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 1.22 by root, Tue Aug 4 18:33:30 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 # contains this node's noderef 9 $NODE # contains this node's node ID
10 NODE # returns this node's noderef 10 NODE # returns this node's node ID
11 NODE $port # returns the noderef of the port
12 11
12 $SELF # receiving/own port id in rcv callbacks
13
14 # initialise the node so it can send/receive messages
15 configure;
16
17 # ports are message destinations
18
19 # sending messages
13 snd $port, type => data...; 20 snd $port, type => data...;
21 snd $port, @msg;
22 snd @msg_with_first_element_being_a_port;
14 23
15 $SELF # receiving/own port id in rcv callbacks 24 # creating/using ports, the simple way
25 my $simple_port = port { my @msg = @_ };
16 26
17 rcv $port, smartmatch => $cb->($port, @msg); 27 # creating/using ports, tagged message matching
18 28 my $port = port;
19 # examples:
20 rcv $port2, ping => sub { snd $_[0], "pong"; 0 }; 29 rcv $port, ping => sub { snd $_[0], "pong" };
21 rcv $port1, pong => sub { warn "pong received\n" }; 30 rcv $port, pong => sub { warn "pong received\n" };
22 snd $port2, ping => $port1;
23 31
24 # more, smarter, matches (_any_ is exported by this module) 32 # create a port on another node
25 rcv $port, [child_died => $pid] => sub { ... 33 my $port = spawn $node, $initfunc, @initdata;
26 rcv $port, [_any_, _any_, 3] => sub { .. $_[2] is 3 34
35 # 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.
27 49
28=head1 DESCRIPTION 50=head1 DESCRIPTION
29 51
30This module (-family) implements a simple message passing framework. 52This module (-family) implements a simple message passing framework.
31 53
32Despite its simplicity, you can securely message other processes running 54Despite its simplicity, you can securely message other processes running
33on the same or other hosts. 55on the same or other hosts, and you can supervise entities remotely.
34 56
57For an introduction to this module family, see the L<AnyEvent::MP::Intro>
58manual page and the examples under F<eg/>.
59
35At the moment, this module family is severly brokena nd underdocumented, 60At the moment, this module family is a bit underdocumented.
36so do not use. This was uploaded mainly to reserve the CPAN namespace -
37stay tuned!
38 61
39=head1 CONCEPTS 62=head1 CONCEPTS
40 63
41=over 4 64=over 4
42 65
43=item port 66=item port
44 67
45A port is something you can send messages to with the C<snd> function, and 68A port is something you can send messages to (with the C<snd> function).
46you can register C<rcv> handlers with. All C<rcv> handlers will receive
47messages they match, messages will not be queued.
48 69
70Ports allow you to register C<rcv> handlers that can match all or just
71some messages. Messages send to ports will not be queued, regardless of
72anything was listening for them or not.
73
49=item port id - C<noderef#portname> 74=item port ID - C<nodeid#portname>
50 75
51A port id is always the noderef, a hash-mark (C<#>) as separator, followed 76A port ID is the concatenation of a node ID, a hash-mark (C<#>) as
52by a port name (a printable string of unspecified format). 77separator, and a port name (a printable string of unspecified format).
53 78
54=item node 79=item node
55 80
56A node is a single process containing at least one port - the node 81A node is a single process containing at least one port - the node port,
57port. You can send messages to node ports to let them create new ports, 82which enables nodes to manage each other remotely, and to create new
58among other things. 83ports.
59 84
60Initially, nodes are either private (single-process only) or hidden 85Nodes are either public (have one or more listening ports) or private
61(connected to a master node only). Only when they epxlicitly "become 86(no listening ports). Private nodes cannot talk to other private nodes
62public" can you send them messages from unrelated other nodes. 87currently.
63 88
64=item noderef - C<host:port,host:port...>, C<id@noderef>, C<id> 89=item node ID - C<[a-za-Z0-9_\-.:]+>
65 90
66A noderef is a string that either uniquely identifies a given node (for 91A node ID is a string that uniquely identifies the node within a
67private and hidden nodes), or contains a recipe on how to reach a given 92network. Depending on the configuration used, node IDs can look like a
68node (for public nodes). 93hostname, a hostname and a port, or a random string. AnyEvent::MP itself
94doesn't interpret node IDs in any way.
95
96=item binds - C<ip:port>
97
98Nodes can only talk to each other by creating some kind of connection to
99each other. To do this, nodes should listen on one or more local transport
100endpoints - binds. Currently, only standard C<ip:port> specifications can
101be used, which specify TCP ports to listen on.
102
103=item seeds - C<host:port>
104
105When a node starts, it knows nothing about the network. To teach the node
106about the network it first has to contact some other node within the
107network. This node is called a seed.
108
109Seeds are transport endpoint(s) of as many nodes as one wants. Those nodes
110are expected to be long-running, and at least one of those should always
111be available. When nodes run out of connections (e.g. due to a network
112error), they try to re-establish connections to some seednodes again to
113join the network.
114
115Apart from being sued for seeding, seednodes are not special in any way -
116every public node can be a seednode.
69 117
70=back 118=back
71 119
72=head1 VARIABLES/FUNCTIONS 120=head1 VARIABLES/FUNCTIONS
73 121
75 123
76=cut 124=cut
77 125
78package AnyEvent::MP; 126package AnyEvent::MP;
79 127
80use AnyEvent::MP::Base; 128use AnyEvent::MP::Kernel;
81 129
82use common::sense; 130use common::sense;
83 131
84use Carp (); 132use Carp ();
85 133
86use AE (); 134use AE ();
87 135
88use base "Exporter"; 136use base "Exporter";
89 137
90our $VERSION = '0.02'; 138our $VERSION = $AnyEvent::MP::Kernel::VERSION;
139
91our @EXPORT = qw( 140our @EXPORT = qw(
92 NODE $NODE *SELF node_of _any_ 141 NODE $NODE *SELF node_of after
93 become_slave become_public 142 configure
94 snd rcv mon kil reg psub 143 snd rcv mon mon_guard kil reg psub spawn
95 port 144 port
96); 145);
97 146
98our $SELF; 147our $SELF;
99 148
103 kil $SELF, die => $msg; 152 kil $SELF, die => $msg;
104} 153}
105 154
106=item $thisnode = NODE / $NODE 155=item $thisnode = NODE / $NODE
107 156
108The C<NODE> function returns, and the C<$NODE> variable contains 157The C<NODE> function returns, and the C<$NODE> variable contains, the node
109the noderef of the local node. The value is initialised by a call 158ID of the node running in the current process. This value is initialised by
110to C<become_public> or C<become_slave>, after which all local port 159a call to C<configure>.
111identifiers become invalid.
112 160
113=item $noderef = node_of $portid 161=item $nodeid = node_of $port
114 162
115Extracts and returns the noderef from a portid or a noderef. 163Extracts and returns the node ID from a port ID or a node ID.
164
165=item configure key => value...
166
167Before 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
169to know is its own name, and optionally it should know the addresses of
170some other nodes in the network to discover other nodes.
171
172This function configures a node - it must be called exactly once (or
173never) before calling other AnyEvent::MP functions.
174
175=over 4
176
177=item step 1, gathering configuration from profiles
178
179The function first looks up a profile in the aemp configuration (see the
180L<aemp> commandline utility). The profile name can be specified via the
181named C<profile> parameter. If it is missing, then the nodename (F<uname
182-n>) will be used as profile name.
183
184The profile data is then gathered as follows:
185
186First, all remaining key => value pairs (all of which are conviniently
187undocumented at the moment) will be interpreted as configuration
188data. Then they will be overwritten by any values specified in the global
189default configuration (see the F<aemp> utility), then the chain of
190profiles chosen by the profile name (and any C<parent> attributes).
191
192That means that the values specified in the profile have highest priority
193and the values specified directly via C<configure> have lowest priority,
194and can only be used to specify defaults.
195
196If the profile specifies a node ID, then this will become the node ID of
197this process. If not, then the profile name will be used as node ID. The
198special node ID of C<anon/> will be replaced by a random node ID.
199
200=item step 2, bind listener sockets
201
202The next step is to look up the binds in the profile, followed by binding
203aemp protocol listeners on all binds specified (it is possible and valid
204to have no binds, meaning that the node cannot be contacted form the
205outside. This means the node cannot talk to other nodes that also have no
206binds, but it can still talk to all "normal" nodes).
207
208If the profile does not specify a binds list, then a default of C<*> is
209used, meaning the node will bind on a dynamically-assigned port on every
210local IP address it finds.
211
212=item step 3, connect to seed nodes
213
214As the last step, the seeds list from the profile is passed to the
215L<AnyEvent::MP::Global> module, which will then use it to keep
216connectivity with at least one node at any point in time.
217
218=back
219
220Example: become a distributed node using the locla node name as profile.
221This should be the most common form of invocation for "daemon"-type nodes.
222
223 configure
224
225Example: become an anonymous node. This form is often used for commandline
226clients.
227
228 configure nodeid => "anon/";
229
230Example: configure a node using a profile called seed, which si suitable
231for a seed node as it binds on all local addresses on a fixed port (4040,
232customary 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)"
116 245
117=item $SELF 246=item $SELF
118 247
119Contains the current port id while executing C<rcv> callbacks or C<psub> 248Contains the current port id while executing C<rcv> callbacks or C<psub>
120blocks. 249blocks.
121 250
122=item SELF, %SELF, @SELF... 251=item *SELF, SELF, %SELF, @SELF...
123 252
124Due to some quirks in how perl exports variables, it is impossible to 253Due to some quirks in how perl exports variables, it is impossible to
125just export C<$SELF>, all the symbols called C<SELF> are exported by this 254just export C<$SELF>, all the symbols named C<SELF> are exported by this
126module, but only C<$SELF> is currently used. 255module, but only C<$SELF> is currently used.
127 256
128=item snd $portid, type => @data 257=item snd $port, type => @data
129 258
130=item snd $portid, @msg 259=item snd $port, @msg
131 260
132Send the given message to the given port ID, which can identify either 261Send the given message to the given port, which can identify either a
133a local or a remote port, and can be either a string or soemthignt hat 262local or a remote port, and must be a port ID.
134stringifies a sa port ID (such as a port object :).
135 263
136While the message can be about anything, it is highly recommended to use a 264While the message can be almost anything, it is highly recommended to
137string as first element (a portid, or some word that indicates a request 265use a string as first element (a port ID, or some word that indicates a
138type etc.). 266request type etc.) and to consist if only simple perl values (scalars,
267arrays, hashes) - if you think you need to pass an object, think again.
139 268
140The message data effectively becomes read-only after a call to this 269The message data logically becomes read-only after a call to this
141function: modifying any argument is not allowed and can cause many 270function: modifying any argument (or values referenced by them) is
142problems. 271forbidden, as there can be considerable time between the call to C<snd>
272and the time the message is actually being serialised - in fact, it might
273never be copied as within the same process it is simply handed to the
274receiving port.
143 275
144The type of data you can transfer depends on the transport protocol: when 276The type of data you can transfer depends on the transport protocol: when
145JSON is used, then only strings, numbers and arrays and hashes consisting 277JSON is used, then only strings, numbers and arrays and hashes consisting
146of those are allowed (no objects). When Storable is used, then anything 278of those are allowed (no objects). When Storable is used, then anything
147that Storable can serialise and deserialise is allowed, and for the local 279that Storable can serialise and deserialise is allowed, and for the local
148node, anything can be passed. 280node, anything can be passed. Best rely only on the common denominator of
281these.
149 282
150=item kil $portid[, @reason] 283=item $local_port = port
151 284
152Kill the specified port with the given C<@reason>. 285Create a new local port object and returns its port ID. Initially it has
286no callbacks set and will throw an error when it receives messages.
153 287
154If no C<@reason> is specified, then the port is killed "normally" (linked 288=item $local_port = port { my @msg = @_ }
155ports will not be kileld, or even notified).
156 289
157Otherwise, linked ports get killed with the same reason (second form of 290Creates a new local port, and returns its ID. Semantically the same as
158C<mon>, see below). 291creating a port and calling C<rcv $port, $callback> on it.
159 292
160Runtime errors while evaluating C<rcv> callbacks or inside C<psub> blocks 293The block will be called for every message received on the port, with the
161will be reported as reason C<< die => $@ >>. 294global variable C<$SELF> set to the port ID. Runtime errors will cause the
295port 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.
162 297
163Transport/communication errors are reported as C<< transport_error => 298If you want to stop/destroy the port, simply C<kil> it:
164$message >>.
165 299
166=item $guard = mon $portid, $cb->(@reason) 300 my $port = port {
301 my @msg = @_;
302 ...
303 kil $SELF;
304 };
167 305
168=item $guard = mon $portid, $otherport 306=cut
169 307
170=item $guard = mon $portid, $otherport, @msg 308sub rcv($@);
171 309
310sub _kilme {
311 die "received message on port without callback";
312}
313
314sub port(;&) {
315 my $id = "$UNIQ." . $ID++;
316 my $port = "$NODE#$id";
317
318 rcv $port, shift || \&_kilme;
319
320 $port
321}
322
323=item rcv $local_port, $callback->(@msg)
324
325Replaces the default callback on the specified port. There is no way to
326remove the default callback: use C<sub { }> to disable it, or better
327C<kil> the port when it is no longer needed.
328
329The global C<$SELF> (exported by this module) contains C<$port> while
330executing the callback. Runtime errors during callback execution will
331result in the port being C<kil>ed.
332
333The default callback received all messages not matched by a more specific
334C<tag> match.
335
336=item rcv $local_port, tag => $callback->(@msg_without_tag), ...
337
338Register (or replace) callbacks to be called on messages starting with the
339given tag on the given port (and return the port), or unregister it (when
340C<$callback> is C<$undef> or missing). There can only be one callback
341registered for each tag.
342
343The original message will be passed to the callback, after the first
344element (the tag) has been removed. The callback will use the same
345environment as the default callback (see above).
346
347Example: create a port and bind receivers on it in one go.
348
349 my $port = rcv port,
350 msg1 => sub { ... },
351 msg2 => sub { ... },
352 ;
353
354Example: create a port, bind receivers and send it in a message elsewhere
355in one go:
356
357 snd $otherport, reply =>
358 rcv port,
359 msg1 => sub { ... },
360 ...
361 ;
362
363Example: 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, $otherport;
370 };
371
372=cut
373
374sub rcv($@) {
375 my $port = shift;
376 my ($nodeid, $portid) = split /#/, $port, 2;
377
378 $NODE{$nodeid} == $NODE{""}
379 or Carp::croak "$port: rcv can only be called on local ports, caught";
380
381 while (@_) {
382 if (ref $_[0]) {
383 if (my $self = $PORT_DATA{$portid}) {
384 "AnyEvent::MP::Port" eq ref $self
385 or Carp::croak "$port: rcv can only be called on message matching ports, caught";
386
387 $self->[2] = shift;
388 } else {
389 my $cb = shift;
390 $PORT{$portid} = sub {
391 local $SELF = $port;
392 eval { &$cb }; _self_die if $@;
393 };
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
411 };
412
413 "AnyEvent::MP::Port" eq ref $self
414 or Carp::croak "$port: rcv can only be called on message matching ports, caught";
415
416 my ($tag, $cb) = splice @_, 0, 2;
417
418 if (defined $cb) {
419 $self->[1]{$tag} = $cb;
420 } else {
421 delete $self->[1]{$tag};
422 }
423 }
424 }
425
426 $port
427}
428
429=item $closure = psub { BLOCK }
430
431Remembers C<$SELF> and creates a closure out of the BLOCK. When the
432closure is executed, sets up the environment in the same way as in C<rcv>
433callbacks, i.e. runtime errors will cause the port to get C<kil>ed.
434
435This is useful when you register callbacks from C<rcv> callbacks:
436
437 rcv delayed_reply => sub {
438 my ($delay, @reply) = @_;
439 my $timer = AE::timer $delay, 0, psub {
440 snd @reply, $SELF;
441 };
442 };
443
444=cut
445
446sub 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 }
464 }
465}
466
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 $port dies
474
172Monitor the given port and do something when the port is killed. 475Monitor the given port and do something when the port is killed or
476messages to it were lost, and optionally return a guard that can be used
477to stop monitoring again.
173 478
479C<mon> effectively guarantees that, in the absence of hardware failures,
480after starting the monitor, either all messages sent to the port will
481arrive, or the monitoring action will be invoked after possible message
482loss has been detected. No messages will be lost "in between" (after
483the first lost message no further messages will be received by the
484port). After the monitoring action was invoked, further messages might get
485delivered again.
486
487Note that monitoring-actions are one-shot: once messages are lost (and a
488monitoring alert was raised), they are removed and will not trigger again.
489
174In the first form, the callback is simply called with any number 490In the first form (callback), the callback is simply called with any
175of C<@reason> elements (no @reason means that the port was deleted 491number of C<@reason> elements (no @reason means that the port was deleted
176"normally"). Note also that I<< the callback B<must> never die >>, so use 492"normally"). Note also that I<< the callback B<must> never die >>, so use
177C<eval> if unsure. 493C<eval> if unsure.
178 494
179In the second form, the other port will be C<kil>'ed with C<@reason>, iff 495In the second form (another port given), the other port (C<$rcvport>)
180a @reason was specified, i.e. on "normal" kils nothing happens, while 496will be C<kil>'ed with C<@reason>, iff a @reason was specified, i.e. on
181under all other conditions, the other port is killed with the same reason. 497"normal" kils nothing happens, while under all other conditions, the other
498port is killed with the same reason.
182 499
500The third form (kill self) is the same as the second form, except that
501C<$rvport> defaults to C<$SELF>.
502
183In the last form, a message of the form C<@msg, @reason> will be C<snd>. 503In the last form (message), a message of the form C<@msg, @reason> will be
504C<snd>.
505
506As a rule of thumb, monitoring requests should always monitor a port from
507a local port (or callback). The reason is that kill messages might get
508lost, just like any other message. Another less obvious reason is that
509even monitoring requests can get lost (for exmaple, when the connection
510to the other node goes down permanently). When monitoring a port locally
511these problems do not exist.
184 512
185Example: call a given callback when C<$port> is killed. 513Example: call a given callback when C<$port> is killed.
186 514
187 mon $port, sub { warn "port died because of <@_>\n" }; 515 mon $port, sub { warn "port died because of <@_>\n" };
188 516
189Example: kill ourselves when C<$port> is killed abnormally. 517Example: kill ourselves when C<$port> is killed abnormally.
190 518
191 mon $port, $self; 519 mon $port;
192 520
193Example: send us a restart message another C<$port> is killed. 521Example: send us a restart message when another C<$port> is killed.
194 522
195 mon $port, $self => "restart"; 523 mon $port, $self => "restart";
196 524
197=cut 525=cut
198 526
199sub mon { 527sub mon {
200 my ($noderef, $port, $cb) = ((split /#/, shift, 2), shift); 528 my ($nodeid, $port) = split /#/, shift, 2;
201 529
202 my $node = $NODE{$noderef} || add_node $noderef; 530 my $node = $NODE{$nodeid} || add_node $nodeid;
203 531
204 #TODO: ports must not be references 532 my $cb = @_ ? shift : $SELF || Carp::croak 'mon: called with one argument only, but $SELF not set,';
205 if (!ref $cb or "AnyEvent::MP::Port" eq ref $cb) { 533
534 unless (ref $cb) {
206 if (@_) { 535 if (@_) {
207 # send a kill info message 536 # send a kill info message
208 my (@msg) = ($cb, @_); 537 my (@msg) = ($cb, @_);
209 $cb = sub { snd @msg, @_ }; 538 $cb = sub { snd @msg, @_ };
210 } else { 539 } else {
226is killed, the references will be freed. 555is killed, the references will be freed.
227 556
228Optionally returns a guard that will stop the monitoring. 557Optionally returns a guard that will stop the monitoring.
229 558
230This function is useful when you create e.g. timers or other watchers and 559This function is useful when you create e.g. timers or other watchers and
231want to free them when the port gets killed: 560want to free them when the port gets killed (note the use of C<psub>):
232 561
233 $port->rcv (start => sub { 562 $port->rcv (start => sub {
234 my $timer; $timer = mon_guard $port, AE::timer 1, 1, sub { 563 my $timer; $timer = mon_guard $port, AE::timer 1, 1, psub {
235 undef $timer if 0.9 < rand; 564 undef $timer if 0.9 < rand;
236 }); 565 });
237 }); 566 });
238 567
239=cut 568=cut
240 569
241sub mon_guard { 570sub mon_guard {
242 my ($port, @refs) = @_; 571 my ($port, @refs) = @_;
243 572
573 #TODO: mon-less form?
574
244 mon $port, sub { 0 && @refs } 575 mon $port, sub { 0 && @refs }
245} 576}
246 577
247=item $local_port = port 578=item kil $port[, @reason]
248 579
249Create a new local port object that supports message matching. 580Kill the specified port with the given C<@reason>.
250 581
251=item $portid = port { my @msg = @_; $finished } 582If no C<@reason> is specified, then the port is killed "normally" (ports
583monitoring other ports will not necessarily die because a port dies
584"normally").
252 585
253Creates a "mini port", that is, a very lightweight port without any 586Otherwise, linked ports get killed with the same reason (second form of
254pattern matching behind it, and returns its ID. 587C<mon>, see above).
255 588
256The block will be called for every message received on the port. When the 589Runtime errors while evaluating C<rcv> callbacks or inside C<psub> blocks
257callback returns a true value its job is considered "done" and the port 590will be reported as reason C<< die => $@ >>.
258will be destroyed. Otherwise it will stay alive.
259 591
260The message will be passed as-is, no extra argument (i.e. no port id) will 592Transport/communication errors are reported as C<< transport_error =>
261be passed to the callback. 593$message >>.
262 594
263If you need the local port id in the callback, this works nicely:
264
265 my $port; $port = miniport {
266 snd $otherport, reply => $port;
267 };
268
269=cut 595=cut
270 596
271sub port(;&) { 597=item $port = spawn $node, $initfunc[, @initdata]
272 my $id = "$UNIQ." . $ID++;
273 my $port = "$NODE#$id";
274 598
275 if (@_) { 599Creates a port on the node C<$node> (which can also be a port ID, in which
276 my $cb = shift; 600case it's the node where that port resides).
277 $PORT{$id} = sub { 601
278 local $SELF = $port; 602The port ID of the newly created port is returned immediately, and it is
279 eval { 603possible to immediately start sending messages or to monitor the port.
280 &$cb 604
281 and kil $id; 605After the port has been created, the init function is called on the remote
606node, in the same context as a C<rcv> callback. This function must be a
607fully-qualified function name (e.g. C<MyApp::Chat::Server::init>). To
608specify a function in the main program, use C<::name>.
609
610If the function doesn't exist, then the node tries to C<require>
611the package, then the package above the package and so on (e.g.
612C<MyApp::Chat::Server>, C<MyApp::Chat>, C<MyApp>) until the function
613exists or it runs out of package names.
614
615The init function is then called with the newly-created port as context
616object (C<$SELF>) and the C<@initdata> values as arguments.
617
618A common idiom is to pass a local port, immediately monitor the spawned
619port, and in the remote init function, immediately monitor the passed
620local port. This two-way monitoring ensures that both ports get cleaned up
621when there is a problem.
622
623Example: 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 {
282 }; 636 ...
283 _self_die if $@;
284 };
285 } else {
286 my $self = bless {
287 id => "$NODE#$id",
288 }, "AnyEvent::MP::Port";
289
290 $PORT_DATA{$id} = $self;
291 $PORT{$id} = sub {
292 local $SELF = $port;
293
294 eval {
295 for (@{ $self->{rc0}{$_[0]} }) {
296 $_ && &{$_->[0]}
297 && undef $_;
298 }
299
300 for (@{ $self->{rcv}{$_[0]} }) {
301 $_ && [@_[1 .. @{$_->[1]}]] ~~ $_->[1]
302 && &{$_->[0]}
303 && undef $_;
304 }
305
306 for (@{ $self->{any} }) {
307 $_ && [@_[0 .. $#{$_->[1]}]] ~~ $_->[1]
308 && &{$_->[0]}
309 && undef $_;
310 }
311 };
312 _self_die if $@;
313 }; 637 };
314 } 638 }
315 639
316 $port
317}
318
319=item reg $portid, $name
320
321Registers the given port under the name C<$name>. If the name already
322exists it is replaced.
323
324A port can only be registered under one well known name.
325
326A port automatically becomes unregistered when it is killed.
327
328=cut 640=cut
329 641
330sub reg(@) { 642sub _spawn {
331 my ($portid, $name) = @_; 643 my $port = shift;
644 my $init = shift;
332 645
333 $REG{$name} = $portid; 646 local $SELF = "$NODE#$port";
334} 647 eval {
335 648 &{ load_func $init }
336=item rcv $portid, tagstring => $callback->(@msg), ...
337
338=item rcv $portid, $smartmatch => $callback->(@msg), ...
339
340=item rcv $portid, [$smartmatch...] => $callback->(@msg), ...
341
342Register callbacks to be called on matching messages on the given port.
343
344The callback has to return a true value when its work is done, after
345which is will be removed, or a false value in which case it will stay
346registered.
347
348The global C<$SELF> (exported by this module) contains C<$portid> while
349executing the callback.
350
351Runtime errors wdurign callback execution will result in the port being
352C<kil>ed.
353
354If the match is an array reference, then it will be matched against the
355first elements of the message, otherwise only the first element is being
356matched.
357
358Any element in the match that is specified as C<_any_> (a function
359exported by this module) matches any single element of the message.
360
361While not required, it is highly recommended that the first matching
362element is a string identifying the message. The one-string-only match is
363also the most efficient match (by far).
364
365=cut
366
367sub rcv($@) {
368 my ($noderef, $port) = split /#/, shift, 2;
369
370 ($NODE{$noderef} || add_node $noderef) == $NODE{""}
371 or Carp::croak "$noderef#$port: rcv can only be called on local ports, caught";
372
373 my $self = $PORT_DATA{$port}
374 or Carp::croak "$noderef#$port: rcv can only be called on message matching ports, caught";
375
376 "AnyEvent::MP::Port" eq ref $self
377 or Carp::croak "$noderef#$port: rcv can only be called on message matching ports, caught";
378
379 while (@_) {
380 my ($match, $cb) = splice @_, 0, 2;
381
382 if (!ref $match) {
383 push @{ $self->{rc0}{$match} }, [$cb];
384 } elsif (("ARRAY" eq ref $match && !ref $match->[0])) {
385 my ($type, @match) = @$match;
386 @match
387 ? push @{ $self->{rcv}{$match->[0]} }, [$cb, \@match]
388 : push @{ $self->{rc0}{$match->[0]} }, [$cb];
389 } else {
390 push @{ $self->{any} }, [$cb, $match];
391 }
392 }
393}
394
395=item $closure = psub { BLOCK }
396
397Remembers C<$SELF> and creates a closure out of the BLOCK. When the
398closure is executed, sets up the environment in the same way as in C<rcv>
399callbacks, i.e. runtime errors will cause the port to get C<kil>ed.
400
401This is useful when you register callbacks from C<rcv> callbacks:
402
403 rcv delayed_reply => sub {
404 my ($delay, @reply) = @_;
405 my $timer = AE::timer $delay, 0, psub {
406 snd @reply, $SELF;
407 };
408 }; 649 };
409
410=cut
411
412sub psub(&) {
413 my $cb = shift;
414
415 my $port = $SELF
416 or Carp::croak "psub can only be called from within rcv or psub callbacks, not";
417
418 sub {
419 local $SELF = $port;
420
421 if (wantarray) {
422 my @res = eval { &$cb };
423 _self_die if $@; 650 _self_die if $@;
424 @res 651}
425 } else { 652
426 my $res = eval { &$cb }; 653sub spawn(@) {
427 _self_die if $@; 654 my ($nodeid, undef) = split /#/, shift, 2;
428 $res 655
429 } 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, $callback
669
670Either sends the given message, or call the given callback, after the
671specified number of seconds.
672
673This is simply a utility function that comes in handy at times - the
674AnyEvent::MP author is not convinced of the wisdom of having it, though,
675so it may go away in the future.
676
677=cut
678
679sub after($@) {
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;
430 } 687 };
431} 688}
432 689
433=back 690=back
434 691
435=head1 FUNCTIONS FOR NODES 692=head1 AnyEvent::MP vs. Distributed Erlang
693
694AnyEvent::MP got lots of its ideas from distributed Erlang (Erlang node
695== aemp node, Erlang process == aemp port), so many of the documents and
696programming techniques employed by Erlang apply to AnyEvent::MP. Here is a
697sample:
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
704Despite the similarities, there are also some important differences:
436 705
437=over 4 706=over 4
438 707
439=item become_public endpoint... 708=item * Node IDs are arbitrary strings in AEMP.
440 709
441Tells the node to become a public node, i.e. reachable from other nodes. 710Erlang relies on special naming and DNS to work everywhere in the same
711way. AEMP relies on each node somehow knowing its own address(es) (e.g. by
712configuraiton or DNS), but will otherwise discover other odes itself.
442 713
443If no arguments are given, or the first argument is C<undef>, then 714=item * Erlang has a "remote ports are like local ports" philosophy, AEMP
444AnyEvent::MP tries to bind on port C<4040> on all IP addresses that the 715uses "local ports are like remote ports".
445local nodename resolves to.
446 716
447Otherwise the first argument must be an array-reference with transport 717The failure modes for local ports are quite different (runtime errors
448endpoints ("ip:port", "hostname:port") or port numbers (in which case the 718only) then for remote ports - when a local port dies, you I<know> it dies,
449local nodename is used as hostname). The endpoints are all resolved and 719when a connection to another node dies, you know nothing about the other
450will become the node reference. 720port.
451 721
452=cut 722Erlang pretends remote ports are as reliable as local ports, even when
723they are not.
724
725AEMP encourages a "treat remote ports differently" philosophy, with local
726ports being the special case/exception, where transport errors cannot
727occur.
728
729=item * Erlang uses processes and a mailbox, AEMP does not queue.
730
731Erlang uses processes that selectively receive messages, and therefore
732needs a queue. AEMP is event based, queuing messages would serve no
733useful purpose. For the same reason the pattern-matching abilities of
734AnyEvent::MP are more limited, as there is little need to be able to
735filter messages without dequeing them.
736
737(But see L<Coro::MP> for a more Erlang-like process model on top of AEMP).
738
739=item * Erlang sends are synchronous, AEMP sends are asynchronous.
740
741Sending messages in Erlang is synchronous and blocks the process (and
742so does not need a queue that can overflow). AEMP sends are immediate,
743connection establishment is handled in the background.
744
745=item * Erlang suffers from silent message loss, AEMP does not.
746
747Erlang makes few guarantees on messages delivery - messages can get lost
748without any of the processes realising it (i.e. you send messages a, b,
749and c, and the other side only receives messages a and c).
750
751AEMP guarantees correct ordering, and the guarantee that after one message
752is lost, all following ones sent to the same port are lost as well, until
753monitoring raises an error, so there are no silent "holes" in the message
754sequence.
755
756=item * Erlang can send messages to the wrong port, AEMP does not.
757
758In Erlang it is quite likely that a node that restarts reuses a process ID
759known to other nodes for a completely different process, causing messages
760destined for that process to end up in an unrelated process.
761
762AEMP never reuses port IDs, so old messages or old port IDs floating
763around in the network will not be sent to an unrelated port.
764
765=item * Erlang uses unprotected connections, AEMP uses secure
766authentication and can use TLS.
767
768AEMP can use a proven protocol - TLS - to protect connections and
769securely authenticate nodes.
770
771=item * The AEMP protocol is optimised for both text-based and binary
772communications.
773
774The AEMP protocol, unlike the Erlang protocol, supports both programming
775language independent text-only protocols (good for debugging) and binary,
776language-specific serialisers (e.g. Storable). By default, unless TLS is
777used, the protocol is actually completely text-based.
778
779It has also been carefully designed to be implementable in other languages
780with a minimum of work while gracefully degrading functionality to make the
781protocol simple.
782
783=item * AEMP has more flexible monitoring options than Erlang.
784
785In Erlang, you can chose to receive I<all> exit signals as messages
786or I<none>, there is no in-between, so monitoring single processes is
787difficult to implement. Monitoring in AEMP is more flexible than in
788Erlang, as one can choose between automatic kill, exit message or callback
789on a per-process basis.
790
791=item * Erlang tries to hide remote/local connections, AEMP does not.
792
793Monitoring in Erlang is not an indicator of process death/crashes, in the
794same way as linking is (except linking is unreliable in Erlang).
795
796In AEMP, you don't "look up" registered port names or send to named ports
797that might or might not be persistent. Instead, you normally spawn a port
798on the remote node. The init function monitors you, and you monitor the
799remote port. Since both monitors are local to the node, they are much more
800reliable (no need for C<spawn_link>).
801
802This also saves round-trips and avoids sending messages to the wrong port
803(hard to do in Erlang).
453 804
454=back 805=back
455 806
456=head1 NODE MESSAGES 807=head1 RATIONALE
457
458Nodes understand the following messages sent to them. Many of them take
459arguments called C<@reply>, which will simply be used to compose a reply
460message - C<$reply[0]> is the port to reply to, C<$reply[1]> the type and
461the remaining arguments are simply the message data.
462 808
463=over 4 809=over 4
464 810
465=cut 811=item Why strings for port and node IDs, why not objects?
466 812
467=item lookup => $name, @reply 813We considered "objects", but found that the actual number of methods
814that can be called are quite low. Since port and node IDs travel over
815the network frequently, the serialising/deserialising would add lots of
816overhead, as well as having to keep a proxy object everywhere.
468 817
469Replies with the port ID of the specified well-known port, or C<undef>. 818Strings can easily be printed, easily serialised etc. and need no special
819procedures to be "valid".
470 820
471=item devnull => ... 821And as a result, a miniport consists of a single closure stored in a
822global hash - it can't become much cheaper.
472 823
473Generic data sink/CPU heat conversion. 824=item Why favour JSON, why not a real serialising format such as Storable?
474 825
475=item relay => $port, @msg 826In fact, any AnyEvent::MP node will happily accept Storable as framing
827format, but currently there is no way to make a node use Storable by
828default (although all nodes will accept it).
476 829
477Simply forwards the message to the given port. 830The default framing protocol is JSON because a) JSON::XS is many times
831faster for small messages and b) most importantly, after years of
832experience we found that object serialisation is causing more problems
833than it solves: Just like function calls, objects simply do not travel
834easily over the network, mostly because they will always be a copy, so you
835always have to re-think your design.
478 836
479=item eval => $string[ @reply] 837Keeping your messages simple, concentrating on data structures rather than
480 838objects, will keep your messages clean, tidy and efficient.
481Evaluates the given string. If C<@reply> is given, then a message of the
482form C<@reply, $@, @evalres> is sent.
483
484Example: crash another node.
485
486 snd $othernode, eval => "exit";
487
488=item time => @reply
489
490Replies the the current node time to C<@reply>.
491
492Example: tell the current node to send the current time to C<$myport> in a
493C<timereply> message.
494
495 snd $NODE, time => $myport, timereply => 1, 2;
496 # => snd $myport, timereply => 1, 2, <time>
497 839
498=back 840=back
499 841
500=head1 SEE ALSO 842=head1 SEE ALSO
843
844L<AnyEvent::MP::Intro> - a gentle introduction.
845
846L<AnyEvent::MP::Kernel> - more, lower-level, stuff.
847
848L<AnyEvent::MP::Global> - network maintainance and port groups, to find
849your applications.
501 850
502L<AnyEvent>. 851L<AnyEvent>.
503 852
504=head1 AUTHOR 853=head1 AUTHOR
505 854

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines