit";	751	AEMP guarantees correct ordering, and the guarantee that after one message
		752	is lost, all following ones sent to the same port are lost as well, until
		753	monitoring raises an error, so there are no silent "holes" in the message
		754	sequence.
413		755
414	=cut	756	=item * Erlang can send messages to the wrong port, AEMP does not.
415		757
416	rcv "", eval => sub {	758	In Erlang it is quite likely that a node that restarts reuses a process ID
417	my (undef, $string, @reply) = @_;	759	known to other nodes for a completely different process, causing messages
418	my @res = eval $string;	760	destined for that process to end up in an unrelated process.
419	snd @reply, "$@", @res if @reply;
420	};
421		761
422	=item time => @reply	762	AEMP never reuses port IDs, so old messages or old port IDs floating
		763	around in the network will not be sent to an unrelated port.
423		764
424	Replies the the current node time to C<@reply>.	765	=item * Erlang uses unprotected connections, AEMP uses secure
		766	authentication and can use TLS.
425		767
426	Example: tell the current node to send the current time to C<$myport> in a	768	AEMP can use a proven protocol - TLS - to protect connections and
427	C<timereply> message.	769	securely authenticate nodes.
428		770
429	snd $NODE, time => $myport, timereply => 1, 2;	771	=item * The AEMP protocol is optimised for both text-based and binary
430	# => snd $myport, timereply => 1, 2, <time>	772	communications.
431		773
432	=cut	774	The AEMP protocol, unlike the Erlang protocol, supports both programming
		775	language independent text-only protocols (good for debugging) and binary,
		776	language-specific serialisers (e.g. Storable). By default, unless TLS is
		777	used, the protocol is actually completely text-based.
433		778
434	rcv "", time => sub { shift; snd @_, AE::time };	779	It has also been carefully designed to be implementable in other languages
		780	with a minimum of work while gracefully degrading functionality to make the
		781	protocol simple.
		782
		783	=item * AEMP has more flexible monitoring options than Erlang.
		784
		785	In Erlang, you can chose to receive I<all> exit signals as messages
		786	or I<none>, there is no in-between, so monitoring single processes is
		787	difficult to implement. Monitoring in AEMP is more flexible than in
		788	Erlang, as one can choose between automatic kill, exit message or callback
		789	on a per-process basis.
		790
		791	=item * Erlang tries to hide remote/local connections, AEMP does not.
		792
		793	Monitoring in Erlang is not an indicator of process death/crashes, in the
		794	same way as linking is (except linking is unreliable in Erlang).
		795
		796	In AEMP, you don't "look up" registered port names or send to named ports
		797	that might or might not be persistent. Instead, you normally spawn a port
		798	on the remote node. The init function monitors you, and you monitor the
		799	remote port. Since both monitors are local to the node, they are much more
		800	reliable (no need for C<spawn_link>).
		801
		802	This also saves round-trips and avoids sending messages to the wrong port
		803	(hard to do in Erlang).
435		804
436	=back	805	=back
437		806
		807	=head1 RATIONALE
		808
		809	=over 4
		810
		811	=item Why strings for port and node IDs, why not objects?
		812
		813	We considered "objects", but found that the actual number of methods
		814	that can be called are quite low. Since port and node IDs travel over
		815	the network frequently, the serialising/deserialising would add lots of
		816	overhead, as well as having to keep a proxy object everywhere.
		817
		818	Strings can easily be printed, easily serialised etc. and need no special
		819	procedures to be "valid".
		820
		821	And as a result, a miniport consists of a single closure stored in a
		822	global hash - it can't become much cheaper.
		823
		824	=item Why favour JSON, why not a real serialising format such as Storable?
		825
		826	In fact, any AnyEvent::MP node will happily accept Storable as framing
		827	format, but currently there is no way to make a node use Storable by
		828	default (although all nodes will accept it).
		829
		830	The default framing protocol is JSON because a) JSON::XS is many times
		831	faster for small messages and b) most importantly, after years of
		832	experience we found that object serialisation is causing more problems
		833	than it solves: Just like function calls, objects simply do not travel
		834	easily over the network, mostly because they will always be a copy, so you
		835	always have to re-think your design.
		836
		837	Keeping your messages simple, concentrating on data structures rather than
		838	objects, will keep your messages clean, tidy and efficient.
		839
		840	=back
		841
438	=head1 SEE ALSO	842	=head1 SEE ALSO
		843
		844	L<AnyEvent::MP::Intro> - a gentle introduction.
		845
		846	L<AnyEvent::MP::Kernel> - more, lower-level, stuff.
		847
		848	L<AnyEvent::MP::Global> - network maintainance and port groups, to find
		849	your applications.
439		850
440	L<AnyEvent>.	851	L<AnyEvent>.
441		852
442	=head1 AUTHOR	853	=head1 AUTHOR
443		854

Diff Legend

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

Comparing AnyEvent-MP/MP.pm (file contents): Revision 1.5 by root, Sat Aug 1 07:44:02 2009 UTC vs. Revision 1.75 by root, Mon Aug 31 13:18:06 2009 UTC

Diff Legend

Comparing AnyEvent-MP/MP.pm (file contents):
Revision 1.5 by root, Sat Aug 1 07:44:02 2009 UTC vs.
Revision 1.75 by root, Mon Aug 31 13:18:06 2009 UTC