ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/AnyEvent-MP/MP.pm
(Generate patch)

Comparing AnyEvent-MP/MP.pm (file contents):
Revision 1.37 by root, Fri Aug 7 16:47:23 2009 UTC vs.
Revision 1.71 by root, Sun Aug 30 19:52:56 2009 UTC

8 8
9 $NODE # contains this node's noderef 9 $NODE # contains this node's noderef
10 NODE # returns this node's noderef 10 NODE # returns this node's noderef
11 NODE $port # returns the noderef of the port 11 NODE $port # returns the noderef of the port
12 12
13 $SELF # receiving/own port id in rcv callbacks
14
15 # initialise the node so it can send/receive messages
16 initialise_node;
17
18 # ports are message endpoints
19
20 # sending messages
13 snd $port, type => data...; 21 snd $port, type => data...;
22 snd $port, @msg;
23 snd @msg_with_first_element_being_a_port;
14 24
15 $SELF # receiving/own port id in rcv callbacks 25 # creating/using ports, the simple way
26 my $simple_port = port { my @msg = @_; 0 };
16 27
17 rcv $port, smartmatch => $cb->($port, @msg); 28 # creating/using ports, tagged message matching
18 29 my $port = port;
19 # examples:
20 rcv $port2, ping => sub { snd $_[0], "pong"; 0 }; 30 rcv $port, ping => sub { snd $_[0], "pong"; 0 };
21 rcv $port1, pong => sub { warn "pong received\n" }; 31 rcv $port, pong => sub { warn "pong received\n"; 0 };
22 snd $port2, ping => $port1;
23 32
24 # more, smarter, matches (_any_ is exported by this module) 33 # create a port on another node
25 rcv $port, [child_died => $pid] => sub { ... 34 my $port = spawn $node, $initfunc, @initdata;
26 rcv $port, [_any_, _any_, 3] => sub { .. $_[2] is 3
27 35
28 # monitoring 36 # monitoring
29 mon $port, $cb->(@msg) # callback is invoked on death 37 mon $port, $cb->(@msg) # callback is invoked on death
30 mon $port, $otherport # kill otherport on abnormal death 38 mon $port, $otherport # kill otherport on abnormal death
31 mon $port, $otherport, @msg # send message on death 39 mon $port, $otherport, @msg # send message on death
32 40
41=head1 CURRENT STATUS
42
43 bin/aemp - stable.
44 AnyEvent::MP - stable API, should work.
45 AnyEvent::MP::Intro - uptodate, but incomplete.
46 AnyEvent::MP::Kernel - mostly stable.
47 AnyEvent::MP::Global - stable API, protocol not yet final.
48
49 stay tuned.
50
33=head1 DESCRIPTION 51=head1 DESCRIPTION
34 52
35This module (-family) implements a simple message passing framework. 53This module (-family) implements a simple message passing framework.
36 54
37Despite its simplicity, you can securely message other processes running 55Despite its simplicity, you can securely message other processes running
38on the same or other hosts. 56on the same or other hosts, and you can supervise entities remotely.
39 57
40For an introduction to this module family, see the L<AnyEvent::MP::Intro> 58For an introduction to this module family, see the L<AnyEvent::MP::Intro>
41manual page. 59manual page and the examples under F<eg/>.
42 60
43At the moment, this module family is severly broken and underdocumented, 61At the moment, this module family is a bit underdocumented.
44so do not use. This was uploaded mainly to reserve the CPAN namespace -
45stay tuned! The basic API should be finished, however.
46 62
47=head1 CONCEPTS 63=head1 CONCEPTS
48 64
49=over 4 65=over 4
50 66
51=item port 67=item port
52 68
53A port is something you can send messages to (with the C<snd> function). 69A port is something you can send messages to (with the C<snd> function).
54 70
55Some ports allow you to register C<rcv> handlers that can match specific 71Ports allow you to register C<rcv> handlers that can match all or just
56messages. All C<rcv> handlers will receive messages they match, messages 72some messages. Messages send to ports will not be queued, regardless of
57will not be queued. 73anything was listening for them or not.
58 74
59=item port id - C<noderef#portname> 75=item port ID - C<nodeid#portname>
60 76
61A port id is normaly the concatenation of a noderef, a hash-mark (C<#>) as 77A port ID is the concatenation of a node ID, a hash-mark (C<#>) as
62separator, and a port name (a printable string of unspecified format). An 78separator, and a port name (a printable string of unspecified format).
63exception is the the node port, whose ID is identical to its node
64reference.
65 79
66=item node 80=item node
67 81
68A node is a single process containing at least one port - the node 82A node is a single process containing at least one port - the node port,
69port. You can send messages to node ports to find existing ports or to 83which enables nodes to manage each other remotely, and to create new
70create new ports, among other things. 84ports.
71 85
72Nodes are either private (single-process only), slaves (connected to a 86Nodes are either public (have one or more listening ports) or private
73master node only) or public nodes (connectable from unrelated nodes). 87(no listening ports). Private nodes cannot talk to other private nodes
88currently.
74 89
75=item noderef - C<host:port,host:port...>, C<id@noderef>, C<id> 90=item node ID - C<[a-za-Z0-9_\-.:]+>
76 91
77A node reference is a string that either simply identifies the node (for 92A node ID is a string that uniquely identifies the node within a
78private and slave nodes), or contains a recipe on how to reach a given 93network. Depending on the configuration used, node IDs can look like a
79node (for public nodes). 94hostname, a hostname and a port, or a random string. AnyEvent::MP itself
95doesn't interpret node IDs in any way.
