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

Comparing AnyEvent-Fork-RPC/RPC.pm (file contents):
Revision 1.1 by root, Wed Apr 17 15:55:59 2013 UTC vs.
Revision 1.45 by root, Thu Sep 12 15:15:49 2019 UTC

…		…
12	->require ("MyModule")	12	->require ("MyModule")
13	->AnyEvent::Fork::RPC::run (	13	->AnyEvent::Fork::RPC::run (
14	"MyModule::server",	14	"MyModule::server",
15	);	15	);
16		16
		17	use AnyEvent;
		18
17	my $cv = AE::cv;	19	my $cv = AE::cv;
18		20
19	$rpc->(1, 2, 3, sub {	21	$rpc->(1, 2, 3, sub {
20	print "MyModule::server returned @_\n";	22	print "MyModule::server returned @_\n";
21	$cv->send;	23	$cv->send;
…		…
24	$cv->recv;	26	$cv->recv;
25		27
26	=head1 DESCRIPTION	28	=head1 DESCRIPTION
27		29
28	This module implements a simple RPC protocol and backend for processes	30	This module implements a simple RPC protocol and backend for processes
29	created via L<AnyEvent::Fork>, allowing you to call a function in the	31	created via L<AnyEvent::Fork> or L<AnyEvent::Fork::Remote>, allowing you
30	child process and receive its return values (up to 4GB serialised).	32	to call a function in the child process and receive its return values (up
		33	to 4GB serialised).
31		34
32	It implements two different backends: a synchronous one that works like a	35	It implements two different backends: a synchronous one that works like a
33	normal function call, and an asynchronous one that can run multiple jobs	36	normal function call, and an asynchronous one that can run multiple jobs
34	concurrently in the child, using AnyEvent.	37	concurrently in the child, using AnyEvent.
35		38
36	It also implements an asynchronous event mechanism from the child to the	39	It also implements an asynchronous event mechanism from the child to the
37	parent, that could be used for progress indications or other information.	40	parent, that could be used for progress indications or other information.
38		41
		42	=head1 EXAMPLES
		43
		44	=head2 Example 1: Synchronous Backend
		45
		46	Here is a simple example that implements a backend that executes C<unlink>
		47	and C<rmdir> calls, and reports their status back. It also reports the
		48	number of requests it has processed every three requests, which is clearly
		49	silly, but illustrates the use of events.
		50
		51	First the parent process:
		52
		53	use AnyEvent;
		54	use AnyEvent::Fork;
		55	use AnyEvent::Fork::RPC;
		56
		57	my $done = AE::cv;
		58
		59	my $rpc = AnyEvent::Fork
		60	->new
		61	->require ("MyWorker")
		62	->AnyEvent::Fork::RPC::run ("MyWorker::run",
		63	on_error => sub { warn "ERROR: $_[0]"; exit 1 },
		64	on_event => sub { warn "$_[0] requests handled\n" },
		65	on_destroy => $done,
		66	);
		67
		68	for my $id (1..6) {
		69	$rpc->(rmdir => "/tmp/somepath/$id", sub {
		70	$_[0]
		71	or warn "/tmp/somepath/$id: $_[1]\n";
		72	});
		73	}
		74
		75	undef $rpc;
		76
		77	$done->recv;
		78
		79	The parent creates the process, queues a few rmdir's. It then forgets
		80	about the C<$rpc> object, so that the child exits after it has handled the
		81	requests, and then it waits till the requests have been handled.
		82
		83	The child is implemented using a separate module, C<MyWorker>, shown here:
		84
		85	package MyWorker;
		86
		87	my $count;
		88
		89	sub run {
		90	my ($cmd, $path) = @_;
		91
		92	AnyEvent::Fork::RPC::event ($count)
		93	unless ++$count % 3;
		94
		95	my $status = $cmd eq "rmdir" ? rmdir $path
		96	: $cmd eq "unlink" ? unlink $path
		97	: die "fatal error, illegal command '$cmd'";
		98
		99	$status or (0, "$!")
		100	}
		101
		102	1
		103
		104	The C<run> function first sends a "progress" event every three calls, and
		105	then executes C<rmdir> or C<unlink>, depending on the first parameter (or
		106	dies with a fatal error - obviously, you must never let this happen :).
		107
		108	Eventually it returns the status value true if the command was successful,
		109	or the status value 0 and the stringified error message.
		110
		111	On my system, running the first code fragment with the given
		112	F<MyWorker.pm> in the current directory yields:
		113
		114	/tmp/somepath/1: No such file or directory
		115	/tmp/somepath/2: No such file or directory
		116	3 requests handled
		117	/tmp/somepath/3: No such file or directory
		118	/tmp/somepath/4: No such file or directory
		119	/tmp/somepath/5: No such file or directory
		120	6 requests handled
		121	/tmp/somepath/6: No such file or directory
		122
		123	Obviously, none of the directories I am trying to delete even exist. Also,
		124	the events and responses are processed in exactly the same order as
		125	they were created in the child, which is true for both synchronous and
		126	asynchronous backends.
		127
		128	Note that the parentheses in the call to C<AnyEvent::Fork::RPC::event> are
		129	not optional. That is because the function isn't defined when the code is
		130	compiled. You can make sure it is visible by pre-loading the correct
		131	backend module in the call to C<require>:
		132
		133	->require ("AnyEvent::Fork::RPC::Sync", "MyWorker")
		134
		135	Since the backend module declares the C<event> function, loading it first
		136	ensures that perl will correctly interpret calls to it.
		137
		138	And as a final remark, there is a fine module on CPAN that can
		139	asynchronously C<rmdir> and C<unlink> and a lot more, and more efficiently
		140	than this example, namely L<IO::AIO>.
		141
		142	=head3 Example 1a: the same with the asynchronous backend
		143
		144	This example only shows what needs to be changed to use the async backend
		145	instead. Doing this is not very useful, the purpose of this example is
		146	to show the minimum amount of change that is required to go from the
		147	synchronous to the asynchronous backend.
		148
		149	To use the async backend in the previous example, you need to add the
		150	C<async> parameter to the C<AnyEvent::Fork::RPC::run> call:
		151
		152	->AnyEvent::Fork::RPC::run ("MyWorker::run",
		153	async => 1,
		154	...
		155
		156	And since the function call protocol is now changed, you need to adopt
		157	C<MyWorker::run> to the async API.
		158
		159	First, you need to accept the extra initial C<$done> callback:
		160
		161	sub run {
		162	my ($done, $cmd, $path) = @_;
		163
		164	And since a response is now generated when C<$done> is called, as opposed
		165	to when the function returns, we need to call the C<$done> function with
		166	the status:
		167
		168	$done->($status or (0, "$!"));
		169
		170	A few remarks are in order. First, it's quite pointless to use the async
		171	backend for this example - but it I<is> possible. Second, you can call
		172	C<$done> before or after returning from the function. Third, having both
		173	returned from the function and having called the C<$done> callback, the
		174	child process may exit at any time, so you should call C<$done> only when
		175	you really I<are> done.
		176
		177	=head2 Example 2: Asynchronous Backend
		178
		179	This example implements multiple count-downs in the child, using
		180	L<AnyEvent> timers. While this is a bit silly (one could use timers in the
		181	parent just as well), it illustrates the ability to use AnyEvent in the
		182	child and the fact that responses can arrive in a different order then the
		183	requests.
		184
		185	It also shows how to embed the actual child code into a C<__DATA__>
		186	section, so it doesn't need any external files at all.
		187
		188	And when your parent process is often busy, and you have stricter timing
		189	requirements, then running timers in a child process suddenly doesn't look
		190	so silly anymore.
		191
		192	Without further ado, here is the code:
		193
		194	use AnyEvent;
		195	use AnyEvent::Fork;
		196	use AnyEvent::Fork::RPC;
		197
		198	my $done = AE::cv;
		199
		200	my $rpc = AnyEvent::Fork
		201	->new
		202	->require ("AnyEvent::Fork::RPC::Async")
		203	->eval (do { local $/; <DATA> })
		204	->AnyEvent::Fork::RPC::run ("run",
		205	async => 1,
		206	on_error => sub { warn "ERROR: $_[0]"; exit 1 },
		207	on_event => sub { print $_[0] },
		208	on_destroy => $done,
		209	);
		210
		211	for my $count (3, 2, 1) {
		212	$rpc->($count, sub {
		213	warn "job $count finished\n";
		214	});
		215	}
		216
		217	undef $rpc;
		218
		219	$done->recv;
		220
		221	__DATA__
		222
		223	# this ends up in main, as we don't use a package declaration
		224
		225	use AnyEvent;
		226
		227	sub run {
		228	my ($done, $count) = @_;
		229
		230	my $n;
		231
		232	AnyEvent::Fork::RPC::event "starting to count up to $count\n";
		233
		234	my $w; $w = AE::timer 1, 1, sub {
		235	++$n;
		236
		237	AnyEvent::Fork::RPC::event "count $n of $count\n";
		238
		239	if ($n == $count) {
		240	undef $w;
		241	$done->();
		242	}
		243	};
		244	}
		245
		246	The parent part (the one before the C<__DATA__> section) isn't very
		247	different from the earlier examples. It sets async mode, preloads
		248	the backend module (so the C<AnyEvent::Fork::RPC::event> function is
		249	declared), uses a slightly different C<on_event> handler (which we use
		250	simply for logging purposes) and then, instead of loading a module with
		251	the actual worker code, it C<eval>'s the code from the data section in the
		252	child process.
		253
		254	It then starts three countdowns, from 3 to 1 seconds downwards, destroys
		255	the rpc object so the example finishes eventually, and then just waits for
		256	the stuff to trickle in.
		257
		258	The worker code uses the event function to log some progress messages, but
		259	mostly just creates a recurring one-second timer.
		260
		261	The timer callback increments a counter, logs a message, and eventually,
		262	when the count has been reached, calls the finish callback.
		263
		264	On my system, this results in the following output. Since all timers fire
		265	at roughly the same time, the actual order isn't guaranteed, but the order
		266	shown is very likely what you would get, too.
		267
		268	starting to count up to 3
		269	starting to count up to 2
		270	starting to count up to 1
		271	count 1 of 3
		272	count 1 of 2
		273	count 1 of 1
		274	job 1 finished
		275	count 2 of 2
		276	job 2 finished
		277	count 2 of 3
		278	count 3 of 3
		279	job 3 finished
		280
		281	While the overall ordering isn't guaranteed, the async backend still
		282	guarantees that events and responses are delivered to the parent process
		283	in the exact same ordering as they were generated in the child process.
		284
		285	And unless your system is I<very> busy, it should clearly show that the
		286	job started last will finish first, as it has the lowest count.
		287
		288	This concludes the async example. Since L<AnyEvent::Fork> does not
		289	actually fork, you are free to use about any module in the child, not just
		290	L<AnyEvent>, but also L<IO::AIO>, or L<Tk> for example.
		291
		292	=head2 Example 3: Asynchronous backend with Coro
		293
		294	With L<Coro> you can create a nice asynchronous backend implementation by
		295	defining an rpc server function that creates a new Coro thread for every
		296	request that calls a function "normally", i.e. the parameters from the
		297	parent process are passed to it, and any return values are returned to the
		298	parent process, e.g.:
		299
		300	package My::Arith;
		301
		302	sub add {
		303	return $_[0] + $_[1];
		304	}
		305
		306	sub mul {
		307	return $_[0] * $_[1];
		308	}
		309
		310	sub run {
		311	my ($done, $func, @arg) = @_;
		312
		313	Coro::async_pool {
		314	$done->($func->(@arg));
		315	};
		316	}
		317
		318	The C<run> function creates a new thread for every invocation, using the
		319	first argument as function name, and calls the C<$done> callback on it's
		320	return values. This makes it quite natural to define the C<add> and C<mul>
		321	functions to add or multiply two numbers and return the result.
		322
		323	Since this is the asynchronous backend, it's quite possible to define RPC
		324	function that do I/O or wait for external events - their execution will
		325	overlap as needed.
		326
		327	The above could be used like this:
		328
		329	my $rpc = AnyEvent::Fork
		330	->new
		331	->require ("MyWorker")
		332	->AnyEvent::Fork::RPC::run ("My::Arith::run",
		333	on_error => ..., on_event => ..., on_destroy => ...,
		334	);
		335
		336	$rpc->(add => 1, 3, Coro::rouse_cb); say Coro::rouse_wait;
		337	$rpc->(mul => 3, 2, Coro::rouse_cb); say Coro::rouse_wait;
		338
		339	The C<say>'s will print C<4> and C<6>.
		340
		341	=head2 Example 4: Forward AnyEvent::Log messages using C<on_event>
		342
		343	This partial example shows how to use the C<event> function to forward
		344	L<AnyEvent::Log> messages to the parent.
		345
		346	For this, the parent needs to provide a suitable C<on_event>:
		347
		348	->AnyEvent::Fork::RPC::run (
		349	on_event => sub {
		350	if ($_[0] eq "ae_log") {
		351	my (undef, $level, $message) = @_;
		352	AE::log $level, $message;
		353	} else {
		354	# other event types
		355	}
		356	},
		357	)
		358
		359	In the child, as early as possible, the following code should reconfigure
		360	L<AnyEvent::Log> to log via C<AnyEvent::Fork::RPC::event>:
		361
		362	$AnyEvent::Log::LOG->log_cb (sub {
		363	my ($timestamp, $orig_ctx, $level, $message) = @{+shift};
		364
		365	if (defined &AnyEvent::Fork::RPC::event) {
		366	AnyEvent::Fork::RPC::event (ae_log => $level, $message);
		367	} else {
		368	warn "[$$ before init] $message\n";
		369	}
		370	});
		371
		372	There is an important twist - the C<AnyEvent::Fork::RPC::event> function
		373	is only defined when the child is fully initialised. If you redirect the
		374	log messages in your C<init> function for example, then the C<event>
		375	function might not yet be available. This is why the log callback checks
		376	whether the function is there using C<defined>, and only then uses it to
		377	log the message.
		378
39	=head1 PARENT PROCESS USAGE	379	=head1 PARENT PROCESS USAGE
40		380
41	This module exports nothing, and only implements a single function:	381	This module exports nothing, and only implements a single function:
42		382
43	=over 4	383	=over 4
…		…
50		390
51	use Errno ();	391	use Errno ();
52	use Guard ();	392	use Guard ();
53		393
54	use AnyEvent;	394	use AnyEvent;
55	#use AnyEvent::Fork;
56		395
57	our $VERSION = 0.1;	396	our $VERSION = 1.25;
58		397
59	=item my $rpc = AnyEvent::Fork::RPC::run $fork, $function, [key => value...]	398	=item my $rpc = AnyEvent::Fork::RPC::run $fork, $function, [key => value...]
60		399
61	The traditional way to call it. But it is way cooler to call it in the	400	The traditional way to call it. But it is way cooler to call it in the
62	following way:	401	following way:
…		…
82	Called on (fatal) errors, with a descriptive (hopefully) message. If	421	Called on (fatal) errors, with a descriptive (hopefully) message. If
83	this callback is not provided, but C<on_event> is, then the C<on_event>	422	this callback is not provided, but C<on_event> is, then the C<on_event>
84	callback is called with the first argument being the string C<error>,	423	callback is called with the first argument being the string C<error>,
85	followed by the error message.	424	followed by the error message.
86		425
87	If neither handler is provided it prints the error to STDERR and will	426	If neither handler is provided, then the error is reported with loglevel
88	start failing badly.	427	C<error> via C<AE::log>.
89		428
90	=item on_event => $cb->(...)	429	=item on_event => $cb->(...)
91		430
92	Called for every call to the C<AnyEvent::Fork::RPC::event> function in the	431	Called for every call to the C<AnyEvent::Fork::RPC::event> function in the
93	child, with the arguments of that function passed to the callback.	432	child, with the arguments of that function passed to the callback.
94		433
95	Also called on errors when no C<on_error> handler is provided.	434	Also called on errors when no C<on_error> handler is provided.
96		435
		436	=item on_destroy => $cb->()
		437
		438	Called when the C<$rpc> object has been destroyed and all requests have
		439	been successfully handled. This is useful when you queue some requests and
		440	want the child to go away after it has handled them. The problem is that
		441	the parent must not exit either until all requests have been handled, and
		442	this can be accomplished by waiting for this callback.
		443
97	=item init => $function (default none)	444	=item init => $function (default: none)
98		445
99	When specified (by name), this function is called in the child as the very	446	When specified (by name), this function is called in the child as the very
100	first thing when taking over the process, with all the arguments normally	447	first thing when taking over the process, with all the arguments normally
101	passed to the C<AnyEvent::Fork::run> function, except the communications	448	passed to the C<AnyEvent::Fork::run> function, except the communications
102	socket.	449	socket.
103		450
104	It can be used to do one-time things in the child such as storing passed	451	It can be used to do one-time things in the child such as storing passed
105	parameters or opening database connections.	452	parameters or opening database connections.
106		453
		454	It is called very early - before the serialisers are created or the
		455	C<$function> name is resolved into a function reference, so it could be
		456	used to load any modules that provide the serialiser or function. It can
		457	not, however, create events.
		458
		459	=item done => $function (default: C<CORE::exit>)
		460
		461	The function to call when the asynchronous backend detects an end of file
		462	condition when reading from the communications socket I<and> there are no
		463	outstanding requests. It's ignored by the synchronous backend.
		464
		465	By overriding this you can prolong the life of a RPC process after e.g.
		466	the parent has exited by running the event loop in the provided function
		467	(or simply calling it, for example, when your child process uses L<EV> you
		468	could provide L<EV::run> as C<done> function).
		469
		470	Of course, in that case you are responsible for exiting at the appropriate
		471	time and not returning from
		472
107	=item async => $boolean (default: 0)	473	=item async => $boolean (default: C<0>)
108		474
109	The default server used in the child does all I/O blockingly, and only	475	The default server used in the child does all I/O blockingly, and only
110	allows a single RPC call to execute concurrently.	476	allows a single RPC call to execute concurrently.
111		477
112	Setting C<async> to a true value switches to another implementation that	478	Setting C<async> to a true value switches to another implementation that
113	uses L<AnyEvent> in the child and allows multiple concurrent RPC calls.	479	uses L<AnyEvent> in the child and allows multiple concurrent RPC calls (it
		480	does not support recursion in the event loop however, blocking condvar
		481	calls will fail).
114		482
115	The actual API in the child is documented in the section that describes	483	The actual API in the child is documented in the section that describes
116	the calling semantics of the returned C<$rpc> function.	484	the calling semantics of the returned C<$rpc> function.
117		485
118	=item serialiser => $string (default: '(sub { pack "(w/a)", @_ }, sub { unpack "(w/a)", shift })')	486	If you want to pre-load the actual back-end modules to enable memory
		487	sharing, then you should load C<AnyEvent::Fork::RPC::Sync> for
		488	synchronous, and C<AnyEvent::Fork::RPC::Async> for asynchronous mode.
		489
		490	If you use a template process and want to fork both sync and async
		491	children, then it is permissible to load both modules.
		492
		493	=item serialiser => $string (default: C<$AnyEvent::Fork::RPC::STRING_SERIALISER>)
119		494
120	All arguments, result data and event data have to be serialised to be	495	All arguments, result data and event data have to be serialised to be
121	transferred between the processes. For this, they have to be frozen and	496	transferred between the processes. For this, they have to be frozen and
122	thawed in both parent and child processes.	497	thawed in both parent and child processes.
123		498
124	By default, only octet strings can be passed between the processes, which	499	By default, only octet strings can be passed between the processes,
125	is reasonably fast and efficient.	500	which is reasonably fast and efficient and requires no extra modules
		501	(the C<AnyEvent::Fork::RPC> distribution does not provide these extra
		502	serialiser modules).
126		503
127	For more complicated use cases, you can provide your own freeze and thaw	504	For more complicated use cases, you can provide your own freeze and thaw
128	functions, by specifying a string with perl source code. It's supposed to	505	functions, by specifying a string with perl source code. It's supposed to
129	return two code references when evaluated: the first receives a list of	506	return two code references when evaluated: the first receives a list of
130	perl values and must return an octet string. The second receives the octet	507	perl values and must return an octet string. The second receives the octet
131	string and must return the original list of values.	508	string and must return the original list of values.
132		509
		510	If you need an external module for serialisation, then you can either
		511	pre-load it into your L<AnyEvent::Fork> process, or you can add a C<use>
		512	or C<require> statement into the serialiser string. Or both.
		513
		514	Here are some examples - all of them are also available as global
		515	variables that make them easier to use.
		516
		517	=over 4
		518
		519	=item C<$AnyEvent::Fork::RPC::STRING_SERIALISER> - octet strings only
		520
		521	This serialiser (currently the default) concatenates length-prefixes octet
		522	strings, and is the default. That means you can only pass (and return)
		523	strings containing character codes 0-255.
		524
		525	The main advantages of this serialiser are the high speed and that it
		526	doesn't need another module. The main disadvantage is that you are very
		527	limited in what you can pass - only octet strings.
		528
		529	Implementation:
		530
		531	(
		532	sub { pack "(w/a)", @_ },
		533	sub { unpack "(w/a)", shift }
		534	)
		535
		536	=item C<$AnyEvent::Fork::RPC::CBOR_XS_SERIALISER> - uses L<CBOR::XS>
		537
		538	This serialiser creates CBOR::XS arrays - you have to make sure the
		539	L<CBOR::XS> module is installed for this serialiser to work. It can be
		540	beneficial for sharing when you preload the L<CBOR::XS> module in a template
		541	process.
		542
		543	L<CBOR::XS> is about as fast as the octet string serialiser, but supports
		544	complex data structures (similar to JSON) and is faster than any of the
		545	other serialisers. If you have the L<CBOR::XS> module available, it's the
		546	best choice.
		547
		548	The encoder enables C<allow_sharing> (so this serialisation method can
		549	encode cyclic and self-referencing data structures).
		550
		551	Implementation:
		552
		553	use CBOR::XS ();
		554	(
		555	sub { CBOR::XS::encode_cbor_sharing \@_ },
		556	sub { @{ CBOR::XS::decode_cbor shift } }
		557	)
		558
		559	=item C<$AnyEvent::Fork::RPC::JSON_SERIALISER> - uses L<JSON::XS> or L<JSON>
		560
		561	This serialiser creates JSON arrays - you have to make sure the L<JSON>
		562	module is installed for this serialiser to work. It can be beneficial for
		563	sharing when you preload the L<JSON> module in a template process.
		564
		565	L<JSON> (with L<JSON::XS> installed) is slower than the octet string
		566	serialiser, but usually much faster than L<Storable>, unless big chunks of
		567	binary data need to be transferred.
		568
		569	Implementation:
		570
		571	use JSON ();
		572	(
		573	sub { JSON::encode_json \@_ },
		574	sub { @{ JSON::decode_json shift } }
		575	)
		576
		577	=item C<$AnyEvent::Fork::RPC::STORABLE_SERIALISER> - L<Storable>
		578
		579	This serialiser uses L<Storable>, which means it has high chance of
		580	serialising just about anything you throw at it, at the cost of having
		581	very high overhead per operation. It also comes with perl. It should be
		582	used when you need to serialise complex data structures.
		583
		584	Implementation:
		585
		586	use Storable ();
		587	(
		588	sub { Storable::freeze \@_ },
		589	sub { @{ Storable::thaw shift } }
		590	)
		591
		592	=item C<$AnyEvent::Fork::RPC::NSTORABLE_SERIALISER> - portable Storable
		593
		594	This serialiser also uses L<Storable>, but uses it's "network" format
		595	to serialise data, which makes it possible to talk to different
		596	perl binaries (for example, when talking to a process created with
		597	L<AnyEvent::Fork::Remote>).
		598
		599	Implementation:
		600
		601	use Storable ();
		602	(
		603	sub { Storable::nfreeze \@_ },
		604	sub { @{ Storable::thaw shift } }
		605	)
		606
133	=back	607	=back
134		608
		609	=item buflen => $bytes (default: C<512 - 16>)
		610
		611	The starting size of the read buffer for request and response data.
		612
		613	C<AnyEvent::Fork::RPC> ensures that the buffer for reeading request and
		614	response data is large enough for at leats aingle request or response, and
		615	will dynamically enlarge the buffer if needed.
		616
		617	While this ensures that memory is not overly wasted, it typically leads
		618	to having to do one syscall per request, which can be inefficient in some
		619	cases. In such cases, it can be beneficient to increase the buffer size to
		620	hold more than one request.
		621
		622	=item buflen_req => $bytes (default: same as C<buflen>)
		623
		624	Overrides C<buflen> for request data (as read by the forked process).
		625
		626	=item buflen_res => $bytes (default: same as C<buflen>)
		627
		628	Overrides C<buflen> for response data (replies read by the parent process).
		629
		630	=back
		631
		632	See the examples section earlier in this document for some actual
		633	examples.
		634
135	=cut	635	=cut
136		636
137	our $SERIALISE_STRINGS = '(sub { pack "(w/a)", @_ }, sub { unpack "(w/a)", shift })';	637	our $STRING_SERIALISER = '(sub { pack "(w/a)", @_ }, sub { unpack "(w/a)", shift })';
		638	our $CBOR_XS_SERIALISER = 'use CBOR::XS (); (sub { CBOR::XS::encode_cbor_sharing \@_ }, sub { @{ CBOR::XS::decode_cbor shift } })';
		639	our $JSON_SERIALISER = 'use JSON (); (sub { JSON::encode_json \@_ }, sub { @{ JSON::decode_json shift } })';
		640	our $STORABLE_SERIALISER = 'use Storable (); (sub { Storable::freeze \@_ }, sub { @{ Storable::thaw shift } })';
		641	our $NSTORABLE_SERIALISER = 'use Storable (); (sub { Storable::nfreeze \@_ }, sub { @{ Storable::thaw shift } })';
138		642
139	sub run {	643	sub run {
140	my ($self, $function, %arg) = @_;	644	my ($self, $function, %arg) = @_;
141		645
142	my $serialiser = delete $arg{serialiser} \|\| $SERIALISE_STRINGS;	646	my $serialiser = delete $arg{serialiser} \|\| $STRING_SERIALISER;
143	my $on_event = delete $arg{on_event};	647	my $on_event = delete $arg{on_event};
144	my $on_error = delete $arg{on_error};	648	my $on_error = delete $arg{on_error};
		649	my $on_destroy = delete $arg{on_destroy};
145		650
146	# default for on_error is to on_event, if specified	651	# default for on_error is to on_event, if specified
147	$on_error \|\|= $on_event	652	$on_error \|\|= $on_event
148	? sub { $on_event->(error => shift) }	653	? sub { $on_event->(error => shift) }
149	: sub { die "AnyEvent::Fork::RPC: uncaught error: $_[0].\n" };	654	: sub { AE::log die => "AnyEvent::Fork::RPC: uncaught error: $_[0]." };
150		655
151	# default for on_event is to raise an error	656	# default for on_event is to raise an error
152	$on_event \|\|= sub { $on_error->("event received, but no on_event handler") };	657	$on_event \|\|= sub { $on_error->("event received, but no on_event handler") };
153		658
154	my ($f, $t) = eval $serialiser; die $@ if $@;	659	my ($f, $t) = eval $serialiser; die $@ if $@;
155		660
156	my (@rcb, $fh, $shutdown, $wbuf, $ww, $rbuf, $rw);	661	my (@rcb, %rcb, $fh, $shutdown, $wbuf, $ww);
		662	my ($rlen, $rbuf, $rw) = $arg{buflen_res} \|\| $arg{buflen} \|\| 512 - 16;
157		663
158	my $wcb = sub {	664	my $wcb = sub {
159	my $len = syswrite $fh, $wbuf;	665	my $len = syswrite $fh, $wbuf;
160		666
161	if (!defined $len) {	667	unless (defined $len) {
162	if ($! != Errno::EAGAIN && $! != Errno::EWOULDBLOCK) {	668	if ($! != Errno::EAGAIN && $! != Errno::EWOULDBLOCK) {
163	undef $rw; undef $ww; # it ends here	669	undef $rw; undef $ww; # it ends here
164	$on_error->("$!");	670	$on_error->("$!");
165	}	671	}
166	}	672	}
…		…
173	}	679	}
174	};	680	};
175		681
176	my $module = "AnyEvent::Fork::RPC::" . ($arg{async} ? "Async" : "Sync");	682	my $module = "AnyEvent::Fork::RPC::" . ($arg{async} ? "Async" : "Sync");
177		683
178	$self->require ($module)	684	$self->eval ("use $module 2 ()")
179	->send_arg ($function, $arg{init}, $serialiser)	685	->send_arg (
		686	function => $function,
		687	init => $arg{init},
		688	serialiser => $serialiser,
		689	done => $arg{done} \|\| "$module\::do_exit",
		690	rlen => $arg{buflen_req} \|\| $arg{buflen} \|\| 512 - 16,
		691	-10 # the above are 10 arguments
		692	)
180	->run ("$module\::run", sub {	693	->run ("$module\::run", sub {
181	$fh = shift;	694	$fh = shift
		695	or return $on_error->("connection failed");
		696
		697	my ($id, $len);
182	$rw = AE::io $fh, 0, sub {	698	$rw = AE::io $fh, 0, sub {
		699	$rlen = $rlen * 2 + 16 if $rlen - 128 < length $rbuf;
183	my $len = sysread $fh, $rbuf, 512 + length $rbuf, length $rbuf;	700	$len = sysread $fh, $rbuf, $rlen - length $rbuf, length $rbuf;
184		701
185	if ($len) {	702	if ($len) {
186	while (5 <= length $rbuf) {	703	while (8 <= length $rbuf) {
187	$len = unpack "L", $rbuf;	704	($id, $len) = unpack "NN", $rbuf;
188	if (4 + $len <= length $rbuf) {	705	8 + $len <= length $rbuf
		706	or last;
		707
189	my @r = $t->(substr $rbuf, 4, $len);	708	my @r = $t->(substr $rbuf, 8, $len);
190	substr $rbuf, 0, $len + 4, "";	709	substr $rbuf, 0, 8 + $len, "";
191		710
192	if (pop @r) {	711	if ($id) {
193	$on_event->(@r);
194	} elsif (@rcb) {	712	if (@rcb) {
195	(shift @rcb)->(@r);	713	(shift @rcb)->(@r);
		714	} elsif (my $cb = delete $rcb{$id}) {
		715	$cb->(@r);
196	} else {	716	} else {
197	undef $rw; undef $ww;	717	undef $rw; undef $ww;
198	$on_error->("unexpected data from child");	718	$on_error->("unexpected data from child");
199	}	719	}
		720	} else {
		721	$on_event->(@r);
200	}	722	}
201	}	723	}
202	} elsif (defined $len) {	724	} elsif (defined $len) {
203	undef $rw; undef $ww; # it ends here	725	undef $rw; undef $ww; # it ends here
		726
		727	if (@rcb \|\| %rcb) {
204	$on_error->("unexpected eof")	728	$on_error->("unexpected eof");
205	if @rcb;	729	} else {
		730	$on_destroy->()
		731	if $on_destroy;
		732	}
206	} elsif ($! != Errno::EAGAIN && $! != Errno::EWOULDBLOCK) {	733	} elsif ($! != Errno::EAGAIN && $! != Errno::EWOULDBLOCK) {
207	undef $rw; undef $ww; # it ends here	734	undef $rw; undef $ww; # it ends here
208	$on_error->("read: $!");	735	$on_error->("read: $!");
209	}	736	}
210	};	737	};
…		…
212	$ww \|\|= AE::io $fh, 1, $wcb;	739	$ww \|\|= AE::io $fh, 1, $wcb;
213	});	740	});
214		741
215	my $guard = Guard::guard {	742	my $guard = Guard::guard {
216	$shutdown = 1;	743	$shutdown = 1;
217	$ww \|\|= $fh && AE::io $fh, 1, $wcb;	744
		745	shutdown $fh, 1 if $fh && !$ww;
218	};	746	};
219		747
		748	my $id;
		749
		750	$arg{async}
220	sub {	751	? sub {
221	push @rcb, pop;	752	$id = ($id == 0xffffffff ? 0 : $id) + 1;
		753	$id = ($id == 0xffffffff ? 0 : $id) + 1 while exists $rcb{$id}; # rarely loops
222		754
		755	$rcb{$id} = pop;
		756
223	$guard; # keep it alive	757	$guard if 0; # keep it alive
224		758
225	$wbuf .= pack "L/a*", &$f;	759	$wbuf .= pack "NN/a*", $id, &$f;
226	$ww \|\|= $fh && AE::io $fh, 1, $wcb;	760	$ww \|\|= $fh && AE::io $fh, 1, $wcb;
227	}	761	}
		762	: sub {
		763	push @rcb, pop;
		764
		765	$guard; # keep it alive
		766
		767	$wbuf .= pack "N/a*", &$f;
		768	$ww \|\|= $fh && AE::io $fh, 1, $wcb;
		769	}
228	}	770	}
229		771
		772	=item $rpc->(..., $cb->(...))
		773
		774	The RPC object returned by C<AnyEvent::Fork::RPC::run> is actually a code
		775	reference. There are two things you can do with it: call it, and let it go
		776	out of scope (let it get destroyed).
		777
		778	If C<async> was false when C<$rpc> was created (the default), then, if you
		779	call C<$rpc>, the C<$function> is invoked with all arguments passed to
		780	C<$rpc> except the last one (the callback). When the function returns, the
		781	callback will be invoked with all the return values.
		782
		783	If C<async> was true, then the C<$function> receives an additional
		784	initial argument, the result callback. In this case, returning from
		785	C<$function> does nothing - the function only counts as "done" when the
		786	result callback is called, and any arguments passed to it are considered
		787	the return values. This makes it possible to "return" from event handlers
		788	or e.g. Coro threads.
		789
		790	The other thing that can be done with the RPC object is to destroy it. In
		791	this case, the child process will execute all remaining RPC calls, report
		792	their results, and then exit.
		793
		794	See the examples section earlier in this document for some actual
		795	examples.
		796
230	=back	797	=back
231		798
232	=head1 CHILD PROCESS USAGE	799	=head1 CHILD PROCESS USAGE
233		800
234	These functions are not available in this module. They are only available	801	The following function is not available in this module. They are only
235	in the namespace of this module when the child is running, without	802	available in the namespace of this module when the child is running,
236	having to load any extra module. They are part of the child-side API of	803	without having to load any extra modules. They are part of the child-side
237	L<AnyEvent::Fork::RPC>.	804	API of L<AnyEvent::Fork::RPC>.
238		805
239	=over 4	806	=over 4
240
241	=item AnyEvent::Fork::RPC::quit
242
243	This function can be called to gracefully stop the child process when it
244	is idle.
245
246	After this function is called, the process stops handling incoming RPC
247	requests, but outstanding events and function return values will be sent
248	to the parent. When all data has been sent, the process calls C<exit>.
249
250	Since the parent might not expect the child to exit at random points in
251	time, it is often better to signal the parent by sending an C<event> and
252	letting the parent close down the child process.
253		807
254	=item AnyEvent::Fork::RPC::event ...	808	=item AnyEvent::Fork::RPC::event ...
255		809
256	Send an event to the parent. Events are a bit like RPC calls made by the	810	Send an event to the parent. Events are a bit like RPC calls made by the
257	child process to the parent, except that there is no notion of return	811	child process to the parent, except that there is no notion of return
258	values.	812	values.
259		813
		814	See the examples section earlier in this document for some actual
		815	examples.
		816
		817	Note: the event data, like any data send to the parent, might not be sent
		818	immediatelly but queued for later sending, so there is no guarantee that
		819	the event has been sent to the parent when the call returns - when you
		820	e.g. exit directly after calling this function, the parent might never
		821	receive the event.
		822
260	=back	823	=back
261		824
		825	=head2 PROCESS EXIT
		826
		827	If and when the child process exits depends on the backend and
		828	configuration. Apart from explicit exits (e.g. by calling C<exit>) or
		829	runtime conditions (uncaught exceptions, signals etc.), the backends exit
		830	under these conditions:
		831
		832	=over 4
		833
		834	=item Synchronous Backend
		835
		836	The synchronous backend is very simple: when the process waits for another
		837	request to arrive and the writing side (usually in the parent) is closed,
		838	it will exit normally, i.e. as if your main program reached the end of the
		839	file.
		840
		841	That means that if your parent process exits, the RPC process will usually
		842	exit as well, either because it is idle anyway, or because it executes a
		843	request. In the latter case, you will likely get an error when the RPc
		844	process tries to send the results to the parent (because agruably, you
		845	shouldn't exit your parent while there are still outstanding requests).
		846
		847	The process is usually quiescent when it happens, so it should rarely be a
		848	problem, and C<END> handlers can be used to clean up.
		849
		850	=item Asynchronous Backend
		851
		852	For the asynchronous backend, things are more complicated: Whenever it
		853	listens for another request by the parent, it might detect that the socket
		854	was closed (e.g. because the parent exited). It will sotp listening for
		855	new requests and instead try to write out any remaining data (if any) or
		856	simply check whether the socket can be written to. After this, the RPC
		857	process is effectively done - no new requests are incoming, no outstanding
		858	request data can be written back.
		859
		860	Since chances are high that there are event watchers that the RPC server
		861	knows nothing about (why else would one use the async backend if not for
		862	the ability to register watchers?), the event loop would often happily
		863	continue.
		864
		865	This is why the asynchronous backend explicitly calls C<CORE::exit> when
		866	it is done (under other circumstances, such as when there is an I/O error
		867	and there is outstanding data to write, it will log a fatal message via
		868	L<AnyEvent::Log>, also causing the program to exit).
		869
		870	You can override this by specifying a function name to call via the C<done>
		871	parameter instead.
		872
		873	=back
		874
		875	=head1 ADVANCED TOPICS
		876
		877	=head2 Choosing a backend
		878
		879	So how do you decide which backend to use? Well, that's your problem to
		880	solve, but here are some thoughts on the matter:
		881
		882	=over 4
		883
		884	=item Synchronous
		885
		886	The synchronous backend does not rely on any external modules (well,
		887	except L<common::sense>, which works around a bug in how perl's warning
		888	system works). This keeps the process very small, for example, on my
		889	system, an empty perl interpreter uses 1492kB RSS, which becomes 2020kB
		890	after C<use warnings; use strict> (for people who grew up with C64s around
		891	them this is probably shocking every single time they see it). The worker
		892	process in the first example in this document uses 1792kB.
		893
		894	Since the calls are done synchronously, slow jobs will keep newer jobs
		895	from executing.
		896
		897	The synchronous backend also has no overhead due to running an event loop
		898	- reading requests is therefore very efficient, while writing responses is
		899	less so, as every response results in a write syscall.
		900
		901	If the parent process is busy and a bit slow reading responses, the child
		902	waits instead of processing further requests. This also limits the amount
		903	of memory needed for buffering, as never more than one response has to be
		904	buffered.
		905
		906	The API in the child is simple - you just have to define a function that
		907	does something and returns something.
		908
		909	It's hard to use modules or code that relies on an event loop, as the
		910	child cannot execute anything while it waits for more input.
		911
		912	=item Asynchronous
		913
		914	The asynchronous backend relies on L<AnyEvent>, which tries to be small,
		915	but still comes at a price: On my system, the worker from example 1a uses
		916	3420kB RSS (for L<AnyEvent>, which loads L<EV>, which needs L<XSLoader>
		917	which in turn loads a lot of other modules such as L<warnings>, L<strict>,
		918	L<vars>, L<Exporter>...).
		919
		920	It batches requests and responses reasonably efficiently, doing only as
		921	few reads and writes as needed, but needs to poll for events via the event
		922	loop.
		923
		924	Responses are queued when the parent process is busy. This means the child
		925	can continue to execute any queued requests. It also means that a child
		926	might queue a lot of responses in memory when it generates them and the
		927	parent process is slow accepting them.
		928
		929	The API is not a straightforward RPC pattern - you have to call a
		930	"done" callback to pass return values and signal completion. Also, more
		931	importantly, the API starts jobs as fast as possible - when 1000 jobs
		932	are queued and the jobs are slow, they will all run concurrently. The
		933	child must implement some queueing/limiting mechanism if this causes
		934	problems. Alternatively, the parent could limit the amount of rpc calls
		935	that are outstanding.
		936
		937	Blocking use of condvars is not supported (in the main thread, outside of
		938	e.g. L<Coro> threads).
		939
		940	Using event-based modules such as L<IO::AIO>, L<Gtk2>, L<Tk> and so on is
		941	easy.
		942
		943	=back
		944
		945	=head2 Passing file descriptors
		946
		947	Unlike L<AnyEvent::Fork>, this module has no in-built file handle or file
		948	descriptor passing abilities.
		949
		950	The reason is that passing file descriptors is extraordinary tricky
		951	business, and conflicts with efficient batching of messages.
		952
		953	There still is a method you can use: Create a
		954	C<AnyEvent::Util::portable_socketpair> and C<send_fh> one half of it to
		955	the process before you pass control to C<AnyEvent::Fork::RPC::run>.
		956
		957	Whenever you want to pass a file descriptor, send an rpc request to the
		958	child process (so it expects the descriptor), then send it over the other
		959	half of the socketpair. The child should fetch the descriptor from the
		960	half it has passed earlier.
		961
		962	Here is some (untested) pseudocode to that effect:
		963
		964	use AnyEvent::Util;
		965	use AnyEvent::Fork;
		966	use AnyEvent::Fork::RPC;
		967	use IO::FDPass;
		968
		969	my ($s1, $s2) = AnyEvent::Util::portable_socketpair;
		970
		971	my $rpc = AnyEvent::Fork
		972	->new
		973	->send_fh ($s2)
		974	->require ("MyWorker")
		975	->AnyEvent::Fork::RPC::run ("MyWorker::run"
		976	init => "MyWorker::init",
		977	);
		978
		979	undef $s2; # no need to keep it around
		980
		981	# pass an fd
		982	$rpc->("i'll send some fd now, please expect it!", my $cv = AE::cv);
		983
		984	IO::FDPass fileno $s1, fileno $handle_to_pass;
		985
		986	$cv->recv;
		987
		988	The MyWorker module could look like this:
		989
		990	package MyWorker;
		991
		992	use IO::FDPass;
		993
		994	my $s2;
		995
		996	sub init {
		997	$s2 = $_[0];
		998	}
		999
		1000	sub run {
		1001	if ($_[0] eq "i'll send some fd now, please expect it!") {
		1002	my $fd = IO::FDPass::recv fileno $s2;
		1003	...
		1004	}
		1005	}
		1006
		1007	Of course, this might be blocking if you pass a lot of file descriptors,
		1008	so you might want to look into L<AnyEvent::FDpasser> which can handle the
		1009	gory details.
		1010
		1011	=head1 EXCEPTIONS
		1012
		1013	There are no provisions whatsoever for catching exceptions at this time -
		1014	in the child, exceptions might kill the process, causing calls to be lost
		1015	and the parent encountering a fatal error. In the parent, exceptions in
		1016	the result callback will not be caught and cause undefined behaviour.
		1017
262	=head1 SEE ALSO	1018	=head1 SEE ALSO
263		1019
264	L<AnyEvent::Fork> (to create the processes in the first place),	1020	L<AnyEvent::Fork>, to create the processes in the first place.
		1021
		1022	L<AnyEvent::Fork::Remote>, likewise, but helpful for remote processes.
		1023
265	L<AnyEvent::Fork::Pool> (to manage whole pools of processes).	1024	L<AnyEvent::Fork::Pool>, to manage whole pools of processes.
266		1025
267	=head1 AUTHOR AND CONTACT INFORMATION	1026	=head1 AUTHOR AND CONTACT INFORMATION
268		1027
269	Marc Lehmann <schmorp@schmorp.de>	1028	Marc Lehmann <schmorp@schmorp.de>
270	http://software.schmorp.de/pkg/AnyEvent-Fork-RPC	1029	http://software.schmorp.de/pkg/AnyEvent-Fork-RPC

Diff Legend

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

Comparing AnyEvent-Fork-RPC/RPC.pm (file contents): Revision 1.1 by root, Wed Apr 17 15:55:59 2013 UTC vs. Revision 1.45 by root, Thu Sep 12 15:15:49 2019 UTC

Diff Legend

Comparing AnyEvent-Fork-RPC/RPC.pm (file contents):
Revision 1.1 by root, Wed Apr 17 15:55:59 2013 UTC vs.
Revision 1.45 by root, Thu Sep 12 15:15:49 2019 UTC