1 |
=head1 NAME |
2 |
|
3 |
AnyEvent::MP - multi-processing/message-passing framework |
4 |
|
5 |
=head1 SYNOPSIS |
6 |
|
7 |
use AnyEvent::MP; |
8 |
|
9 |
$NODE # contains this node's noderef |
10 |
NODE # returns this node's noderef |
11 |
NODE $port # returns the noderef of the port |
12 |
|
13 |
$SELF # receiving/own port id in rcv callbacks |
14 |
|
15 |
# ports are message endpoints |
16 |
|
17 |
# sending messages |
18 |
snd $port, type => data...; |
19 |
snd $port, @msg; |
20 |
snd @msg_with_first_element_being_a_port; |
21 |
|
22 |
# miniports |
23 |
my $miniport = port { my @msg = @_; 0 }; |
24 |
|
25 |
# full ports |
26 |
my $port = port; |
27 |
rcv $port, smartmatch => $cb->(@msg); |
28 |
rcv $port, ping => sub { snd $_[0], "pong"; 0 }; |
29 |
rcv $port, pong => sub { warn "pong received\n"; 0 }; |
30 |
|
31 |
# remote ports |
32 |
my $port = spawn $node, $initfunc, @initdata; |
33 |
|
34 |
# more, smarter, matches (_any_ is exported by this module) |
35 |
rcv $port, [child_died => $pid] => sub { ... |
36 |
rcv $port, [_any_, _any_, 3] => sub { .. $_[2] is 3 |
37 |
|
38 |
# monitoring |
39 |
mon $port, $cb->(@msg) # callback is invoked on death |
40 |
mon $port, $otherport # kill otherport on abnormal death |
41 |
mon $port, $otherport, @msg # send message on death |
42 |
|
43 |
=head1 DESCRIPTION |
44 |
|
45 |
This module (-family) implements a simple message passing framework. |
46 |
|
47 |
Despite its simplicity, you can securely message other processes running |
48 |
on the same or other hosts. |
49 |
|
50 |
For an introduction to this module family, see the L<AnyEvent::MP::Intro> |
51 |
manual page. |
52 |
|
53 |
At the moment, this module family is severly broken and underdocumented, |
54 |
so do not use. This was uploaded mainly to reserve the CPAN namespace - |
55 |
stay tuned! The basic API should be finished, however. |
56 |
|
57 |
=head1 CONCEPTS |
58 |
|
59 |
=over 4 |
60 |
|
61 |
=item port |
62 |
|
63 |
A port is something you can send messages to (with the C<snd> function). |
64 |
|
65 |
Some ports allow you to register C<rcv> handlers that can match specific |
66 |
messages. All C<rcv> handlers will receive messages they match, messages |
67 |
will not be queued. |
68 |
|
69 |
=item port id - C<noderef#portname> |
70 |
|
71 |
A port id is normaly the concatenation of a noderef, a hash-mark (C<#>) as |
72 |
separator, and a port name (a printable string of unspecified format). An |
73 |
exception is the the node port, whose ID is identical to its node |
74 |
reference. |
75 |
|
76 |
=item node |
77 |
|
78 |
A node is a single process containing at least one port - the node |
79 |
port. You can send messages to node ports to find existing ports or to |
80 |
create new ports, among other things. |
81 |
|
82 |
Nodes are either private (single-process only), slaves (connected to a |
83 |
master node only) or public nodes (connectable from unrelated nodes). |
84 |
|
85 |
=item noderef - C<host:port,host:port...>, C<id@noderef>, C<id> |
86 |
|
87 |
A node reference is a string that either simply identifies the node (for |
88 |
private and slave nodes), or contains a recipe on how to reach a given |
89 |
node (for public nodes). |
90 |
|
91 |
This recipe is simply a comma-separated list of C<address:port> pairs (for |
92 |
TCP/IP, other protocols might look different). |
93 |
|
94 |
Node references come in two flavours: resolved (containing only numerical |
95 |
addresses) or unresolved (where hostnames are used instead of addresses). |
96 |
|
97 |
Before using an unresolved node reference in a message you first have to |
98 |
resolve it. |
99 |
|
100 |
=back |
101 |
|
102 |
=head1 VARIABLES/FUNCTIONS |
103 |
|
104 |
=over 4 |
105 |
|
106 |
=cut |
107 |
|
108 |
package AnyEvent::MP; |
109 |
|
110 |
use AnyEvent::MP::Base; |
111 |
|
112 |
use common::sense; |
113 |
|
114 |
use Carp (); |
115 |
|
116 |
use AE (); |
117 |
|
118 |
use base "Exporter"; |
119 |
|
120 |
our $VERSION = $AnyEvent::MP::Base::VERSION; |
121 |
|
122 |
our @EXPORT = qw( |
123 |
NODE $NODE *SELF node_of _any_ |
124 |
resolve_node initialise_node |
125 |
snd rcv mon kil reg psub spawn |
126 |
port |
127 |
); |
128 |
|
129 |
our $SELF; |
130 |
|
131 |
sub _self_die() { |
132 |
my $msg = $@; |
133 |
$msg =~ s/\n+$// unless ref $msg; |
134 |
kil $SELF, die => $msg; |
135 |
} |
136 |
|
137 |
=item $thisnode = NODE / $NODE |
138 |
|
139 |
The C<NODE> function returns, and the C<$NODE> variable contains |
140 |
the noderef of the local node. The value is initialised by a call |
141 |
to C<become_public> or C<become_slave>, after which all local port |
142 |
identifiers become invalid. |
143 |
|
144 |
=item $noderef = node_of $port |
145 |
|
146 |
Extracts and returns the noderef from a portid or a noderef. |
147 |
|
148 |
=item initialise_node $noderef, $seednode, $seednode... |
149 |
|
150 |
=item initialise_node "slave/", $master, $master... |
151 |
|
152 |
Before a node can talk to other nodes on the network it has to initialise |
153 |
itself - the minimum a node needs to know is it's own name, and optionally |
154 |
it should know the noderefs of some other nodes in the network. |
155 |
|
156 |
This function initialises a node - it must be called exactly once (or |
157 |
never) before calling other AnyEvent::MP functions. |
158 |
|
159 |
All arguments are noderefs, which can be either resolved or unresolved. |
160 |
|
161 |
There are two types of networked nodes, public nodes and slave nodes: |
162 |
|
163 |
=over 4 |
164 |
|
165 |
=item public nodes |
166 |
|
167 |
For public nodes, C<$noderef> must either be a (possibly unresolved) |
168 |
noderef, in which case it will be resolved, or C<undef> (or missing), in |
169 |
which case the noderef will be guessed. |
170 |
|
171 |
Afterwards, the node will bind itself on all endpoints and try to connect |
172 |
to all additional C<$seednodes> that are specified. Seednodes are optional |
173 |
and can be used to quickly bootstrap the node into an existing network. |
174 |
|
175 |
=item slave nodes |
176 |
|
177 |
When the C<$noderef> is the special string C<slave/>, then the node will |
178 |
become a slave node. Slave nodes cannot be contacted from outside and will |
179 |
route most of their traffic to the master node that they attach to. |
180 |
|
181 |
At least one additional noderef is required: The node will try to connect |
182 |
to all of them and will become a slave attached to the first node it can |
183 |
successfully connect to. |
184 |
|
185 |
=back |
186 |
|
187 |
This function will block until all nodes have been resolved and, for slave |
188 |
nodes, until it has successfully established a connection to a master |
189 |
server. |
190 |
|
191 |
Example: become a public node listening on the default node. |
192 |
|
193 |
initialise_node; |
194 |
|
195 |
Example: become a public node, and try to contact some well-known master |
196 |
servers to become part of the network. |
197 |
|
198 |
initialise_node undef, "master1", "master2"; |
199 |
|
200 |
Example: become a public node listening on port C<4041>. |
201 |
|
202 |
initialise_node 4041; |
203 |
|
204 |
Example: become a public node, only visible on localhost port 4044. |
205 |
|
206 |
initialise_node "locahost:4044"; |
207 |
|
208 |
Example: become a slave node to any of the specified master servers. |
209 |
|
210 |
initialise_node "slave/", "master1", "192.168.13.17", "mp.example.net"; |
211 |
|
212 |
=item $cv = resolve_node $noderef |
213 |
|
214 |
Takes an unresolved node reference that may contain hostnames and |
215 |
abbreviated IDs, resolves all of them and returns a resolved node |
216 |
reference. |
217 |
|
218 |
In addition to C<address:port> pairs allowed in resolved noderefs, the |
219 |
following forms are supported: |
220 |
|
221 |
=over 4 |
222 |
|
223 |
=item the empty string |
224 |
|
225 |
An empty-string component gets resolved as if the default port (4040) was |
226 |
specified. |
227 |
|
228 |
=item naked port numbers (e.g. C<1234>) |
229 |
|
230 |
These are resolved by prepending the local nodename and a colon, to be |
231 |
further resolved. |
232 |
|
233 |
=item hostnames (e.g. C<localhost:1234>, C<localhost>) |
234 |
|
235 |
These are resolved by using AnyEvent::DNS to resolve them, optionally |
236 |
looking up SRV records for the C<aemp=4040> port, if no port was |
237 |
specified. |
238 |
|
239 |
=back |
240 |
|
241 |
=item $SELF |
242 |
|
243 |
Contains the current port id while executing C<rcv> callbacks or C<psub> |
244 |
blocks. |
245 |
|
246 |
=item SELF, %SELF, @SELF... |
247 |
|
248 |
Due to some quirks in how perl exports variables, it is impossible to |
249 |
just export C<$SELF>, all the symbols called C<SELF> are exported by this |
250 |
module, but only C<$SELF> is currently used. |
251 |
|
252 |
=item snd $port, type => @data |
253 |
|
254 |
=item snd $port, @msg |
255 |
|
256 |
Send the given message to the given port ID, which can identify either |
257 |
a local or a remote port, and can be either a string or soemthignt hat |
258 |
stringifies a sa port ID (such as a port object :). |
259 |
|
260 |
While the message can be about anything, it is highly recommended to use a |
261 |
string as first element (a portid, or some word that indicates a request |
262 |
type etc.). |
263 |
|
264 |
The message data effectively becomes read-only after a call to this |
265 |
function: modifying any argument is not allowed and can cause many |
266 |
problems. |
267 |
|
268 |
The type of data you can transfer depends on the transport protocol: when |
269 |
JSON is used, then only strings, numbers and arrays and hashes consisting |
270 |
of those are allowed (no objects). When Storable is used, then anything |
271 |
that Storable can serialise and deserialise is allowed, and for the local |
272 |
node, anything can be passed. |
273 |
|
274 |
=item $local_port = port |
275 |
|
276 |
Create a new local port object that can be used either as a pattern |
277 |
matching port ("full port") or a single-callback port ("miniport"), |
278 |
depending on how C<rcv> callbacks are bound to the object. |
279 |
|
280 |
=item $port = port { my @msg = @_; $finished } |
281 |
|
282 |
Creates a "miniport", that is, a very lightweight port without any pattern |
283 |
matching behind it, and returns its ID. Semantically the same as creating |
284 |
a port and calling C<rcv $port, $callback> on it. |
285 |
|
286 |
The block will be called for every message received on the port. When the |
287 |
callback returns a true value its job is considered "done" and the port |
288 |
will be destroyed. Otherwise it will stay alive. |
289 |
|
290 |
The message will be passed as-is, no extra argument (i.e. no port id) will |
291 |
be passed to the callback. |
292 |
|
293 |
If you need the local port id in the callback, this works nicely: |
294 |
|
295 |
my $port; $port = port { |
296 |
snd $otherport, reply => $port; |
297 |
}; |
298 |
|
299 |
=cut |
300 |
|
301 |
sub rcv($@); |
302 |
|
303 |
sub port(;&) { |
304 |
my $id = "$UNIQ." . $ID++; |
305 |
my $port = "$NODE#$id"; |
306 |
|
307 |
if (@_) { |
308 |
rcv $port, shift; |
309 |
} else { |
310 |
$PORT{$id} = sub { }; # nop |
311 |
} |
312 |
|
313 |
$port |
314 |
} |
315 |
|
316 |
=item reg $port, $name |
317 |
|
318 |
=item reg $name |
319 |
|
320 |
Registers the given port (or C<$SELF><<< if missing) under the name |
321 |
C<$name>. If the name already exists it is replaced. |
322 |
|
323 |
A port can only be registered under one well known name. |
324 |
|
325 |
A port automatically becomes unregistered when it is killed. |
326 |
|
327 |
=cut |
328 |
|
329 |
sub reg(@) { |
330 |
my $port = @_ > 1 ? shift : $SELF || Carp::croak 'reg: called with one argument only, but $SELF not set,'; |
331 |
|
332 |
$REG{$_[0]} = $port; |
333 |
} |
334 |
|
335 |
=item rcv $port, $callback->(@msg) |
336 |
|
337 |
Replaces the callback on the specified miniport (after converting it to |
338 |
one if required). |
339 |
|
340 |
=item rcv $port, tagstring => $callback->(@msg), ... |
341 |
|
342 |
=item rcv $port, $smartmatch => $callback->(@msg), ... |
343 |
|
344 |
=item rcv $port, [$smartmatch...] => $callback->(@msg), ... |
345 |
|
346 |
Register callbacks to be called on matching messages on the given full |
347 |
port (after converting it to one if required) and return the port. |
348 |
|
349 |
The callback has to return a true value when its work is done, after |
350 |
which is will be removed, or a false value in which case it will stay |
351 |
registered. |
352 |
|
353 |
The global C<$SELF> (exported by this module) contains C<$port> while |
354 |
executing the callback. |
355 |
|
356 |
Runtime errors during callback execution will result in the port being |
357 |
C<kil>ed. |
358 |
|
359 |
If the match is an array reference, then it will be matched against the |
360 |
first elements of the message, otherwise only the first element is being |
361 |
matched. |
362 |
|
363 |
Any element in the match that is specified as C<_any_> (a function |
364 |
exported by this module) matches any single element of the message. |
365 |
|
366 |
While not required, it is highly recommended that the first matching |
367 |
element is a string identifying the message. The one-string-only match is |
368 |
also the most efficient match (by far). |
369 |
|
370 |
Example: create a port and bind receivers on it in one go. |
371 |
|
372 |
my $port = rcv port, |
373 |
msg1 => sub { ...; 0 }, |
374 |
msg2 => sub { ...; 0 }, |
375 |
; |
376 |
|
377 |
Example: create a port, bind receivers and send it in a message elsewhere |
378 |
in one go: |
379 |
|
380 |
snd $otherport, reply => |
381 |
rcv port, |
382 |
msg1 => sub { ...; 0 }, |
383 |
... |
384 |
; |
385 |
|
386 |
=cut |
387 |
|
388 |
sub rcv($@) { |
389 |
my $port = shift; |
390 |
my ($noderef, $portid) = split /#/, $port, 2; |
391 |
|
392 |
($NODE{$noderef} || add_node $noderef) == $NODE{""} |
393 |
or Carp::croak "$port: rcv can only be called on local ports, caught"; |
394 |
|
395 |
if (@_ == 1) { |
396 |
my $cb = shift; |
397 |
delete $PORT_DATA{$portid}; |
398 |
$PORT{$portid} = sub { |
399 |
local $SELF = $port; |
400 |
eval { |
401 |
&$cb |
402 |
and kil $port; |
403 |
}; |
404 |
_self_die if $@; |
405 |
}; |
406 |
} else { |
407 |
my $self = $PORT_DATA{$portid} ||= do { |
408 |
my $self = bless { |
409 |
id => $port, |
410 |
}, "AnyEvent::MP::Port"; |
411 |
|
412 |
$PORT{$portid} = sub { |
413 |
local $SELF = $port; |
414 |
|
415 |
eval { |
416 |
for (@{ $self->{rc0}{$_[0]} }) { |
417 |
$_ && &{$_->[0]} |
418 |
&& undef $_; |
419 |
} |
420 |
|
421 |
for (@{ $self->{rcv}{$_[0]} }) { |
422 |
$_ && [@_[1 .. @{$_->[1]}]] ~~ $_->[1] |
423 |
&& &{$_->[0]} |
424 |
&& undef $_; |
425 |
} |
426 |
|
427 |
for (@{ $self->{any} }) { |
428 |
$_ && [@_[0 .. $#{$_->[1]}]] ~~ $_->[1] |
429 |
&& &{$_->[0]} |
430 |
&& undef $_; |
431 |
} |
432 |
}; |
433 |
_self_die if $@; |
434 |
}; |
435 |
|
436 |
$self |
437 |
}; |
438 |
|
439 |
"AnyEvent::MP::Port" eq ref $self |
440 |
or Carp::croak "$port: rcv can only be called on message matching ports, caught"; |
441 |
|
442 |
while (@_) { |
443 |
my ($match, $cb) = splice @_, 0, 2; |
444 |
|
445 |
if (!ref $match) { |
446 |
push @{ $self->{rc0}{$match} }, [$cb]; |
447 |
} elsif (("ARRAY" eq ref $match && !ref $match->[0])) { |
448 |
my ($type, @match) = @$match; |
449 |
@match |
450 |
? push @{ $self->{rcv}{$match->[0]} }, [$cb, \@match] |
451 |
: push @{ $self->{rc0}{$match->[0]} }, [$cb]; |
452 |
} else { |
453 |
push @{ $self->{any} }, [$cb, $match]; |
454 |
} |
455 |
} |
456 |
} |
457 |
|
458 |
$port |
459 |
} |
460 |
|
461 |
=item $closure = psub { BLOCK } |
462 |
|
463 |
Remembers C<$SELF> and creates a closure out of the BLOCK. When the |
464 |
closure is executed, sets up the environment in the same way as in C<rcv> |
465 |
callbacks, i.e. runtime errors will cause the port to get C<kil>ed. |
466 |
|
467 |
This is useful when you register callbacks from C<rcv> callbacks: |
468 |
|
469 |
rcv delayed_reply => sub { |
470 |
my ($delay, @reply) = @_; |
471 |
my $timer = AE::timer $delay, 0, psub { |
472 |
snd @reply, $SELF; |
473 |
}; |
474 |
}; |
475 |
|
476 |
=cut |
477 |
|
478 |
sub psub(&) { |
479 |
my $cb = shift; |
480 |
|
481 |
my $port = $SELF |
482 |
or Carp::croak "psub can only be called from within rcv or psub callbacks, not"; |
483 |
|
484 |
sub { |
485 |
local $SELF = $port; |
486 |
|
487 |
if (wantarray) { |
488 |
my @res = eval { &$cb }; |
489 |
_self_die if $@; |
490 |
@res |
491 |
} else { |
492 |
my $res = eval { &$cb }; |
493 |
_self_die if $@; |
494 |
$res |
495 |
} |
496 |
} |
497 |
} |
498 |
|
499 |
=item $guard = mon $port, $cb->(@reason) |
500 |
|
501 |
=item $guard = mon $port, $rcvport |
502 |
|
503 |
=item $guard = mon $port |
504 |
|
505 |
=item $guard = mon $port, $rcvport, @msg |
506 |
|
507 |
Monitor the given port and do something when the port is killed or |
508 |
messages to it were lost, and optionally return a guard that can be used |
509 |
to stop monitoring again. |
510 |
|
511 |
C<mon> effectively guarantees that, in the absence of hardware failures, |
512 |
that after starting the monitor, either all messages sent to the port |
513 |
will arrive, or the monitoring action will be invoked after possible |
514 |
message loss has been detected. No messages will be lost "in between" |
515 |
(after the first lost message no further messages will be received by the |
516 |
port). After the monitoring action was invoked, further messages might get |
517 |
delivered again. |
518 |
|
519 |
In the first form (callback), the callback is simply called with any |
520 |
number of C<@reason> elements (no @reason means that the port was deleted |
521 |
"normally"). Note also that I<< the callback B<must> never die >>, so use |
522 |
C<eval> if unsure. |
523 |
|
524 |
In the second form (another port given), the other port (C<$rcvport>) |
525 |
will be C<kil>'ed with C<@reason>, iff a @reason was specified, i.e. on |
526 |
"normal" kils nothing happens, while under all other conditions, the other |
527 |
port is killed with the same reason. |
528 |
|
529 |
The third form (kill self) is the same as the second form, except that |
530 |
C<$rvport> defaults to C<$SELF>. |
531 |
|
532 |
In the last form (message), a message of the form C<@msg, @reason> will be |
533 |
C<snd>. |
534 |
|
535 |
As a rule of thumb, monitoring requests should always monitor a port from |
536 |
a local port (or callback). The reason is that kill messages might get |
537 |
lost, just like any other message. Another less obvious reason is that |
538 |
even monitoring requests can get lost (for exmaple, when the connection |
539 |
to the other node goes down permanently). When monitoring a port locally |
540 |
these problems do not exist. |
541 |
|
542 |
Example: call a given callback when C<$port> is killed. |
543 |
|
544 |
mon $port, sub { warn "port died because of <@_>\n" }; |
545 |
|
546 |
Example: kill ourselves when C<$port> is killed abnormally. |
547 |
|
548 |
mon $port; |
549 |
|
550 |
Example: send us a restart message when another C<$port> is killed. |
551 |
|
552 |
mon $port, $self => "restart"; |
553 |
|
554 |
=cut |
555 |
|
556 |
sub mon { |
557 |
my ($noderef, $port) = split /#/, shift, 2; |
558 |
|
559 |
my $node = $NODE{$noderef} || add_node $noderef; |
560 |
|
561 |
my $cb = @_ ? shift : $SELF || Carp::croak 'mon: called with one argument only, but $SELF not set,'; |
562 |
|
563 |
unless (ref $cb) { |
564 |
if (@_) { |
565 |
# send a kill info message |
566 |
my (@msg) = ($cb, @_); |
567 |
$cb = sub { snd @msg, @_ }; |
568 |
} else { |
569 |
# simply kill other port |
570 |
my $port = $cb; |
571 |
$cb = sub { kil $port, @_ if @_ }; |
572 |
} |
573 |
} |
574 |
|
575 |
$node->monitor ($port, $cb); |
576 |
|
577 |
defined wantarray |
578 |
and AnyEvent::Util::guard { $node->unmonitor ($port, $cb) } |
579 |
} |
580 |
|
581 |
=item $guard = mon_guard $port, $ref, $ref... |
582 |
|
583 |
Monitors the given C<$port> and keeps the passed references. When the port |
584 |
is killed, the references will be freed. |
585 |
|
586 |
Optionally returns a guard that will stop the monitoring. |
587 |
|
588 |
This function is useful when you create e.g. timers or other watchers and |
589 |
want to free them when the port gets killed: |
590 |
|
591 |
$port->rcv (start => sub { |
592 |
my $timer; $timer = mon_guard $port, AE::timer 1, 1, sub { |
593 |
undef $timer if 0.9 < rand; |
594 |
}); |
595 |
}); |
596 |
|
597 |
=cut |
598 |
|
599 |
sub mon_guard { |
600 |
my ($port, @refs) = @_; |
601 |
|
602 |
#TODO: mon-less form? |
603 |
|
604 |
mon $port, sub { 0 && @refs } |
605 |
} |
606 |
|
607 |
=item kil $port[, @reason] |
608 |
|
609 |
Kill the specified port with the given C<@reason>. |
610 |
|
611 |
If no C<@reason> is specified, then the port is killed "normally" (linked |
612 |
ports will not be kileld, or even notified). |
613 |
|
614 |
Otherwise, linked ports get killed with the same reason (second form of |
615 |
C<mon>, see below). |
616 |
|
617 |
Runtime errors while evaluating C<rcv> callbacks or inside C<psub> blocks |
618 |
will be reported as reason C<< die => $@ >>. |
619 |
|
620 |
Transport/communication errors are reported as C<< transport_error => |
621 |
$message >>. |
622 |
|
623 |
=cut |
624 |
|
625 |
=item $port = spawn $node, $initfunc[, @initdata] |
626 |
|
627 |
Creates a port on the node C<$node> (which can also be a port ID, in which |
628 |
case it's the node where that port resides). |
629 |
|
630 |
The port ID of the newly created port is return immediately, and it is |
631 |
permissible to immediately start sending messages or monitor the port. |
632 |
|
633 |
After the port has been created, the init function is |
634 |
called. This function must be a fully-qualified function name |
635 |
(e.g. C<MyApp::Chat::Server::init>). To specify a function in the main |
636 |
program, use C<::name>. |
637 |
|
638 |
If the function doesn't exist, then the node tries to C<require> |
639 |
the package, then the package above the package and so on (e.g. |
640 |
C<MyApp::Chat::Server>, C<MyApp::Chat>, C<MyApp>) until the function |
641 |
exists or it runs out of package names. |
642 |
|
643 |
The init function is then called with the newly-created port as context |
644 |
object (C<$SELF>) and the C<@initdata> values as arguments. |
645 |
|
646 |
A common idiom is to pass your own port, monitor the spawned port, and |
647 |
in the init function, monitor the original port. This two-way monitoring |
648 |
ensures that both ports get cleaned up when there is a problem. |
649 |
|
650 |
Example: spawn a chat server port on C<$othernode>. |
651 |
|
652 |
# this node, executed from within a port context: |
653 |
my $server = spawn $othernode, "MyApp::Chat::Server::connect", $SELF; |
654 |
mon $server; |
655 |
|
656 |
# init function on C<$othernode> |
657 |
sub connect { |
658 |
my ($srcport) = @_; |
659 |
|
660 |
mon $srcport; |
661 |
|
662 |
rcv $SELF, sub { |
663 |
... |
664 |
}; |
665 |
} |
666 |
|
667 |
=cut |
668 |
|
669 |
sub _spawn { |
670 |
my $port = shift; |
671 |
my $init = shift; |
672 |
|
673 |
local $SELF = "$NODE#$port"; |
674 |
eval { |
675 |
&{ load_func $init } |
676 |
}; |
677 |
_self_die if $@; |
678 |
} |
679 |
|
680 |
sub spawn(@) { |
681 |
my ($noderef, undef) = split /#/, shift, 2; |
682 |
|
683 |
my $id = "$RUNIQ." . $ID++; |
684 |
|
685 |
$_[0] =~ /::/ |
686 |
or Carp::croak "spawn init function must be a fully-qualified name, caught"; |
687 |
|
688 |
($NODE{$noderef} || add_node $noderef) |
689 |
->send (["", "AnyEvent::MP::_spawn" => $id, @_]); |
690 |
|
691 |
"$noderef#$id" |
692 |
} |
693 |
|
694 |
=back |
695 |
|
696 |
=head1 NODE MESSAGES |
697 |
|
698 |
Nodes understand the following messages sent to them. Many of them take |
699 |
arguments called C<@reply>, which will simply be used to compose a reply |
700 |
message - C<$reply[0]> is the port to reply to, C<$reply[1]> the type and |
701 |
the remaining arguments are simply the message data. |
702 |
|
703 |
While other messages exist, they are not public and subject to change. |
704 |
|
705 |
=over 4 |
706 |
|
707 |
=cut |
708 |
|
709 |
=item lookup => $name, @reply |
710 |
|
711 |
Replies with the port ID of the specified well-known port, or C<undef>. |
712 |
|
713 |
=item devnull => ... |
714 |
|
715 |
Generic data sink/CPU heat conversion. |
716 |
|
717 |
=item relay => $port, @msg |
718 |
|
719 |
Simply forwards the message to the given port. |
720 |
|
721 |
=item eval => $string[ @reply] |
722 |
|
723 |
Evaluates the given string. If C<@reply> is given, then a message of the |
724 |
form C<@reply, $@, @evalres> is sent. |
725 |
|
726 |
Example: crash another node. |
727 |
|
728 |
snd $othernode, eval => "exit"; |
729 |
|
730 |
=item time => @reply |
731 |
|
732 |
Replies the the current node time to C<@reply>. |
733 |
|
734 |
Example: tell the current node to send the current time to C<$myport> in a |
735 |
C<timereply> message. |
736 |
|
737 |
snd $NODE, time => $myport, timereply => 1, 2; |
738 |
# => snd $myport, timereply => 1, 2, <time> |
739 |
|
740 |
=back |
741 |
|
742 |
=head1 AnyEvent::MP vs. Distributed Erlang |
743 |
|
744 |
AnyEvent::MP got lots of its ideas from distributed Erlang (Erlang node |
745 |
== aemp node, Erlang process == aemp port), so many of the documents and |
746 |
programming techniques employed by Erlang apply to AnyEvent::MP. Here is a |
747 |
sample: |
748 |
|
749 |
http://www.Erlang.se/doc/programming_rules.shtml |
750 |
http://Erlang.org/doc/getting_started/part_frame.html # chapters 3 and 4 |
751 |
http://Erlang.org/download/Erlang-book-part1.pdf # chapters 5 and 6 |
752 |
http://Erlang.org/download/armstrong_thesis_2003.pdf # chapters 4 and 5 |
753 |
|
754 |
Despite the similarities, there are also some important differences: |
755 |
|
756 |
=over 4 |
757 |
|
758 |
=item * Node references contain the recipe on how to contact them. |
759 |
|
760 |
Erlang relies on special naming and DNS to work everywhere in the |
761 |
same way. AEMP relies on each node knowing it's own address(es), with |
762 |
convenience functionality. |
763 |
|
764 |
This means that AEMP requires a less tightly controlled environment at the |
765 |
cost of longer node references and a slightly higher management overhead. |
766 |
|
767 |
=item * Erlang uses processes and a mailbox, AEMP does not queue. |
768 |
|
769 |
Erlang uses processes that selctively receive messages, and therefore |
770 |
needs a queue. AEMP is event based, queuing messages would serve no useful |
771 |
purpose. |
772 |
|
773 |
(But see L<Coro::MP> for a more Erlang-like process model on top of AEMP). |
774 |
|
775 |
=item * Erlang sends are synchronous, AEMP sends are asynchronous. |
776 |
|
777 |
Sending messages in Erlang is synchronous and blocks the process. AEMP |
778 |
sends are immediate, connection establishment is handled in the |
779 |
background. |
780 |
|
781 |
=item * Erlang can silently lose messages, AEMP cannot. |
782 |
|
783 |
Erlang makes few guarantees on messages delivery - messages can get lost |
784 |
without any of the processes realising it (i.e. you send messages a, b, |
785 |
and c, and the other side only receives messages a and c). |
786 |
|
787 |
AEMP guarantees correct ordering, and the guarantee that there are no |
788 |
holes in the message sequence. |
789 |
|
790 |
=item * In Erlang, processes can be declared dead and later be found to be |
791 |
alive. |
792 |
|
793 |
In Erlang it can happen that a monitored process is declared dead and |
794 |
linked processes get killed, but later it turns out that the process is |
795 |
still alive - and can receive messages. |
796 |
|
797 |
In AEMP, when port monitoring detects a port as dead, then that port will |
798 |
eventually be killed - it cannot happen that a node detects a port as dead |
799 |
and then later sends messages to it, finding it is still alive. |
800 |
|
801 |
=item * Erlang can send messages to the wrong port, AEMP does not. |
802 |
|
803 |
In Erlang it is quite possible that a node that restarts reuses a process |
804 |
ID known to other nodes for a completely different process, causing |
805 |
messages destined for that process to end up in an unrelated process. |
806 |
|
807 |
AEMP never reuses port IDs, so old messages or old port IDs floating |
808 |
around in the network will not be sent to an unrelated port. |
809 |
|
810 |
=item * Erlang uses unprotected connections, AEMP uses secure |
811 |
authentication and can use TLS. |
812 |
|
813 |
AEMP can use a proven protocol - SSL/TLS - to protect connections and |
814 |
securely authenticate nodes. |
815 |
|
816 |
=item * The AEMP protocol is optimised for both text-based and binary |
817 |
communications. |
818 |
|
819 |
The AEMP protocol, unlike the Erlang protocol, supports both |
820 |
language-independent text-only protocols (good for debugging) and binary, |
821 |
language-specific serialisers (e.g. Storable). |
822 |
|
823 |
It has also been carefully designed to be implementable in other languages |
824 |
with a minimum of work while gracefully degrading fucntionality to make the |
825 |
protocol simple. |
826 |
|
827 |
=item * AEMP has more flexible monitoring options than Erlang. |
828 |
|
829 |
In Erlang, you can chose to receive I<all> exit signals as messages |
830 |
or I<none>, there is no in-between, so monitoring single processes is |
831 |
difficult to implement. Monitoring in AEMP is more flexible than in |
832 |
Erlang, as one can choose between automatic kill, exit message or callback |
833 |
on a per-process basis. |
834 |
|
835 |
=item * Erlang tries to hide remote/local connections, AEMP does not. |
836 |
|
837 |
Monitoring in Erlang is not an indicator of process death/crashes, |
838 |
as linking is (except linking is unreliable in Erlang). |
839 |
|
840 |
In AEMP, you don't "look up" registered port names or send to named ports |
841 |
that might or might not be persistent. Instead, you normally spawn a port |
842 |
on the remote node. The init function monitors the you, and you monitor |
843 |
the remote port. Since both monitors are local to the node, they are much |
844 |
more reliable. |
845 |
|
846 |
This also saves round-trips and avoids sending messages to the wrong port |
847 |
(hard to do in Erlang). |
848 |
|
849 |
=back |
850 |
|
851 |
=head1 SEE ALSO |
852 |
|
853 |
L<AnyEvent>. |
854 |
|
855 |
=head1 AUTHOR |
856 |
|
857 |
Marc Lehmann <schmorp@schmorp.de> |
858 |
http://home.schmorp.de/ |
859 |
|
860 |
=cut |
861 |
|
862 |
1 |
863 |
|