80 96
81This recipe is simply a comma-separated list of C<address:port> pairs (for 97=item binds - C<ip:port>
82TCP/IP, other protocols might look different).
83 98
84Node references come in two flavours: resolved (containing only numerical 99Nodes can only talk to each other by creating some kind of connection to
85addresses) or unresolved (where hostnames are used instead of addresses). 100each other. To do this, nodes should listen on one or more local transport
101endpoints - binds. Currently, only standard C<ip:port> specifications can
102be used, which specify TCP ports to listen on.
86 103
87Before using an unresolved node reference in a message you first have to 104=item seeds - C<host:port>
88resolve it. 105
106When a node starts, it knows nothing about the network. To teach the node
107about the network it first has to contact some other node within the
108network. This node is called a seed.
109
110Seeds are transport endpoint(s) of as many nodes as one wants. Those nodes
111are expected to be long-running, and at least one of those should always
112be available. When nodes run out of connections (e.g. due to a network
113error), they try to re-establish connections to some seednodes again to
114join the network.
115
116Apart from being sued for seeding, seednodes are not special in any way -
117every public node can be a seednode.
89 118
90=back 119=back
91 120
92=head1 VARIABLES/FUNCTIONS 121=head1 VARIABLES/FUNCTIONS
93 122
95 124
96=cut 125=cut
97 126
98package AnyEvent::MP; 127package AnyEvent::MP;
99 128
100use AnyEvent::MP::Base; 129use AnyEvent::MP::Kernel;
101 130
102use common::sense; 131use common::sense;
103 132
104use Carp (); 133use Carp ();
105 134
106use AE (); 135use AE ();
107 136
108use base "Exporter"; 137use base "Exporter";
109 138
110our $VERSION = '0.1'; 139our $VERSION = $AnyEvent::MP::Kernel::VERSION;
140
111our @EXPORT = qw( 141our @EXPORT = qw(
112 NODE $NODE *SELF node_of _any_ 142 NODE $NODE *SELF node_of after
113 resolve_node initialise_node 143 initialise_node
114 snd rcv mon kil reg psub 144 snd rcv mon mon_guard kil reg psub spawn
115 port 145 port
116); 146);
117 147
118our $SELF; 148our $SELF;
119 149
123 kil $SELF, die => $msg; 153 kil $SELF, die => $msg;
124} 154}
125 155
126=item $thisnode = NODE / $NODE 156=item $thisnode = NODE / $NODE
127 157
128The C<NODE> function returns, and the C<$NODE> variable contains 158The C<NODE> function returns, and the C<$NODE> variable contains, the node
129the noderef of the local node. The value is initialised by a call 159ID of the node running in the current process. This value is initialised by
130to C<become_public> or C<become_slave>, after which all local port 160a call to C<initialise_node>.
131identifiers become invalid.
132 161
133=item $noderef = node_of $port 162=item $nodeid = node_of $port
134 163
135Extracts and returns the noderef from a portid or a noderef. 164Extracts and returns the node ID from a port ID or a node ID.
136 165
137=item initialise_node $noderef, $seednode, $seednode... 166=item initialise_node $profile_name, key => value...
138 167
139=item initialise_node "slave/", $master, $master...
140
141Before a node can talk to other nodes on the network it has to initialise 168Before a node can talk to other nodes on the network (i.e. enter
142itself - the minimum a node needs to know is it's own name, and optionally 169"distributed mode") it has to initialise itself - the minimum a node needs
143it should know the noderefs of some other nodes in the network. 170to know is its own name, and optionally it should know the addresses of
171some other nodes in the network to discover other nodes.
144 172
145This function initialises a node - it must be called exactly once (or 173This function initialises a node - it must be called exactly once (or
146never) before calling other AnyEvent::MP functions. 174never) before calling other AnyEvent::MP functions.
147 175
148All arguments are noderefs, which can be either resolved or unresolved. 176The first argument is a profile name. If it is C<undef> or missing, then
177the current nodename will be used instead (i.e. F<uname -n>).
149 178
150There are two types of networked nodes, public nodes and slave nodes: 179The function first looks up the profile in the aemp configuration (see the
180L<aemp> commandline utility). the profile is calculated as follows:
151 181
152=over 4 182First, all remaining key => value pairs (all of which are conviniently
183undocumented at the moment) will be used. Then they will be overwritten by
184any values specified in the global default configuration (see the F<aemp>
185utility), then the chain of profiles selected, if any. That means that
186the values specified in the profile have highest priority and the values
187specified via C<initialise_node> have lowest priority.
153 188
154=item public nodes 189If the profile specifies a node ID, then this will become the node ID of
190this process. If not, then the profile name will be used as node ID. The
191special node ID of C<anon/> will be replaced by a random node ID.
155 192
156For public nodes, C<$noderef> must either be a (possibly unresolved) 193The next step is to look up the binds in the profile, followed by binding
157noderef, in which case it will be resolved, or C<undef> (or missing), in 194aemp protocol listeners on all binds specified (it is possible and valid
158which case the noderef will be guessed. 195to have no binds, meaning that the node cannot be contacted form the
196outside. This means the node cannot talk to other nodes that also have no
197binds, but it can still talk to all "normal" nodes).
159 198
160Afterwards, the node will bind itself on all endpoints and try to connect 199If the profile does not specify a binds list, then a default of C<*> is
161to all additional C<$seednodes> that are specified. Seednodes are optional 200used.
162and can be used to quickly bootstrap the node into an existing network.
163 201
164=item slave nodes 202Lastly, the seeds list from the profile is passed to the
203L<AnyEvent::MP::Global> module, which will then use it to keep
204connectivity with at least on of those seed nodes at any point in time.
165 205
166When the C<$noderef> is the special string C<slave/>, then the node will 206Example: become a distributed node listening on the guessed noderef, or
167become a slave node. Slave nodes cannot be contacted from outside and will 207the one specified via C<aemp> for the current node. This should be the
168route most of their traffic to the master node that they attach to. 208most common form of invocation for "daemon"-type nodes.
169
170At least one additional noderef is required: The node will try to connect
171to all of them and will become a slave attached to the first node it can
172successfully connect to.
173
174=back
175
176This function will block until all nodes have been resolved and, for slave
177nodes, until it has successfully established a connection to a master
178server.
179
180Example: become a public node listening on the default node.
181 209
182 initialise_node; 210 initialise_node;
183 211
184Example: become a public node, and try to contact some well-known master 212Example: become an anonymous node. This form is often used for commandline
185servers to become part of the network. 213clients.
186 214
187 initialise_node undef, "master1", "master2";
188
189Example: become a public node listening on port C<4041>.
190
191 initialise_node 4041; 215 initialise_node "anon/";
192 216
193Example: become a public node, only visible on localhost port 4044. 217Example: become a distributed node. If there is no profile of the given
218name, or no binds list was specified, resolve C<localhost:4044> and bind
219on the resulting addresses.
194 220
195 initialise_node "locahost:4044"; 221 initialise_node "localhost:4044";
196
197Example: become a slave node to any of the specified master servers.
198
199 initialise_node "slave/", "master1", "192.168.13.17", "mp.example.net";
200
201=item $cv = resolve_node $noderef
202
203Takes an unresolved node reference that may contain hostnames and
204abbreviated IDs, resolves all of them and returns a resolved node
205reference.
206
207In addition to C<address:port> pairs allowed in resolved noderefs, the
208following forms are supported:
209
210=over 4
211
212=item the empty string
213
214An empty-string component gets resolved as if the default port (4040) was
215specified.
216
217=item naked port numbers (e.g. C<1234>)
218
219These are resolved by prepending the local nodename and a colon, to be
220further resolved.
221
222=item hostnames (e.g. C<localhost:1234>, C<localhost>)
223
224These are resolved by using AnyEvent::DNS to resolve them, optionally
225looking up SRV records for the C<aemp=4040> port, if no port was
226specified.
227
228=back
229 222
230=item $SELF 223=item $SELF
231 224
232Contains the current port id while executing C<rcv> callbacks or C<psub> 225Contains the current port id while executing C<rcv> callbacks or C<psub>
233blocks. 226blocks.
234 227
235=item SELF, %SELF, @SELF... 228=item *SELF, SELF, %SELF, @SELF...
236 229
237Due to some quirks in how perl exports variables, it is impossible to 230Due to some quirks in how perl exports variables, it is impossible to
238just export C<$SELF>, all the symbols called C<SELF> are exported by this 231just export C<$SELF>, all the symbols named C<SELF> are exported by this
239module, but only C<$SELF> is currently used. 232module, but only C<$SELF> is currently used.
240 233
241=item snd $port, type => @data 234=item snd $port, type => @data
242 235
243=item snd $port, @msg 236=item snd $port, @msg
244 237
245Send the given message to the given port ID, which can identify either 238Send the given message to the given port, which can identify either a
246a local or a remote port, and can be either a string or soemthignt hat 239local or a remote port, and must be a port ID.
247stringifies a sa port ID (such as a port object :).
248 240
249While the message can be about anything, it is highly recommended to use a 241While the message can be almost anything, it is highly recommended to
250string as first element (a portid, or some word that indicates a request 242use a string as first element (a port ID, or some word that indicates a
251type etc.). 243request type etc.) and to consist if only simple perl values (scalars,
244arrays, hashes) - if you think you need to pass an object, think again.
252 245
253The message data effectively becomes read-only after a call to this 246The message data logically becomes read-only after a call to this
254function: modifying any argument is not allowed and can cause many 247function: modifying any argument (or values referenced by them) is
255problems. 248forbidden, as there can be considerable time between the call to C<snd>
249and the time the message is actually being serialised - in fact, it might
250never be copied as within the same process it is simply handed to the
251receiving port.
256 252
257The type of data you can transfer depends on the transport protocol: when 253The type of data you can transfer depends on the transport protocol: when
258JSON is used, then only strings, numbers and arrays and hashes consisting 254JSON is used, then only strings, numbers and arrays and hashes consisting
259of those are allowed (no objects). When Storable is used, then anything 255of those are allowed (no objects). When Storable is used, then anything
260that Storable can serialise and deserialise is allowed, and for the local 256that Storable can serialise and deserialise is allowed, and for the local
261node, anything can be passed. 257node, anything can be passed. Best rely only on the common denominator of
258these.
262 259
263=item $local_port = port 260=item $local_port = port
264 261
265Create a new local port object that can be used either as a pattern 262Create a new local port object and returns its port ID. Initially it has
266matching port ("full port") or a single-callback port ("miniport"), 263no callbacks set and will throw an error when it receives messages.
267depending on how C<rcv> callbacks are bound to the object.
268 264
269=item $port = port { my @msg = @_; $finished } 265=item $local_port = port { my @msg = @_ }
270 266
271Creates a "miniport", that is, a very lightweight port without any pattern 267Creates a new local port, and returns its ID. Semantically the same as
272matching behind it, and returns its ID. Semantically the same as creating
273a port and calling C<rcv $port, $callback> on it. 268creating a port and calling C<rcv $port, $callback> on it.
274 269
275The block will be called for every message received on the port. When the 270The block will be called for every message received on the port, with the
276callback returns a true value its job is considered "done" and the port 271global variable C<$SELF> set to the port ID. Runtime errors will cause the
277will be destroyed. Otherwise it will stay alive. 272port to be C<kil>ed. The message will be passed as-is, no extra argument
273(i.e. no port ID) will be passed to the callback.
278 274
279The message will be passed as-is, no extra argument (i.e. no port id) will 275If you want to stop/destroy the port, simply C<kil> it:
280be passed to the callback.
281 276
282If you need the local port id in the callback, this works nicely: 277 my $port = port {
283 278 my @msg = @_;
284 my $port; $port = port { 279 ...
285 snd $otherport, reply => $port; 280 kil $SELF;
286 }; 281 };
287 282
288=cut 283=cut
289 284
290sub rcv($@); 285sub rcv($@);
286
287sub _kilme {
288 die "received message on port without callback";
289}
291 290
292sub port(;&) { 291sub port(;&) {
293 my $id = "$UNIQ." . $ID++; 292 my $id = "$UNIQ." . $ID++;
294 my $port = "$NODE#$id"; 293 my $port = "$NODE#$id";
295 294
296 if (@_) { 295 rcv $port, shift || \&_kilme;
297 rcv $port, shift;
298 } else {
299 $PORT{$id} = sub { }; # nop
300 }
301 296
302 $port 297 $port
303} 298}
304 299
305=item reg $port, $name
306
307=item reg $name
308
309Registers the given port (or C<$SELF><<< if missing) under the name
310C<$name>. If the name already exists it is replaced.
311
312A port can only be registered under one well known name.
313
314A port automatically becomes unregistered when it is killed.
315
316=cut
317
318sub reg(@) {
319 my $port = @_ > 1 ? shift : $SELF || Carp::croak 'reg: called with one argument only, but $SELF not set,';
320
321 $REG{$_[0]} = $port;
322}
323
324=item rcv $port, $callback->(@msg) 300=item rcv $local_port, $callback->(@msg)
325 301
326Replaces the callback on the specified miniport (after converting it to 302Replaces the default callback on the specified port. There is no way to
327one if required). 303remove the default callback: use C<sub { }> to disable it, or better
328 304C<kil> the port when it is no longer needed.
329=item rcv $port, tagstring => $callback->(@msg), ...
330
331=item rcv $port, $smartmatch => $callback->(@msg), ...
332
333=item rcv $port, [$smartmatch...] => $callback->(@msg), ...
334
335Register callbacks to be called on matching messages on the given full
336port (after converting it to one if required) and return the port.
337
338The callback has to return a true value when its work is done, after
339which is will be removed, or a false value in which case it will stay
340registered.
341 305
342The global C<$SELF> (exported by this module) contains C<$port> while 306The global C<$SELF> (exported by this module) contains C<$port> while
343executing the callback. 307executing the callback. Runtime errors during callback execution will
308result in the port being C<kil>ed.
344 309
345Runtime errors wdurign callback execution will result in the port being 310The default callback received all messages not matched by a more specific
346C<kil>ed. 311C<tag> match.
347 312
348If the match is an array reference, then it will be matched against the 313=item rcv $local_port, tag => $callback->(@msg_without_tag), ...
349first elements of the message, otherwise only the first element is being
350matched.
351 314
352Any element in the match that is specified as C<_any_> (a function 315Register (or replace) callbacks to be called on messages starting with the
353exported by this module) matches any single element of the message. 316given tag on the given port (and return the port), or unregister it (when
317C<$callback> is C<$undef> or missing). There can only be one callback
318registered for each tag.
354 319
355While not required, it is highly recommended that the first matching 320The original message will be passed to the callback, after the first
356element is a string identifying the message. The one-string-only match is 321element (the tag) has been removed. The callback will use the same
357also the most efficient match (by far). 322environment as the default callback (see above).
358 323
359Example: create a port and bind receivers on it in one go. 324Example: create a port and bind receivers on it in one go.
360 325
361 my $port = rcv port, 326 my $port = rcv port,
362 msg1 => sub { ...; 0 }, 327 msg1 => sub { ... },
363 msg2 => sub { ...; 0 }, 328 msg2 => sub { ... },
364 ; 329 ;
365 330
366Example: create a port, bind receivers and send it in a message elsewhere 331Example: create a port, bind receivers and send it in a message elsewhere
367in one go: 332in one go:
368 333
369 snd $otherport, reply => 334 snd $otherport, reply =>
370 rcv port, 335 rcv port,
371 msg1 => sub { ...; 0 }, 336 msg1 => sub { ... },
372 ... 337 ...
373 ; 338 ;
339
340Example: temporarily register a rcv callback for a tag matching some port
341(e.g. for a rpc reply) and unregister it after a message was received.
342
343 rcv $port, $otherport => sub {
344 my @reply = @_;
345
346 rcv $SELF, $otherport;
347 };
374 348
375=cut 349=cut
376 350
377sub rcv($@) { 351sub rcv($@) {
378 my $port = shift; 352 my $port = shift;
379 my ($noderef, $portid) = split /#/, $port, 2; 353 my ($noderef, $portid) = split /#/, $port, 2;
380 354
381 ($NODE{$noderef} || add_node $noderef) == $NODE{""} 355 $NODE{$noderef} == $NODE{""}
382 or Carp::croak "$port: rcv can only be called on local ports, caught"; 356 or Carp::croak "$port: rcv can only be called on local ports, caught";
383 357
384 if (@_ == 1) { 358 while (@_) {
359 if (ref $_[0]) {
360 if (my $self = $PORT_DATA{$portid}) {
361 "AnyEvent::MP::Port" eq ref $self
362 or Carp::croak "$port: rcv can only be called on message matching ports, caught";
363
364 $self->[2] = shift;
365 } else {
385 my $cb = shift; 366 my $cb = shift;
386 delete $PORT_DATA{$portid};
387 $PORT{$portid} = sub { 367 $PORT{$portid} = sub {
388 local $SELF = $port; 368 local $SELF = $port;
389 eval { 369 eval { &$cb }; _self_die if $@;
390 &$cb 370 };
391 and kil $port;
392 }; 371 }
393 _self_die if $@; 372 } elsif (defined $_[0]) {
394 };
395 } else {
396 my $self = $PORT_DATA{$portid} ||= do { 373 my $self = $PORT_DATA{$portid} ||= do {
397 my $self = bless { 374 my $self = bless [$PORT{$port} || sub { }, { }, $port], "AnyEvent::MP::Port";
398 id => $port,
399 }, "AnyEvent::MP::Port";
400 375
401 $PORT{$portid} = sub { 376 $PORT{$portid} = sub {
402 local $SELF = $port; 377 local $SELF = $port;
403 378
404 eval {
405 for (@{ $self->{rc0}{$_[0]} }) { 379 if (my $cb = $self->[1]{$_[0]}) {
406 $_ && &{$_->[0]} 380 shift;
407 && undef $_; 381 eval { &$cb }; _self_die if $@;
408 } 382 } else {
409
410 for (@{ $self->{rcv}{$_[0]} }) {
411 $_ && [@_[1 .. @{$_->[1]}]] ~~ $_->[1]
412 && &{$_->[0]} 383 &{ $self->[0] };
413 && undef $_;
414 }
415
416 for (@{ $self->{any} }) {
417 $_ && [@_[0 .. $#{$_->[1]}]] ~~ $_->[1]
418 && &{$_->[0]}
419 && undef $_;
420 } 384 }
421 }; 385 };
422 _self_die if $@; 386
387 $self
423 }; 388 };
424 389
425 $self
426 };
427
428 "AnyEvent::MP::Port" eq ref $self 390 "AnyEvent::MP::Port" eq ref $self
429 or Carp::croak "$port: rcv can only be called on message matching ports, caught"; 391 or Carp::croak "$port: rcv can only be called on message matching ports, caught";
430 392
431 while (@_) {
432 my ($match, $cb) = splice @_, 0, 2; 393 my ($tag, $cb) = splice @_, 0, 2;
433 394
434 if (!ref $match) { 395 if (defined $cb) {
435 push @{ $self->{rc0}{$match} }, [$cb]; 396 $self->[1]{$tag} = $cb;
436 } elsif (("ARRAY" eq ref $match && !ref $match->[0])) {
437 my ($type, @match) = @$match;
438 @match
439 ? push @{ $self->{rcv}{$match->[0]} }, [$cb, \@match]
440 : push @{ $self->{rc0}{$match->[0]} }, [$cb];
441 } else { 397 } else {
442 push @{ $self->{any} }, [$cb, $match]; 398 delete $self->[1]{$tag};
443 } 399 }
444 } 400 }
445 } 401 }
446 402
447 $port 403 $port
483 $res 439 $res
484 } 440 }
485 } 441 }
486} 442}
487 443
488=item $guard = mon $port, $cb->(@reason) 444=item $guard = mon $port, $cb->(@reason) # call $cb when $port dies
489 445
490=item $guard = mon $port, $rcvport 446=item $guard = mon $port, $rcvport # kill $rcvport when $port dies
491 447
492=item $guard = mon $port 448=item $guard = mon $port # kill $SELF when $port dies
493 449
494=item $guard = mon $port, $rcvport, @msg 450=item $guard = mon $port, $rcvport, @msg # send a message when $port dies
495 451
496Monitor the given port and do something when the port is killed, and 452Monitor the given port and do something when the port is killed or
497optionally return a guard that can be used to stop monitoring again. 453messages to it were lost, and optionally return a guard that can be used
454to stop monitoring again.
455
456C<mon> effectively guarantees that, in the absence of hardware failures,
457after starting the monitor, either all messages sent to the port will
458arrive, or the monitoring action will be invoked after possible message
459loss has been detected. No messages will be lost "in between" (after
460the first lost message no further messages will be received by the
461port). After the monitoring action was invoked, further messages might get
462delivered again.
463
464Note that monitoring-actions are one-shot: once messages are lost (and a
465monitoring alert was raised), they are removed and will not trigger again.
498 466
499In the first form (callback), the callback is simply called with any 467In the first form (callback), the callback is simply called with any
500number of C<@reason> elements (no @reason means that the port was deleted 468number of C<@reason> elements (no @reason means that the port was deleted
501"normally"). Note also that I<< the callback B<must> never die >>, so use 469"normally"). Note also that I<< the callback B<must> never die >>, so use
502C<eval> if unsure. 470C<eval> if unsure.
503 471
504In the second form (another port given), the other port (C<$rcvport) 472In the second form (another port given), the other port (C<$rcvport>)
505will be C<kil>'ed with C<@reason>, iff a @reason was specified, i.e. on 473will be C<kil>'ed with C<@reason>, iff a @reason was specified, i.e. on
506"normal" kils nothing happens, while under all other conditions, the other 474"normal" kils nothing happens, while under all other conditions, the other
507port is killed with the same reason. 475port is killed with the same reason.
508 476
509The third form (kill self) is the same as the second form, except that 477The third form (kill self) is the same as the second form, except that
536sub mon { 504sub mon {
537 my ($noderef, $port) = split /#/, shift, 2; 505 my ($noderef, $port) = split /#/, shift, 2;
538 506
539 my $node = $NODE{$noderef} || add_node $noderef; 507 my $node = $NODE{$noderef} || add_node $noderef;
540 508
541 my $cb = @_ ? $_[0] : $SELF || Carp::croak 'mon: called with one argument only, but $SELF not set,'; 509 my $cb = @_ ? shift : $SELF || Carp::croak 'mon: called with one argument only, but $SELF not set,';
542 510
543 unless (ref $cb) { 511 unless (ref $cb) {
544 if (@_) { 512 if (@_) {
545 # send a kill info message 513 # send a kill info message
546 my (@msg) = @_; 514 my (@msg) = ($cb, @_);
547 $cb = sub { snd @msg, @_ }; 515 $cb = sub { snd @msg, @_ };
548 } else { 516 } else {
549 # simply kill other port 517 # simply kill other port
550 my $port = $cb; 518 my $port = $cb;
551 $cb = sub { kil $port, @_ if @_ }; 519 $cb = sub { kil $port, @_ if @_ };
564is killed, the references will be freed. 532is killed, the references will be freed.
565 533
566Optionally returns a guard that will stop the monitoring. 534Optionally returns a guard that will stop the monitoring.
567 535
568This function is useful when you create e.g. timers or other watchers and 536This function is useful when you create e.g. timers or other watchers and
569want to free them when the port gets killed: 537want to free them when the port gets killed (note the use of C<psub>):
570 538
571 $port->rcv (start => sub { 539 $port->rcv (start => sub {
572 my $timer; $timer = mon_guard $port, AE::timer 1, 1, sub { 540 my $timer; $timer = mon_guard $port, AE::timer 1, 1, psub {
573 undef $timer if 0.9 < rand; 541 undef $timer if 0.9 < rand;
574 }); 542 });
575 }); 543 });
576 544
577=cut 545=cut
586 554
587=item kil $port[, @reason] 555=item kil $port[, @reason]
588 556
589Kill the specified port with the given C<@reason>. 557Kill the specified port with the given C<@reason>.
590 558
591If no C<@reason> is specified, then the port is killed "normally" (linked 559If no C<@reason> is specified, then the port is killed "normally" (ports
592ports will not be kileld, or even notified). 560monitoring other ports will not necessarily die because a port dies
561"normally").
593 562
594Otherwise, linked ports get killed with the same reason (second form of 563Otherwise, linked ports get killed with the same reason (second form of
595C<mon>, see below). 564C<mon>, see above).
596 565
597Runtime errors while evaluating C<rcv> callbacks or inside C<psub> blocks 566Runtime errors while evaluating C<rcv> callbacks or inside C<psub> blocks
598will be reported as reason C<< die => $@ >>. 567will be reported as reason C<< die => $@ >>.
599 568
600Transport/communication errors are reported as C<< transport_error => 569Transport/communication errors are reported as C<< transport_error =>
601$message >>. 570$message >>.
602 571
603=back
604
605=head1 NODE MESSAGES
606
607Nodes understand the following messages sent to them. Many of them take
608arguments called C<@reply>, which will simply be used to compose a reply
609message - C<$reply[0]> is the port to reply to, C<$reply[1]> the type and
610the remaining arguments are simply the message data.
611
612While other messages exist, they are not public and subject to change.
613
614=over 4
615
616=cut 572=cut
617 573
618=item lookup => $name, @reply 574=item $port = spawn $node, $initfunc[, @initdata]
619 575
620Replies with the port ID of the specified well-known port, or C<undef>. 576Creates a port on the node C<$node> (which can also be a port ID, in which
577case it's the node where that port resides).
621 578
622=item devnull => ... 579The port ID of the newly created port is returned immediately, and it is
580possible to immediately start sending messages or to monitor the port.
623 581
624Generic data sink/CPU heat conversion. 582After the port has been created, the init function is called on the remote
583node, in the same context as a C<rcv> callback. This function must be a
584fully-qualified function name (e.g. C<MyApp::Chat::Server::init>). To
585specify a function in the main program, use C<::name>.
625 586
626=item relay => $port, @msg 587If the function doesn't exist, then the node tries to C<require>
588the package, then the package above the package and so on (e.g.
589C<MyApp::Chat::Server>, C<MyApp::Chat>, C<MyApp>) until the function
590exists or it runs out of package names.
627 591
628Simply forwards the message to the given port. 592The init function is then called with the newly-created port as context
593object (C<$SELF>) and the C<@initdata> values as arguments.
629 594
630=item eval => $string[ @reply] 595A common idiom is to pass a local port, immediately monitor the spawned
596port, and in the remote init function, immediately monitor the passed
597local port. This two-way monitoring ensures that both ports get cleaned up
598when there is a problem.
631 599
632Evaluates the given string. If C<@reply> is given, then a message of the 600Example: spawn a chat server port on C<$othernode>.
633form C<@reply, $@, @evalres> is sent.
634 601
635Example: crash another node. 602 # this node, executed from within a port context:
603 my $server = spawn $othernode, "MyApp::Chat::Server::connect", $SELF;
604 mon $server;
636 605
637 snd $othernode, eval => "exit"; 606 # init function on C<$othernode>
607 sub connect {
608 my ($srcport) = @_;
638 609
639=item time => @reply 610 mon $srcport;
640 611
641Replies the the current node time to C<@reply>. 612 rcv $SELF, sub {
613 ...
614 };
615 }
642 616
643Example: tell the current node to send the current time to C<$myport> in a 617=cut
644C<timereply> message.
645 618
646 snd $NODE, time => $myport, timereply => 1, 2; 619sub _spawn {
647 # => snd $myport, timereply => 1, 2, <time> 620 my $port = shift;
621 my $init = shift;
622
623 local $SELF = "$NODE#$port";
624 eval {
625 &{ load_func $init }
626 };
627 _self_die if $@;
628}
629
630sub spawn(@) {
631 my ($noderef, undef) = split /#/, shift, 2;
632
633 my $id = "$RUNIQ." . $ID++;
634
635 $_[0] =~ /::/
636 or Carp::croak "spawn init function must be a fully-qualified name, caught";
637
638 snd_to_func $noderef, "AnyEvent::MP::_spawn" => $id, @_;
639
640 "$noderef#$id"
641}
642
643=item after $timeout, @msg
644
645=item after $timeout, $callback
646
647Either sends the given message, or call the given callback, after the
648specified number of seconds.
649
650This is simply a utility function that comes in handy at times - the
651AnyEvent::MP author is not convinced of the wisdom of having it, though,
652so it may go away in the future.
653
654=cut
655
656sub after($@) {
657 my ($timeout, @action) = @_;
658
659 my $t; $t = AE::timer $timeout, 0, sub {
660 undef $t;
661 ref $action[0]
662 ? $action[0]()
663 : snd @action;
664 };
665}
648 666
649=back 667=back
650 668
651=head1 AnyEvent::MP vs. Distributed Erlang 669=head1 AnyEvent::MP vs. Distributed Erlang
652 670
662 680
663Despite the similarities, there are also some important differences: 681Despite the similarities, there are also some important differences:
664 682
665=over 4 683=over 4
666 684
667=item * Node references contain the recipe on how to contact them. 685=item * Node IDs are arbitrary strings in AEMP.
668 686
669Erlang relies on special naming and DNS to work everywhere in the 687Erlang relies on special naming and DNS to work everywhere in the same
670same way. AEMP relies on each node knowing it's own address(es), with 688way. AEMP relies on each node somehow knowing its own address(es) (e.g. by
671convenience functionality. 689configuraiton or DNS), but will otherwise discover other odes itself.
672 690
673This means that AEMP requires a less tightly controlled environment at the 691=item * Erlang has a "remote ports are like local ports" philosophy, AEMP
674cost of longer node references and a slightly higher management overhead. 692uses "local ports are like remote ports".
693
694The failure modes for local ports are quite different (runtime errors
695only) then for remote ports - when a local port dies, you I<know> it dies,
696when a connection to another node dies, you know nothing about the other
697port.
698
699Erlang pretends remote ports are as reliable as local ports, even when
700they are not.
701
702AEMP encourages a "treat remote ports differently" philosophy, with local
703ports being the special case/exception, where transport errors cannot
704occur.
675 705
676=item * Erlang uses processes and a mailbox, AEMP does not queue. 706=item * Erlang uses processes and a mailbox, AEMP does not queue.
677 707
678Erlang uses processes that selctively receive messages, and therefore 708Erlang uses processes that selectively receive messages, and therefore
679needs a queue. AEMP is event based, queuing messages would serve no useful 709needs a queue. AEMP is event based, queuing messages would serve no
680purpose. 710useful purpose. For the same reason the pattern-matching abilities of
711AnyEvent::MP are more limited, as there is little need to be able to
712filter messages without dequeing them.
681 713
682(But see L<Coro::MP> for a more Erlang-like process model on top of AEMP). 714(But see L<Coro::MP> for a more Erlang-like process model on top of AEMP).
683 715
684=item * Erlang sends are synchronous, AEMP sends are asynchronous. 716=item * Erlang sends are synchronous, AEMP sends are asynchronous.
685 717
686Sending messages in Erlang is synchronous and blocks the process. AEMP 718Sending messages in Erlang is synchronous and blocks the process (and
687sends are immediate, connection establishment is handled in the 719so does not need a queue that can overflow). AEMP sends are immediate,
688background. 720connection establishment is handled in the background.
689 721
690=item * Erlang can silently lose messages, AEMP cannot. 722=item * Erlang suffers from silent message loss, AEMP does not.
691 723
692Erlang makes few guarantees on messages delivery - messages can get lost 724Erlang makes few guarantees on messages delivery - messages can get lost
693without any of the processes realising it (i.e. you send messages a, b, 725without any of the processes realising it (i.e. you send messages a, b,
694and c, and the other side only receives messages a and c). 726and c, and the other side only receives messages a and c).
695 727
696AEMP guarantees correct ordering, and the guarantee that there are no 728AEMP guarantees correct ordering, and the guarantee that after one message
697holes in the message sequence. 729is lost, all following ones sent to the same port are lost as well, until
698 730monitoring raises an error, so there are no silent "holes" in the message
699=item * In Erlang, processes can be declared dead and later be found to be 731sequence.
700alive.
701
702In Erlang it can happen that a monitored process is declared dead and
703linked processes get killed, but later it turns out that the process is
704still alive - and can receive messages.
705
706In AEMP, when port monitoring detects a port as dead, then that port will
707eventually be killed - it cannot happen that a node detects a port as dead
708and then later sends messages to it, finding it is still alive.
709 732
710=item * Erlang can send messages to the wrong port, AEMP does not. 733=item * Erlang can send messages to the wrong port, AEMP does not.
711 734
712In Erlang it is quite possible that a node that restarts reuses a process 735In Erlang it is quite likely that a node that restarts reuses a process ID
713ID known to other nodes for a completely different process, causing 736known to other nodes for a completely different process, causing messages
714messages destined for that process to end up in an unrelated process. 737destined for that process to end up in an unrelated process.
715 738
716AEMP never reuses port IDs, so old messages or old port IDs floating 739AEMP never reuses port IDs, so old messages or old port IDs floating
717around in the network will not be sent to an unrelated port. 740around in the network will not be sent to an unrelated port.
718 741
719=item * Erlang uses unprotected connections, AEMP uses secure 742=item * Erlang uses unprotected connections, AEMP uses secure
720authentication and can use TLS. 743authentication and can use TLS.
721 744
722AEMP can use a proven protocol - SSL/TLS - to protect connections and 745AEMP can use a proven protocol - TLS - to protect connections and
723securely authenticate nodes. 746securely authenticate nodes.
724 747
725=item * The AEMP protocol is optimised for both text-based and binary 748=item * The AEMP protocol is optimised for both text-based and binary
726communications. 749communications.
727 750
728The AEMP protocol, unlike the Erlang protocol, supports both 751The AEMP protocol, unlike the Erlang protocol, supports both programming
729language-independent text-only protocols (good for debugging) and binary, 752language independent text-only protocols (good for debugging) and binary,
730language-specific serialisers (e.g. Storable). 753language-specific serialisers (e.g. Storable). By default, unless TLS is
754used, the protocol is actually completely text-based.
731 755
732It has also been carefully designed to be implementable in other languages 756It has also been carefully designed to be implementable in other languages
733with a minimum of work while gracefully degrading fucntionality to make the 757with a minimum of work while gracefully degrading functionality to make the
734protocol simple. 758protocol simple.
735 759
736=item * AEMP has more flexible monitoring options than Erlang. 760=item * AEMP has more flexible monitoring options than Erlang.
737 761
738In Erlang, you can chose to receive I<all> exit signals as messages 762In Erlang, you can chose to receive I<all> exit signals as messages
741Erlang, as one can choose between automatic kill, exit message or callback 765Erlang, as one can choose between automatic kill, exit message or callback
742on a per-process basis. 766on a per-process basis.
743 767
744=item * Erlang tries to hide remote/local connections, AEMP does not. 768=item * Erlang tries to hide remote/local connections, AEMP does not.
745 769
746Monitoring in Erlang is not an indicator of process death/crashes, 770Monitoring in Erlang is not an indicator of process death/crashes, in the
747as linking is (except linking is unreliable in Erlang). 771same way as linking is (except linking is unreliable in Erlang).
748 772
749In AEMP, you don't "look up" registered port names or send to named ports 773In AEMP, you don't "look up" registered port names or send to named ports
750that might or might not be persistent. Instead, you normally spawn a port 774that might or might not be persistent. Instead, you normally spawn a port
751on the remote node. The init function monitors the you, and you monitor 775on the remote node. The init function monitors you, and you monitor the
752the remote port. Since both monitors are local to the node, they are much 776remote port. Since both monitors are local to the node, they are much more
753more reliable. 777reliable (no need for C<spawn_link>).
754 778
755This also saves round-trips and avoids sending messages to the wrong port 779This also saves round-trips and avoids sending messages to the wrong port
756(hard to do in Erlang). 780(hard to do in Erlang).
757 781
758=back 782=back
759 783
784=head1 RATIONALE
785
786=over 4
787
788=item Why strings for port and node IDs, why not objects?
789
790We considered "objects", but found that the actual number of methods
791that can be called are quite low. Since port and node IDs travel over
792the network frequently, the serialising/deserialising would add lots of
793overhead, as well as having to keep a proxy object everywhere.
794
795Strings can easily be printed, easily serialised etc. and need no special
796procedures to be "valid".
797
798And as a result, a miniport consists of a single closure stored in a
799global hash - it can't become much cheaper.
800
801=item Why favour JSON, why not a real serialising format such as Storable?
802
803In fact, any AnyEvent::MP node will happily accept Storable as framing
804format, but currently there is no way to make a node use Storable by
805default (although all nodes will accept it).
806
807The default framing protocol is JSON because a) JSON::XS is many times
808faster for small messages and b) most importantly, after years of
809experience we found that object serialisation is causing more problems
810than it solves: Just like function calls, objects simply do not travel
811easily over the network, mostly because they will always be a copy, so you
812always have to re-think your design.
813
814Keeping your messages simple, concentrating on data structures rather than
815objects, will keep your messages clean, tidy and efficient.
816
817=back
818
760=head1 SEE ALSO 819=head1 SEE ALSO
820
821L<AnyEvent::MP::Intro> - a gentle introduction.
822
823L<AnyEvent::MP::Kernel> - more, lower-level, stuff.
824
825L<AnyEvent::MP::Global> - network maintainance and port groups, to find
826your applications.
761 827
762L<AnyEvent>. 828L<AnyEvent>.
763 829
764=head1 AUTHOR 830=head1 AUTHOR
765 831

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines