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 1.33 by root, Wed Aug 5 22:40:51 2009 UTC vs.
Revision 1.87 by root, Fri Sep 11 02:32:23 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 - explains most concepts.
45 AnyEvent::MP::Kernel - mostly stable.
46 AnyEvent::MP::Global - stable but incomplete, protocol not yet final.
47
48stay 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
35For an introduction to this module family, see the L<AnyEvent::MP::Intro> 57For an introduction to this module family, see the L<AnyEvent::MP::Intro>
36manual page. 58manual page and the examples under F<eg/>.
37
38At the moment, this module family is severly broken and underdocumented,
39so do not use. This was uploaded mainly to reserve the CPAN namespace -
40stay tuned! The basic API should be finished, however.
41 59
42=head1 CONCEPTS 60=head1 CONCEPTS
43 61
44=over 4 62=over 4
45 63
46=item port 64=item port
47 65
48A port is something you can send messages to (with the C<snd> function). 66Not to be confused with a TCP port, a "port" is something you can send
67messages to (with the C<snd> function).
49 68
50Some ports allow you to register C<rcv> handlers that can match specific 69Ports allow you to register C<rcv> handlers that can match all or just
51messages. All C<rcv> handlers will receive messages they match, messages 70some messages. Messages send to ports will not be queued, regardless of
52will not be queued. 71anything was listening for them or not.
53 72
54=item port id - C<noderef#portname> 73=item port ID - C<nodeid#portname>
55 74
56A port id is normaly the concatenation of a noderef, a hash-mark (C<#>) as 75A port ID is the concatenation of a node ID, a hash-mark (C<#>) as
57separator, and a port name (a printable string of unspecified format). An 76separator, and a port name (a printable string of unspecified format).
58exception is the the node port, whose ID is identical to its node
59reference.
60 77
61=item node 78=item node
62 79
63A node is a single process containing at least one port - the node 80A node is a single process containing at least one port - the node port,
64port. You can send messages to node ports to find existing ports or to 81which enables nodes to manage each other remotely, and to create new
65create new ports, among other things. 82ports.
66 83
67Nodes are either private (single-process only), slaves (connected to a 84Nodes are either public (have one or more listening ports) or private
68master node only) or public nodes (connectable from unrelated nodes). 85(no listening ports). Private nodes cannot talk to other private nodes
86currently.
69 87
70=item noderef - C<host:port,host:port...>, C<id@noderef>, C<id> 88=item node ID - C<[A-Z_][a-zA-Z0-9_\-.:]*>
71 89
72A node reference is a string that either simply identifies the node (for 90A node ID is a string that uniquely identifies the node within a
73private and slave nodes), or contains a recipe on how to reach a given 91network. Depending on the configuration used, node IDs can look like a
74node (for public nodes). 92hostname, a hostname and a port, or a random string. AnyEvent::MP itself
93doesn't interpret node IDs in any way.
75 94
76This recipe is simply a comma-separated list of C<address:port> pairs (for 95=item binds - C<ip:port>
77TCP/IP, other protocols might look different).
78 96
79Node references come in two flavours: resolved (containing only numerical 97Nodes can only talk to each other by creating some kind of connection to
80addresses) or unresolved (where hostnames are used instead of addresses). 98each other. To do this, nodes should listen on one or more local transport
99endpoints - binds. Currently, only standard C<ip:port> specifications can
100be used, which specify TCP ports to listen on.
81 101
82Before using an unresolved node reference in a message you first have to 102=item seed nodes
83resolve it. 103
104When a node starts, it knows nothing about the network. To teach the node
105about the network it first has to contact some other node within the
106network. This node is called a seed.
107
108Apart from the fact that other nodes know them as seed nodes and they have
109to have fixed listening addresses, seed nodes are perfectly normal nodes -
110any node can function as a seed node for others.
111
112In addition to discovering the network, seed nodes are also used to
113maintain the network and to connect nodes that otherwise would have
114trouble connecting. They form the backbone of an AnyEvent::MP network.
115
116Seed nodes are expected to be long-running, and at least one seed node
117should always be available. They should also be relatively responsive - a
118seed node that blocks for long periods will slow down everybody else.
119
120=item seeds - C<host:port>
121
122Seeds are transport endpoint(s) (usually a hostname/IP address and a
123TCP port) of nodes thta should be used as seed nodes.
124
125The nodes listening on those endpoints are expected to be long-running,
126and at least one of those should always be available. When nodes run out
127of connections (e.g. due to a network error), they try to re-establish
128connections to some seednodes again to join the network.
84 129
85=back 130=back
86 131
87=head1 VARIABLES/FUNCTIONS 132=head1 VARIABLES/FUNCTIONS
88 133
90 135
91=cut 136=cut
92 137
93package AnyEvent::MP; 138package AnyEvent::MP;
94 139
95use AnyEvent::MP::Base; 140use AnyEvent::MP::Kernel;
96 141
97use common::sense; 142use common::sense;
98 143
99use Carp (); 144use Carp ();
100 145
101use AE (); 146use AE ();
102 147
103use base "Exporter"; 148use base "Exporter";
104 149
105our $VERSION = '0.1'; 150our $VERSION = $AnyEvent::MP::Kernel::VERSION;
151
106our @EXPORT = qw( 152our @EXPORT = qw(
107 NODE $NODE *SELF node_of _any_ 153 NODE $NODE *SELF node_of after
108 resolve_node initialise_node 154 configure
109 snd rcv mon kil reg psub 155 snd rcv mon mon_guard kil reg psub spawn cal
110 port 156 port
111); 157);
112 158
113our $SELF; 159our $SELF;
114 160
118 kil $SELF, die => $msg; 164 kil $SELF, die => $msg;
119} 165}
120 166
121=item $thisnode = NODE / $NODE 167=item $thisnode = NODE / $NODE
122 168
123The C<NODE> function returns, and the C<$NODE> variable contains 169The C<NODE> function returns, and the C<$NODE> variable contains, the node
124the noderef of the local node. The value is initialised by a call 170ID of the node running in the current process. This value is initialised by
125to C<become_public> or C<become_slave>, after which all local port 171a call to C<configure>.
126identifiers become invalid.
127 172
128=item $noderef = node_of $port 173=item $nodeid = node_of $port
129 174
130Extracts and returns the noderef from a portid or a noderef. 175Extracts and returns the node ID from a port ID or a node ID.
131 176
132=item $cv = resolve_node $noderef 177=item configure $profile, key => value...
133 178
134Takes an unresolved node reference that may contain hostnames and 179=item configure key => value...
135abbreviated IDs, resolves all of them and returns a resolved node
136reference.
137 180
138In addition to C<address:port> pairs allowed in resolved noderefs, the 181Before a node can talk to other nodes on the network (i.e. enter
139following forms are supported: 182"distributed mode") it has to configure itself - the minimum a node needs
183to know is its own name, and optionally it should know the addresses of
184some other nodes in the network to discover other nodes.
185
186This function configures a node - it must be called exactly once (or
187never) before calling other AnyEvent::MP functions.
140 188
141=over 4 189=over 4
142 190
143=item the empty string 191=item step 1, gathering configuration from profiles
144 192
145An empty-string component gets resolved as if the default port (4040) was 193The function first looks up a profile in the aemp configuration (see the
146specified. 194L<aemp> commandline utility). The profile name can be specified via the
195named C<profile> parameter or can simply be the first parameter). If it is
196missing, then the nodename (F<uname -n>) will be used as profile name.
147 197
148=item naked port numbers (e.g. C<1234>) 198The profile data is then gathered as follows:
149 199
150These are resolved by prepending the local nodename and a colon, to be 200First, all remaining key => value pairs (all of which are conveniently
151further resolved. 201undocumented at the moment) will be interpreted as configuration
202data. Then they will be overwritten by any values specified in the global
203default configuration (see the F<aemp> utility), then the chain of
204profiles chosen by the profile name (and any C<parent> attributes).
152 205
153=item hostnames (e.g. C<localhost:1234>, C<localhost>) 206That means that the values specified in the profile have highest priority
207and the values specified directly via C<configure> have lowest priority,
208and can only be used to specify defaults.
154 209
155These are resolved by using AnyEvent::DNS to resolve them, optionally 210If the profile specifies a node ID, then this will become the node ID of
156looking up SRV records for the C<aemp=4040> port, if no port was 211this process. If not, then the profile name will be used as node ID. The
157specified. 212special node ID of C<anon/> will be replaced by a random node ID.
213
214=item step 2, bind listener sockets
215
216The next step is to look up the binds in the profile, followed by binding
217aemp protocol listeners on all binds specified (it is possible and valid
218to have no binds, meaning that the node cannot be contacted form the
219outside. This means the node cannot talk to other nodes that also have no
220binds, but it can still talk to all "normal" nodes).
221
222If the profile does not specify a binds list, then a default of C<*> is
223used, meaning the node will bind on a dynamically-assigned port on every
224local IP address it finds.
225
226=item step 3, connect to seed nodes
227
228As the last step, the seeds list from the profile is passed to the
229L<AnyEvent::MP::Global> module, which will then use it to keep
230connectivity with at least one node at any point in time.
158 231
159=back 232=back
233
234Example: become a distributed node using the local node name as profile.
235This should be the most common form of invocation for "daemon"-type nodes.
236
237 configure
238
239Example: become an anonymous node. This form is often used for commandline
240clients.
241
242 configure nodeid => "anon/";
243
244Example: configure a node using a profile called seed, which si suitable
245for a seed node as it binds on all local addresses on a fixed port (4040,
246customary for aemp).
247
248 # use the aemp commandline utility
249 # aemp profile seed nodeid anon/ binds '*:4040'
250
251 # then use it
252 configure profile => "seed";
253
254 # or simply use aemp from the shell again:
255 # aemp run profile seed
256
257 # or provide a nicer-to-remember nodeid
258 # aemp run profile seed nodeid "$(hostname)"
160 259
161=item $SELF 260=item $SELF
162 261
163Contains the current port id while executing C<rcv> callbacks or C<psub> 262Contains the current port id while executing C<rcv> callbacks or C<psub>
164blocks. 263blocks.
165 264
166=item SELF, %SELF, @SELF... 265=item *SELF, SELF, %SELF, @SELF...
167 266
168Due to some quirks in how perl exports variables, it is impossible to 267Due to some quirks in how perl exports variables, it is impossible to
169just export C<$SELF>, all the symbols called C<SELF> are exported by this 268just export C<$SELF>, all the symbols named C<SELF> are exported by this
170module, but only C<$SELF> is currently used. 269module, but only C<$SELF> is currently used.
171 270
172=item snd $port, type => @data 271=item snd $port, type => @data
173 272
174=item snd $port, @msg 273=item snd $port, @msg
175 274
176Send the given message to the given port ID, which can identify either 275Send the given message to the given port, which can identify either a
177a local or a remote port, and can be either a string or soemthignt hat 276local or a remote port, and must be a port ID.
178stringifies a sa port ID (such as a port object :).
179 277
180While the message can be about anything, it is highly recommended to use a 278While the message can be almost anything, it is highly recommended to
181string as first element (a portid, or some word that indicates a request 279use a string as first element (a port ID, or some word that indicates a
182type etc.). 280request type etc.) and to consist if only simple perl values (scalars,
281arrays, hashes) - if you think you need to pass an object, think again.
183 282
184The message data effectively becomes read-only after a call to this 283The message data logically becomes read-only after a call to this
185function: modifying any argument is not allowed and can cause many 284function: modifying any argument (or values referenced by them) is
186problems. 285forbidden, as there can be considerable time between the call to C<snd>
286and the time the message is actually being serialised - in fact, it might
287never be copied as within the same process it is simply handed to the
288receiving port.
187 289
188The type of data you can transfer depends on the transport protocol: when 290The type of data you can transfer depends on the transport protocol: when
189JSON is used, then only strings, numbers and arrays and hashes consisting 291JSON is used, then only strings, numbers and arrays and hashes consisting
190of those are allowed (no objects). When Storable is used, then anything 292of those are allowed (no objects). When Storable is used, then anything
191that Storable can serialise and deserialise is allowed, and for the local 293that Storable can serialise and deserialise is allowed, and for the local
192node, anything can be passed. 294node, anything can be passed. Best rely only on the common denominator of
295these.
193 296
194=item $local_port = port 297=item $local_port = port
195 298
196Create a new local port object that can be used either as a pattern 299Create a new local port object and returns its port ID. Initially it has
197matching port ("full port") or a single-callback port ("miniport"), 300no callbacks set and will throw an error when it receives messages.
198depending on how C<rcv> callbacks are bound to the object.
199 301
200=item $port = port { my @msg = @_; $finished } 302=item $local_port = port { my @msg = @_ }
201 303
202Creates a "miniport", that is, a very lightweight port without any pattern 304Creates a new local port, and returns its ID. Semantically the same as
203matching behind it, and returns its ID. Semantically the same as creating
204a port and calling C<rcv $port, $callback> on it. 305creating a port and calling C<rcv $port, $callback> on it.
205 306
206The block will be called for every message received on the port. When the 307The block will be called for every message received on the port, with the
207callback returns a true value its job is considered "done" and the port 308global variable C<$SELF> set to the port ID. Runtime errors will cause the
208will be destroyed. Otherwise it will stay alive. 309port to be C<kil>ed. The message will be passed as-is, no extra argument
310(i.e. no port ID) will be passed to the callback.
209 311
210The message will be passed as-is, no extra argument (i.e. no port id) will 312If you want to stop/destroy the port, simply C<kil> it:
211be passed to the callback.
212 313
213If you need the local port id in the callback, this works nicely: 314 my $port = port {
214 315 my @msg = @_;
215 my $port; $port = port { 316 ...
216 snd $otherport, reply => $port; 317 kil $SELF;
217 }; 318 };
218 319
219=cut 320=cut
220 321
221sub rcv($@); 322sub rcv($@);
323
324sub _kilme {
325 die "received message on port without callback";
326}
222 327
223sub port(;&) { 328sub port(;&) {
224 my $id = "$UNIQ." . $ID++; 329 my $id = "$UNIQ." . $ID++;
225 my $port = "$NODE#$id"; 330 my $port = "$NODE#$id";
226 331
227 if (@_) { 332 rcv $port, shift || \&_kilme;
228 rcv $port, shift;
229 } else {
230 $PORT{$id} = sub { }; # nop
231 }
232 333
233 $port 334 $port
234} 335}
235 336
236=item reg $port, $name
237
238Registers the given port under the name C<$name>. If the name already
239exists it is replaced.
240
241A port can only be registered under one well known name.
242
243A port automatically becomes unregistered when it is killed.
244
245=cut
246
247sub reg(@) {
248 my ($port, $name) = @_;
249
250 $REG{$name} = $port;
251}
252
253=item rcv $port, $callback->(@msg) 337=item rcv $local_port, $callback->(@msg)
254 338
255Replaces the callback on the specified miniport (after converting it to 339Replaces the default callback on the specified port. There is no way to
256one if required). 340remove the default callback: use C<sub { }> to disable it, or better
257 341C<kil> the port when it is no longer needed.
258=item rcv $port, tagstring => $callback->(@msg), ...
259
260=item rcv $port, $smartmatch => $callback->(@msg), ...
261
262=item rcv $port, [$smartmatch...] => $callback->(@msg), ...
263
264Register callbacks to be called on matching messages on the given full
265port (after converting it to one if required).
266
267The callback has to return a true value when its work is done, after
268which is will be removed, or a false value in which case it will stay
269registered.
270 342
271The global C<$SELF> (exported by this module) contains C<$port> while 343The global C<$SELF> (exported by this module) contains C<$port> while
272executing the callback. 344executing the callback. Runtime errors during callback execution will
345result in the port being C<kil>ed.
273 346
274Runtime errors wdurign callback execution will result in the port being 347The default callback received all messages not matched by a more specific
275C<kil>ed. 348C<tag> match.
276 349
277If the match is an array reference, then it will be matched against the 350=item rcv $local_port, tag => $callback->(@msg_without_tag), ...
278first elements of the message, otherwise only the first element is being
279matched.
280 351
281Any element in the match that is specified as C<_any_> (a function 352Register (or replace) callbacks to be called on messages starting with the
282exported by this module) matches any single element of the message. 353given tag on the given port (and return the port), or unregister it (when
354C<$callback> is C<$undef> or missing). There can only be one callback
355registered for each tag.
283 356
284While not required, it is highly recommended that the first matching 357The original message will be passed to the callback, after the first
285element is a string identifying the message. The one-string-only match is 358element (the tag) has been removed. The callback will use the same
286also the most efficient match (by far). 359environment as the default callback (see above).
360
361Example: create a port and bind receivers on it in one go.
362
363 my $port = rcv port,
364 msg1 => sub { ... },
365 msg2 => sub { ... },
366 ;
367
368Example: create a port, bind receivers and send it in a message elsewhere
369in one go:
370
371 snd $otherport, reply =>
372 rcv port,
373 msg1 => sub { ... },
374 ...
375 ;
376
377Example: temporarily register a rcv callback for a tag matching some port
378(e.g. for a rpc reply) and unregister it after a message was received.
379
380 rcv $port, $otherport => sub {
381 my @reply = @_;
382
383 rcv $SELF, $otherport;
384 };
287 385
288=cut 386=cut
289 387
290sub rcv($@) { 388sub rcv($@) {
291 my $port = shift; 389 my $port = shift;
292 my ($noderef, $portid) = split /#/, $port, 2; 390 my ($nodeid, $portid) = split /#/, $port, 2;
293 391
294 ($NODE{$noderef} || add_node $noderef) == $NODE{""} 392 $NODE{$nodeid} == $NODE{""}
295 or Carp::croak "$port: rcv can only be called on local ports, caught"; 393 or Carp::croak "$port: rcv can only be called on local ports, caught";
296 394
297 if (@_ == 1) { 395 while (@_) {
396 if (ref $_[0]) {
397 if (my $self = $PORT_DATA{$portid}) {
398 "AnyEvent::MP::Port" eq ref $self
399 or Carp::croak "$port: rcv can only be called on message matching ports, caught";
400
401 $self->[2] = shift;
402 } else {
298 my $cb = shift; 403 my $cb = shift;
299 delete $PORT_DATA{$portid};
300 $PORT{$portid} = sub { 404 $PORT{$portid} = sub {
301 local $SELF = $port; 405 local $SELF = $port;
302 eval { 406 eval { &$cb }; _self_die if $@;
303 &$cb 407 };
304 and kil $port;
305 }; 408 }
306 _self_die if $@; 409 } elsif (defined $_[0]) {
307 };
308 } else {
309 my $self = $PORT_DATA{$portid} ||= do { 410 my $self = $PORT_DATA{$portid} ||= do {
310 my $self = bless { 411 my $self = bless [$PORT{$port} || sub { }, { }, $port], "AnyEvent::MP::Port";
311 id => $port,
312 }, "AnyEvent::MP::Port";
313 412
314 $PORT{$portid} = sub { 413 $PORT{$portid} = sub {
315 local $SELF = $port; 414 local $SELF = $port;
316 415
317 eval {
318 for (@{ $self->{rc0}{$_[0]} }) { 416 if (my $cb = $self->[1]{$_[0]}) {
319 $_ && &{$_->[0]} 417 shift;
320 && undef $_; 418 eval { &$cb }; _self_die if $@;
321 } 419 } else {
322
323 for (@{ $self->{rcv}{$_[0]} }) {
324 $_ && [@_[1 .. @{$_->[1]}]] ~~ $_->[1]
325 && &{$_->[0]} 420 &{ $self->[0] };
326 && undef $_;
327 }
328
329 for (@{ $self->{any} }) {
330 $_ && [@_[0 .. $#{$_->[1]}]] ~~ $_->[1]
331 && &{$_->[0]}
332 && undef $_;
333 } 421 }
334 }; 422 };
335 _self_die if $@; 423
424 $self
336 }; 425 };
337 426
338 $self
339 };
340
341 "AnyEvent::MP::Port" eq ref $self 427 "AnyEvent::MP::Port" eq ref $self
342 or Carp::croak "$port: rcv can only be called on message matching ports, caught"; 428 or Carp::croak "$port: rcv can only be called on message matching ports, caught";
343 429
344 while (@_) {
345 my ($match, $cb) = splice @_, 0, 2; 430 my ($tag, $cb) = splice @_, 0, 2;
346 431
347 if (!ref $match) { 432 if (defined $cb) {
348 push @{ $self->{rc0}{$match} }, [$cb]; 433 $self->[1]{$tag} = $cb;
349 } elsif (("ARRAY" eq ref $match && !ref $match->[0])) {
350 my ($type, @match) = @$match;
351 @match
352 ? push @{ $self->{rcv}{$match->[0]} }, [$cb, \@match]
353 : push @{ $self->{rc0}{$match->[0]} }, [$cb];
354 } else { 434 } else {
355 push @{ $self->{any} }, [$cb, $match]; 435 delete $self->[1]{$tag};
356 } 436 }
357 } 437 }
358 } 438 }
359 439
360 $port 440 $port
396 $res 476 $res
397 } 477 }
398 } 478 }
399} 479}
400 480
401=item $guard = mon $port, $cb->(@reason) 481=item $guard = mon $port, $cb->(@reason) # call $cb when $port dies
402 482
403=item $guard = mon $port, $otherport 483=item $guard = mon $port, $rcvport # kill $rcvport when $port dies
404 484
405=item $guard = mon $port, $otherport, @msg 485=item $guard = mon $port # kill $SELF when $port dies
406 486
487=item $guard = mon $port, $rcvport, @msg # send a message when $port dies
488
407Monitor the given port and do something when the port is killed. 489Monitor the given port and do something when the port is killed or
490messages to it were lost, and optionally return a guard that can be used
491to stop monitoring again.
408 492
409In the first form, the callback is simply called with any number 493In the first form (callback), the callback is simply called with any
410of C<@reason> elements (no @reason means that the port was deleted 494number of C<@reason> elements (no @reason means that the port was deleted
411"normally"). Note also that I<< the callback B<must> never die >>, so use 495"normally"). Note also that I<< the callback B<must> never die >>, so use
412C<eval> if unsure. 496C<eval> if unsure.
413 497
414In the second form, the other port will be C<kil>'ed with C<@reason>, iff 498In the second form (another port given), the other port (C<$rcvport>)
415a @reason was specified, i.e. on "normal" kils nothing happens, while 499will be C<kil>'ed with C<@reason>, if a @reason was specified, i.e. on
416under all other conditions, the other port is killed with the same reason. 500"normal" kils nothing happens, while under all other conditions, the other
501port is killed with the same reason.
417 502
503The third form (kill self) is the same as the second form, except that
504C<$rvport> defaults to C<$SELF>.
505
418In the last form, a message of the form C<@msg, @reason> will be C<snd>. 506In the last form (message), a message of the form C<@msg, @reason> will be
507C<snd>.
508
509Monitoring-actions are one-shot: once messages are lost (and a monitoring
510alert was raised), they are removed and will not trigger again.
511
512As a rule of thumb, monitoring requests should always monitor a port from
513a local port (or callback). The reason is that kill messages might get
514lost, just like any other message. Another less obvious reason is that
515even monitoring requests can get lost (for example, when the connection
516to the other node goes down permanently). When monitoring a port locally
517these problems do not exist.
518
519C<mon> effectively guarantees that, in the absence of hardware failures,
520after starting the monitor, either all messages sent to the port will
521arrive, or the monitoring action will be invoked after possible message
522loss has been detected. No messages will be lost "in between" (after
523the first lost message no further messages will be received by the
524port). After the monitoring action was invoked, further messages might get
525delivered again.
526
527Inter-host-connection timeouts and monitoring depend on the transport
528used. The only transport currently implemented is TCP, and AnyEvent::MP
529relies on TCP to detect node-downs (this can take 10-15 minutes on a
530non-idle connection, and usually around two hours for idle conenctions).
531
532This means that monitoring is good for program errors and cleaning up
533stuff eventually, but they are no replacement for a timeout when you need
534to ensure some maximum latency.
419 535
420Example: call a given callback when C<$port> is killed. 536Example: call a given callback when C<$port> is killed.
421 537
422 mon $port, sub { warn "port died because of <@_>\n" }; 538 mon $port, sub { warn "port died because of <@_>\n" };
423 539
424Example: kill ourselves when C<$port> is killed abnormally. 540Example: kill ourselves when C<$port> is killed abnormally.
425 541
426 mon $port, $self; 542 mon $port;
427 543
428Example: send us a restart message another C<$port> is killed. 544Example: send us a restart message when another C<$port> is killed.
429 545
430 mon $port, $self => "restart"; 546 mon $port, $self => "restart";
431 547
432=cut 548=cut
433 549
434sub mon { 550sub mon {
435 my ($noderef, $port) = split /#/, shift, 2; 551 my ($nodeid, $port) = split /#/, shift, 2;
436 552
437 my $node = $NODE{$noderef} || add_node $noderef; 553 my $node = $NODE{$nodeid} || add_node $nodeid;
438 554
439 my $cb = shift; 555 my $cb = @_ ? shift : $SELF || Carp::croak 'mon: called with one argument only, but $SELF not set,';
440 556
441 unless (ref $cb) { 557 unless (ref $cb) {
442 if (@_) { 558 if (@_) {
443 # send a kill info message 559 # send a kill info message
444 my (@msg) = ($cb, @_); 560 my (@msg) = ($cb, @_);
462is killed, the references will be freed. 578is killed, the references will be freed.
463 579
464Optionally returns a guard that will stop the monitoring. 580Optionally returns a guard that will stop the monitoring.
465 581
466This function is useful when you create e.g. timers or other watchers and 582This function is useful when you create e.g. timers or other watchers and
467want to free them when the port gets killed: 583want to free them when the port gets killed (note the use of C<psub>):
468 584
469 $port->rcv (start => sub { 585 $port->rcv (start => sub {
470 my $timer; $timer = mon_guard $port, AE::timer 1, 1, sub { 586 my $timer; $timer = mon_guard $port, AE::timer 1, 1, psub {
471 undef $timer if 0.9 < rand; 587 undef $timer if 0.9 < rand;
472 }); 588 });
473 }); 589 });
474 590
475=cut 591=cut
476 592
477sub mon_guard { 593sub mon_guard {
478 my ($port, @refs) = @_; 594 my ($port, @refs) = @_;
479 595
596 #TODO: mon-less form?
597
480 mon $port, sub { 0 && @refs } 598 mon $port, sub { 0 && @refs }
481} 599}
482 600
483=item lnk $port1, $port2
484
485Link two ports. This is simply a shorthand for:
486
487 mon $port1, $port2;
488 mon $port2, $port1;
489
490It means that if either one is killed abnormally, the other one gets
491killed as well.
492
493=item kil $port[, @reason] 601=item kil $port[, @reason]
494 602
495Kill the specified port with the given C<@reason>. 603Kill the specified port with the given C<@reason>.
496 604
497If no C<@reason> is specified, then the port is killed "normally" (linked 605If no C<@reason> is specified, then the port is killed "normally" (ports
498ports will not be kileld, or even notified). 606monitoring other ports will not necessarily die because a port dies
607"normally").
499 608
500Otherwise, linked ports get killed with the same reason (second form of 609Otherwise, linked ports get killed with the same reason (second form of
501C<mon>, see below). 610C<mon>, see above).
502 611
503Runtime errors while evaluating C<rcv> callbacks or inside C<psub> blocks 612Runtime errors while evaluating C<rcv> callbacks or inside C<psub> blocks
504will be reported as reason C<< die => $@ >>. 613will be reported as reason C<< die => $@ >>.
505 614
506Transport/communication errors are reported as C<< transport_error => 615Transport/communication errors are reported as C<< transport_error =>
507$message >>. 616$message >>.
508 617
618=cut
619
620=item $port = spawn $node, $initfunc[, @initdata]
621
622Creates a port on the node C<$node> (which can also be a port ID, in which
623case it's the node where that port resides).
624
625The port ID of the newly created port is returned immediately, and it is
626possible to immediately start sending messages or to monitor the port.
627
628After the port has been created, the init function is called on the remote
629node, in the same context as a C<rcv> callback. This function must be a
630fully-qualified function name (e.g. C<MyApp::Chat::Server::init>). To
631specify a function in the main program, use C<::name>.
632
633If the function doesn't exist, then the node tries to C<require>
634the package, then the package above the package and so on (e.g.
635C<MyApp::Chat::Server>, C<MyApp::Chat>, C<MyApp>) until the function
636exists or it runs out of package names.
637
638The init function is then called with the newly-created port as context
639object (C<$SELF>) and the C<@initdata> values as arguments. It I<must>
640call one of the C<rcv> functions to set callbacks on C<$SELF>, otherwise
641the port might not get created.
642
643A common idiom is to pass a local port, immediately monitor the spawned
644port, and in the remote init function, immediately monitor the passed
645local port. This two-way monitoring ensures that both ports get cleaned up
646when there is a problem.
647
648C<spawn> guarantees that the C<$initfunc> has no visible effects on the
649caller before C<spawn> returns (by delaying invocation when spawn is
650called for the local node).
651
652Example: spawn a chat server port on C<$othernode>.
653
654 # this node, executed from within a port context:
655 my $server = spawn $othernode, "MyApp::Chat::Server::connect", $SELF;
656 mon $server;
657
658 # init function on C<$othernode>
659 sub connect {
660 my ($srcport) = @_;
661
662 mon $srcport;
663
664 rcv $SELF, sub {
665 ...
666 };
667 }
668
669=cut
670
671sub _spawn {
672 my $port = shift;
673 my $init = shift;
674
675 # rcv will create the actual port
676 local $SELF = "$NODE#$port";
677 eval {
678 &{ load_func $init }
679 };
680 _self_die if $@;
681}
682
683sub spawn(@) {
684 my ($nodeid, undef) = split /#/, shift, 2;
685
686 my $id = "$RUNIQ." . $ID++;
687
688 $_[0] =~ /::/
689 or Carp::croak "spawn init function must be a fully-qualified name, caught";
690
691 snd_to_func $nodeid, "AnyEvent::MP::_spawn" => $id, @_;
692
693 "$nodeid#$id"
694}
695
696=item after $timeout, @msg
697
698=item after $timeout, $callback
699
700Either sends the given message, or call the given callback, after the
701specified number of seconds.
702
703This is simply a utility function that comes in handy at times - the
704AnyEvent::MP author is not convinced of the wisdom of having it, though,
705so it may go away in the future.
706
707=cut
708
709sub after($@) {
710 my ($timeout, @action) = @_;
711
712 my $t; $t = AE::timer $timeout, 0, sub {
713 undef $t;
714 ref $action[0]
715 ? $action[0]()
716 : snd @action;
717 };
718}
719
720=item cal $port, @msg, $callback[, $timeout]
721
722A simple form of RPC - sends a message to the given C<$port> with the
723given contents (C<@msg>), but adds a reply port to the message.
724
725The reply port is created temporarily just for the purpose of receiving
726the reply, and will be C<kil>ed when no longer needed.
727
728A reply message sent to the port is passed to the C<$callback> as-is.
729
730If an optional time-out (in seconds) is given and it is not C<undef>,
731then the callback will be called without any arguments after the time-out
732elapsed and the port is C<kil>ed.
733
734If no time-out is given, then the local port will monitor the remote port
735instead, so it eventually gets cleaned-up.
736
737Currently this function returns the temporary port, but this "feature"
738might go in future versions unless you can make a convincing case that
739this is indeed useful for something.
740
741=cut
742
743sub cal(@) {
744 my $timeout = ref $_[-1] ? undef : pop;
745 my $cb = pop;
746
747 my $port = port {
748 undef $timeout;
749 kil $SELF;
750 &$cb;
751 };
752
753 if (defined $timeout) {
754 $timeout = AE::timer $timeout, 0, sub {
755 undef $timeout;
756 kil $port;
757 $cb->();
758 };
759 } else {
760 mon $_[0], sub {
761 kil $port;
762 $cb->();
763 };
764 }
765
766 push @_, $port;
767 &snd;
768
769 $port
770}
771
509=back 772=back
510 773
511=head1 FUNCTIONS FOR NODES 774=head1 AnyEvent::MP vs. Distributed Erlang
775
776AnyEvent::MP got lots of its ideas from distributed Erlang (Erlang node
777== aemp node, Erlang process == aemp port), so many of the documents and
778programming techniques employed by Erlang apply to AnyEvent::MP. Here is a
779sample:
780
781 http://www.Erlang.se/doc/programming_rules.shtml
782 http://Erlang.org/doc/getting_started/part_frame.html # chapters 3 and 4
783 http://Erlang.org/download/Erlang-book-part1.pdf # chapters 5 and 6
784 http://Erlang.org/download/armstrong_thesis_2003.pdf # chapters 4 and 5
785
786Despite the similarities, there are also some important differences:
512 787
513=over 4 788=over 4
514 789
515=item initialise_node $noderef, $seednode, $seednode... 790=item * Node IDs are arbitrary strings in AEMP.
516 791
517=item initialise_node "slave/", $master, $master...
518
519Initialises a node - must be called exactly once before calling other
520AnyEvent::MP functions when talking to other nodes is required.
521
522All arguments are noderefs, which can be either resolved or unresolved.
523
524There are two types of networked nodes, public nodes and slave nodes:
525
526=over 4
527
528=item public nodes
529
530For public nodes, C<$noderef> must either be a (possibly unresolved)
531noderef, in which case it will be resolved, or C<undef> (or missing), in
532which case the noderef will be guessed.
533
534Afterwards, the node will bind itself on all endpoints and try to connect
535to all additional C<$seednodes> that are specified. Seednodes are optional
536and can be used to quickly bootstrap the node into an existing network.
537
538=item slave nodes
539
540When the C<$noderef> is the special string C<slave/>, then the node will
541become a slave node. Slave nodes cannot be contacted from outside and will
542route most of their traffic to the master node that they attach to.
543
544At least one additional noderef is required: The node will try to connect
545to all of them and will become a slave attached to the first node it can
546successfully connect to.
547
548=back
549
550This function will block until all nodes have been resolved and, for slave
551nodes, until it has successfully established a connection to a master
552server.
553
554Example: become a public node listening on the default node.
555
556 initialise_node;
557
558Example: become a public node, and try to contact some well-known master
559servers to become part of the network.
560
561 initialise_node undef, "master1", "master2";
562
563Example: become a public node listening on port C<4041>.
564
565 initialise_node 4041;
566
567Example: become a public node, only visible on localhost port 4044.
568
569 initialise_node "locahost:4044";
570
571Example: become a slave node to any of the specified master servers.
572
573 initialise_node "slave/", "master1", "192.168.13.17", "mp.example.net";
574
575=cut
576
577=back
578
579=head1 NODE MESSAGES
580
581Nodes understand the following messages sent to them. Many of them take
582arguments called C<@reply>, which will simply be used to compose a reply
583message - C<$reply[0]> is the port to reply to, C<$reply[1]> the type and
584the remaining arguments are simply the message data.
585
586While other messages exist, they are not public and subject to change.
587
588=over 4
589
590=cut
591
592=item lookup => $name, @reply
593
594Replies with the port ID of the specified well-known port, or C<undef>.
595
596=item devnull => ...
597
598Generic data sink/CPU heat conversion.
599
600=item relay => $port, @msg
601
602Simply forwards the message to the given port.
603
604=item eval => $string[ @reply]
605
606Evaluates the given string. If C<@reply> is given, then a message of the
607form C<@reply, $@, @evalres> is sent.
608
609Example: crash another node.
610
611 snd $othernode, eval => "exit";
612
613=item time => @reply
614
615Replies the the current node time to C<@reply>.
616
617Example: tell the current node to send the current time to C<$myport> in a
618C<timereply> message.
619
620 snd $NODE, time => $myport, timereply => 1, 2;
621 # => snd $myport, timereply => 1, 2, <time>
622
623=back
624
625=head1 AnyEvent::MP vs. Distributed Erlang
626
627AnyEvent::MP got lots of its ideas from distributed erlang (erlang node
628== aemp node, erlang process == aemp port), so many of the documents and
629programming techniques employed by erlang apply to AnyEvent::MP. Here is a
630sample:
631
632 http://www.erlang.se/doc/programming_rules.shtml
633 http://erlang.org/doc/getting_started/part_frame.html # chapters 3 and 4
634 http://erlang.org/download/erlang-book-part1.pdf # chapters 5 and 6
635 http://erlang.org/download/armstrong_thesis_2003.pdf # chapters 4 and 5
636
637Despite the similarities, there are also some important differences:
638
639=over 4
640
641=item * Node references contain the recipe on how to contact them.
642
643Erlang relies on special naming and DNS to work everywhere in the 792Erlang relies on special naming and DNS to work everywhere in the same
644same way. AEMP relies on each node knowing it's own address(es), with 793way. AEMP relies on each node somehow knowing its own address(es) (e.g. by
645convenience functionality. 794configuration or DNS), but will otherwise discover other odes itself.
646 795
647This means that AEMP requires a less tightly controlled environment at the 796=item * Erlang has a "remote ports are like local ports" philosophy, AEMP
648cost of longer node references and a slightly higher management overhead. 797uses "local ports are like remote ports".
798
799The failure modes for local ports are quite different (runtime errors
800only) then for remote ports - when a local port dies, you I<know> it dies,
801when a connection to another node dies, you know nothing about the other
802port.
803
804Erlang pretends remote ports are as reliable as local ports, even when
805they are not.
806
807AEMP encourages a "treat remote ports differently" philosophy, with local
808ports being the special case/exception, where transport errors cannot
809occur.
649 810
650=item * Erlang uses processes and a mailbox, AEMP does not queue. 811=item * Erlang uses processes and a mailbox, AEMP does not queue.
651 812
652Erlang uses processes that selctively receive messages, and therefore 813Erlang uses processes that selectively receive messages, and therefore
653needs a queue. AEMP is event based, queuing messages would serve no useful 814needs a queue. AEMP is event based, queuing messages would serve no
654purpose. 815useful purpose. For the same reason the pattern-matching abilities of
816AnyEvent::MP are more limited, as there is little need to be able to
817filter messages without dequeuing them.
655 818
656(But see L<Coro::MP> for a more erlang-like process model on top of AEMP). 819(But see L<Coro::MP> for a more Erlang-like process model on top of AEMP).
657 820
658=item * Erlang sends are synchronous, AEMP sends are asynchronous. 821=item * Erlang sends are synchronous, AEMP sends are asynchronous.
659 822
660Sending messages in erlang is synchronous and blocks the process. AEMP 823Sending messages in Erlang is synchronous and blocks the process (and
661sends are immediate, connection establishment is handled in the 824so does not need a queue that can overflow). AEMP sends are immediate,
662background. 825connection establishment is handled in the background.
663 826
664=item * Erlang can silently lose messages, AEMP cannot. 827=item * Erlang suffers from silent message loss, AEMP does not.
665 828
666Erlang makes few guarantees on messages delivery - messages can get lost 829Erlang makes few guarantees on messages delivery - messages can get lost
667without any of the processes realising it (i.e. you send messages a, b, 830without any of the processes realising it (i.e. you send messages a, b,
668and c, and the other side only receives messages a and c). 831and c, and the other side only receives messages a and c).
669 832
670AEMP guarantees correct ordering, and the guarantee that there are no 833AEMP guarantees correct ordering, and the guarantee that after one message
671holes in the message sequence. 834is lost, all following ones sent to the same port are lost as well, until
672 835monitoring raises an error, so there are no silent "holes" in the message
673=item * In erlang, processes can be declared dead and later be found to be 836sequence.
674alive.
675
676In erlang it can happen that a monitored process is declared dead and
677linked processes get killed, but later it turns out that the process is
678still alive - and can receive messages.
679
680In AEMP, when port monitoring detects a port as dead, then that port will
681eventually be killed - it cannot happen that a node detects a port as dead
682and then later sends messages to it, finding it is still alive.
683 837
684=item * Erlang can send messages to the wrong port, AEMP does not. 838=item * Erlang can send messages to the wrong port, AEMP does not.
685 839
686In erlang it is quite possible that a node that restarts reuses a process 840In Erlang it is quite likely that a node that restarts reuses a process ID
687ID known to other nodes for a completely different process, causing 841known to other nodes for a completely different process, causing messages
688messages destined for that process to end up in an unrelated process. 842destined for that process to end up in an unrelated process.
689 843
690AEMP never reuses port IDs, so old messages or old port IDs floating 844AEMP never reuses port IDs, so old messages or old port IDs floating
691around in the network will not be sent to an unrelated port. 845around in the network will not be sent to an unrelated port.
692 846
693=item * Erlang uses unprotected connections, AEMP uses secure 847=item * Erlang uses unprotected connections, AEMP uses secure
694authentication and can use TLS. 848authentication and can use TLS.
695 849
696AEMP can use a proven protocol - SSL/TLS - to protect connections and 850AEMP can use a proven protocol - TLS - to protect connections and
697securely authenticate nodes. 851securely authenticate nodes.
698 852
699=item * The AEMP protocol is optimised for both text-based and binary 853=item * The AEMP protocol is optimised for both text-based and binary
700communications. 854communications.
701 855
702The AEMP protocol, unlike the erlang protocol, supports both 856The AEMP protocol, unlike the Erlang protocol, supports both programming
703language-independent text-only protocols (good for debugging) and binary, 857language independent text-only protocols (good for debugging) and binary,
704language-specific serialisers (e.g. Storable). 858language-specific serialisers (e.g. Storable). By default, unless TLS is
859used, the protocol is actually completely text-based.
705 860
706It has also been carefully designed to be implementable in other languages 861It has also been carefully designed to be implementable in other languages
707with a minimum of work while gracefully degrading fucntionality to make the 862with a minimum of work while gracefully degrading functionality to make the
708protocol simple. 863protocol simple.
709 864
865=item * AEMP has more flexible monitoring options than Erlang.
866
867In Erlang, you can chose to receive I<all> exit signals as messages
868or I<none>, there is no in-between, so monitoring single processes is
869difficult to implement. Monitoring in AEMP is more flexible than in
870Erlang, as one can choose between automatic kill, exit message or callback
871on a per-process basis.
872
873=item * Erlang tries to hide remote/local connections, AEMP does not.
874
875Monitoring in Erlang is not an indicator of process death/crashes, in the
876same way as linking is (except linking is unreliable in Erlang).
877
878In AEMP, you don't "look up" registered port names or send to named ports
879that might or might not be persistent. Instead, you normally spawn a port
880on the remote node. The init function monitors you, and you monitor the
881remote port. Since both monitors are local to the node, they are much more
882reliable (no need for C<spawn_link>).
883
884This also saves round-trips and avoids sending messages to the wrong port
885(hard to do in Erlang).
886
710=back 887=back
711 888
889=head1 RATIONALE
890
891=over 4
892
893=item Why strings for port and node IDs, why not objects?
894
895We considered "objects", but found that the actual number of methods
896that can be called are quite low. Since port and node IDs travel over
897the network frequently, the serialising/deserialising would add lots of
898overhead, as well as having to keep a proxy object everywhere.
899
900Strings can easily be printed, easily serialised etc. and need no special
901procedures to be "valid".
902
903And as a result, a miniport consists of a single closure stored in a
904global hash - it can't become much cheaper.
905
906=item Why favour JSON, why not a real serialising format such as Storable?
907
908In fact, any AnyEvent::MP node will happily accept Storable as framing
909format, but currently there is no way to make a node use Storable by
910default (although all nodes will accept it).
911
912The default framing protocol is JSON because a) JSON::XS is many times
913faster for small messages and b) most importantly, after years of
914experience we found that object serialisation is causing more problems
915than it solves: Just like function calls, objects simply do not travel
916easily over the network, mostly because they will always be a copy, so you
917always have to re-think your design.
918
919Keeping your messages simple, concentrating on data structures rather than
920objects, will keep your messages clean, tidy and efficient.
921
922=back
923
712=head1 SEE ALSO 924=head1 SEE ALSO
925
926L<AnyEvent::MP::Intro> - a gentle introduction.
927
928L<AnyEvent::MP::Kernel> - more, lower-level, stuff.
929
930L<AnyEvent::MP::Global> - network maintainance and port groups, to find
931your applications.
932
933L<AnyEvent::MP::LogCatcher> - simple service to display log messages from
934all nodes.
713 935
714L<AnyEvent>. 936L<AnyEvent>.
715 937
716=head1 AUTHOR 938=head1 AUTHOR
717 939

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines