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

Comparing AnyEvent-Fork-RPC/RPC.pm (file contents):
Revision 1.16 by root, Thu Apr 18 14:07:15 2013 UTC vs.
Revision 1.31 by root, Sat Aug 31 16:35:33 2013 UTC

2 2
3AnyEvent::Fork::RPC - simple RPC extension for AnyEvent::Fork 3AnyEvent::Fork::RPC - simple RPC extension for AnyEvent::Fork
4 4
5=head1 SYNOPSIS 5=head1 SYNOPSIS
6 6
7 use AnyEvent::Fork;
7 use AnyEvent::Fork::RPC; 8 use AnyEvent::Fork::RPC;
8 # use AnyEvent::Fork is not needed
9 9
10 my $rpc = AnyEvent::Fork 10 my $rpc = AnyEvent::Fork
11 ->new 11 ->new
12 ->require ("MyModule") 12 ->require ("MyModule")
13 ->AnyEvent::Fork::RPC::run ( 13 ->AnyEvent::Fork::RPC::run (
26 $cv->recv; 26 $cv->recv;
27 27
28=head1 DESCRIPTION 28=head1 DESCRIPTION
29 29
30This module implements a simple RPC protocol and backend for processes 30This module implements a simple RPC protocol and backend for processes
31created via L<AnyEvent::Fork>, allowing you to call a function in the 31created via L<AnyEvent::Fork> or L<AnyEvent::Fork::Remote>, allowing you
32child process and receive its return values (up to 4GB serialised). 32to call a function in the child process and receive its return values (up
33to 4GB serialised).
33 34
34It implements two different backends: a synchronous one that works like a 35It implements two different backends: a synchronous one that works like a
35normal function call, and an asynchronous one that can run multiple jobs 36normal function call, and an asynchronous one that can run multiple jobs
36concurrently in the child, using AnyEvent. 37concurrently in the child, using AnyEvent.
37 38
38It also implements an asynchronous event mechanism from the child to the 39It also implements an asynchronous event mechanism from the child to the
39parent, that could be used for progress indications or other information. 40parent, that could be used for progress indications or other information.
40
41Loading this module also always loads L<AnyEvent::Fork>, so you can make a
42separate C<use AnyEvent::Fork> if you wish, but you don't have to.
43 41
44=head1 EXAMPLES 42=head1 EXAMPLES
45 43
46=head2 Example 1: Synchronous Backend 44=head2 Example 1: Synchronous Backend
47 45
51silly, but illustrates the use of events. 49silly, but illustrates the use of events.
52 50
53First the parent process: 51First the parent process:
54 52
55 use AnyEvent; 53 use AnyEvent;
54 use AnyEvent::Fork;
56 use AnyEvent::Fork::RPC; 55 use AnyEvent::Fork::RPC;
57 56
58 my $done = AE::cv; 57 my $done = AE::cv;
59 58
60 my $rpc = AnyEvent::Fork 59 my $rpc = AnyEvent::Fork
61 ->new 60 ->new
62 ->require ("MyWorker") 61 ->require ("MyWorker")
63 ->AnyEvent::Fork::RPC::run ("MyWorker::run", 62 ->AnyEvent::Fork::RPC::run ("MyWorker::run",
64 on_error => sub { warn "FATAL: $_[0]"; exit 1 }, 63 on_error => sub { warn "ERROR: $_[0]"; exit 1 },
65 on_event => sub { warn "$_[0] requests handled\n" }, 64 on_event => sub { warn "$_[0] requests handled\n" },
66 on_destroy => $done, 65 on_destroy => $done,
67 ); 66 );
68 67
69 for my $id (1..6) { 68 for my $id (1..6) {
191so silly anymore. 190so silly anymore.
192 191
193Without further ado, here is the code: 192Without further ado, here is the code:
194 193
195 use AnyEvent; 194 use AnyEvent;
195 use AnyEvent::Fork;
196 use AnyEvent::Fork::RPC; 196 use AnyEvent::Fork::RPC;
197 197
198 my $done = AE::cv; 198 my $done = AE::cv;
199 199
200 my $rpc = AnyEvent::Fork 200 my $rpc = AnyEvent::Fork
201 ->new 201 ->new
202 ->require ("AnyEvent::Fork::RPC::Async") 202 ->require ("AnyEvent::Fork::RPC::Async")
203 ->eval (do { local $/; <DATA> }) 203 ->eval (do { local $/; <DATA> })
204 ->AnyEvent::Fork::RPC::run ("run", 204 ->AnyEvent::Fork::RPC::run ("run",
205 async => 1, 205 async => 1,
206 on_error => sub { warn "FATAL: $_[0]"; exit 1 }, 206 on_error => sub { warn "ERROR: $_[0]"; exit 1 },
207 on_event => sub { print $_[0] }, 207 on_event => sub { print $_[0] },
208 on_destroy => $done, 208 on_destroy => $done,
209 ); 209 );
210 210
211 for my $count (3, 2, 1) { 211 for my $count (3, 2, 1) {
287 287
288This concludes the async example. Since L<AnyEvent::Fork> does not 288This concludes the async example. Since L<AnyEvent::Fork> does not
289actually fork, you are free to use about any module in the child, not just 289actually fork, you are free to use about any module in the child, not just
290L<AnyEvent>, but also L<IO::AIO>, or L<Tk> for example. 290L<AnyEvent>, but also L<IO::AIO>, or L<Tk> for example.
291 291
292=head2 Example 3: Asynchronous backend with Coro
293
294With L<Coro> you can create a nice asynchronous backend implementation by
295defining an rpc server function that creates a new Coro thread for every
296request that calls a function "normally", i.e. the parameters from the
297parent process are passed to it, and any return values are returned to the
298parent 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
318The C<run> function creates a new thread for every invocation, using the
319first argument as function name, and calls the C<$done> callback on it's
320return values. This makes it quite natural to define the C<add> and C<mul>
321functions to add or multiply two numbers and return the result.
322
323Since this is the asynchronous backend, it's quite possible to define RPC
324function that do I/O or wait for external events - their execution will
325overlap as needed.
326
327The 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 $rpc->(mul => 3, 2, Coro::rouse_cb); say Coro::rouse_wait;
338
339The C<say>'s will print C<4> and C<6>.
340
341=head2 Example 4: Forward AnyEvent::Log messages using C<on_event>
342
343This partial example shows how to use the C<event> function to forward
344L<AnyEvent::Log> messages to the parent.
345
346For this, the parent needs to provide a suitable C<on_event>:
347
348 ->AnyEvent::Fork::RPC::run (
349 on_event => sub {
350 if ($_[0] eq "ae_log") {
351 my (undef, $level, $message) = @_;
352 AE::log $level, $message;
353 } else {
354 # other event types
355 }
356 },
357 )
358
359In the child, as early as possible, the following code should reconfigure
360L<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
372There is an important twist - the C<AnyEvent::Fork::RPC::event> function
373is only defined when the child is fully initialised. If you redirect the
374log messages in your C<init> function for example, then the C<event>
375function might not yet be available. This is why the log callback checks
376whether the fucntion is there using C<defined>, and only then uses it to
377log the message.
378
292=head1 PARENT PROCESS USAGE 379=head1 PARENT PROCESS USAGE
293 380
294This module exports nothing, and only implements a single function: 381This module exports nothing, and only implements a single function:
295 382
296=over 4 383=over 4
303 390
304use Errno (); 391use Errno ();
305use Guard (); 392use Guard ();
306 393
307use AnyEvent; 394use AnyEvent;
308use AnyEvent::Fork; # we don't actually depend on it, this is for convenience
309 395
310our $VERSION = 0.1; 396our $VERSION = 1.1;
311 397
312=item my $rpc = AnyEvent::Fork::RPC::run $fork, $function, [key => value...] 398=item my $rpc = AnyEvent::Fork::RPC::run $fork, $function, [key => value...]
313 399
314The traditional way to call it. But it is way cooler to call it in the 400The traditional way to call it. But it is way cooler to call it in the
315following way: 401following way:
335Called on (fatal) errors, with a descriptive (hopefully) message. If 421Called on (fatal) errors, with a descriptive (hopefully) message. If
336this callback is not provided, but C<on_event> is, then the C<on_event> 422this callback is not provided, but C<on_event> is, then the C<on_event>
337callback is called with the first argument being the string C<error>, 423callback is called with the first argument being the string C<error>,
338followed by the error message. 424followed by the error message.
339 425
340If neither handler is provided it prints the error to STDERR and will 426If neither handler is provided, then the error is reported with loglevel
341start failing badly. 427C<error> via C<AE::log>.
342 428
343=item on_event => $cb->(...) 429=item on_event => $cb->(...)
344 430
345Called for every call to the C<AnyEvent::Fork::RPC::event> function in the 431Called for every call to the C<AnyEvent::Fork::RPC::event> function in the
346child, with the arguments of that function passed to the callback. 432child, with the arguments of that function passed to the callback.
368It is called very early - before the serialisers are created or the 454It is called very early - before the serialisers are created or the
369C<$function> name is resolved into a function reference, so it could be 455C<$function> name is resolved into a function reference, so it could be
370used to load any modules that provide the serialiser or function. It can 456used to load any modules that provide the serialiser or function. It can
371not, however, create events. 457not, however, create events.
372 458
459=item done => $function (default C<CORE::exit>)
460
461The function to call when the asynchronous backend detects an end of file
462condition when reading from the communications socket I<and> there are no
463outstanding requests. It's ignored by the synchronous backend.
464
465By overriding this you can prolong the life of a RPC process after e.g.
466the 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
468could provide L<EV::loop> as C<done> function).
469
470Of course, in that case you are responsible for exiting at the appropriate
471time and not returning from
472
373=item async => $boolean (default: 0) 473=item async => $boolean (default: 0)
374 474
375The default server used in the child does all I/O blockingly, and only 475The default server used in the child does all I/O blockingly, and only
376allows a single RPC call to execute concurrently. 476allows a single RPC call to execute concurrently.
377 477
415=over 4 515=over 4
416 516
417=item octet strings - C<$AnyEvent::Fork::RPC::STRING_SERIALISER> 517=item octet strings - C<$AnyEvent::Fork::RPC::STRING_SERIALISER>
418 518
419This serialiser concatenates length-prefixes octet strings, and is the 519This serialiser concatenates length-prefixes octet strings, and is the
420default. 520default. That means you can only pass (and return) strings containing
521character codes 0-255.
421 522
422Implementation: 523Implementation:
423 524
424 ( 525 (
425 sub { pack "(w/a*)*", @_ }, 526 sub { pack "(w/a*)*", @_ },
446 547
447=item storable - C<$AnyEvent::Fork::RPC::STORABLE_SERIALISER> 548=item storable - C<$AnyEvent::Fork::RPC::STORABLE_SERIALISER>
448 549
449This serialiser uses L<Storable>, which means it has high chance of 550This serialiser uses L<Storable>, which means it has high chance of
450serialising just about anything you throw at it, at the cost of having 551serialising just about anything you throw at it, at the cost of having
451very high overhead per operation. It also comes with perl. 552very high overhead per operation. It also comes with perl. It should be
553used when you need to serialise complex data structures.
452 554
453Implementation: 555Implementation:
454 556
455 use Storable (); 557 use Storable ();
456 ( 558 (
457 sub { Storable::freeze \@_ }, 559 sub { Storable::freeze \@_ },
458 sub { @{ Storable::thaw shift } } 560 sub { @{ Storable::thaw shift } }
459 ) 561 )
460 562
563=item portable storable - C<$AnyEvent::Fork::RPC::NSTORABLE_SERIALISER>
564
565This serialiser also uses L<Storable>, but uses it's "network" format
566to serialise data, which makes it possible to talk to different
567perl binaries (for example, when talking to a process created with
568L<AnyEvent::Fork::Remote>).
569
570Implementation:
571
572 use Storable ();
573 (
574 sub { Storable::nfreeze \@_ },
575 sub { @{ Storable::thaw shift } }
576 )
577
461=back 578=back
462 579
463=back 580=back
464 581
465See the examples section earlier in this document for some actual 582See the examples section earlier in this document for some actual
466examples. 583examples.
467 584
468=cut 585=cut
469 586
470our $STRING_SERIALISER = '(sub { pack "(w/a*)*", @_ }, sub { unpack "(w/a*)*", shift })'; 587our $STRING_SERIALISER = '(sub { pack "(w/a*)*", @_ }, sub { unpack "(w/a*)*", shift })';
471our $JSON_SERIALISER = 'use JSON (); (sub { JSON::encode_json \@_ }, sub { @{ JSON::decode_json shift } })'; 588our $JSON_SERIALISER = 'use JSON (); (sub { JSON::encode_json \@_ }, sub { @{ JSON::decode_json shift } })';
472our $STORABLE_SERIALISER = 'use Storable (); (sub { Storable::freeze \@_ }, sub { @{ Storable::thaw shift } })'; 589our $STORABLE_SERIALISER = 'use Storable (); (sub { Storable::freeze \@_ }, sub { @{ Storable::thaw shift } })';
590our $NSTORABLE_SERIALISER = 'use Storable (); (sub { Storable::nfreeze \@_ }, sub { @{ Storable::thaw shift } })';
473 591
474sub run { 592sub run {
475 my ($self, $function, %arg) = @_; 593 my ($self, $function, %arg) = @_;
476 594
477 my $serialiser = delete $arg{serialiser} || $STRING_SERIALISER; 595 my $serialiser = delete $arg{serialiser} || $STRING_SERIALISER;
480 my $on_destroy = delete $arg{on_destroy}; 598 my $on_destroy = delete $arg{on_destroy};
481 599
482 # default for on_error is to on_event, if specified 600 # default for on_error is to on_event, if specified
483 $on_error ||= $on_event 601 $on_error ||= $on_event
484 ? sub { $on_event->(error => shift) } 602 ? sub { $on_event->(error => shift) }
485 : sub { die "AnyEvent::Fork::RPC: uncaught error: $_[0].\n" }; 603 : sub { AE::log die => "AnyEvent::Fork::RPC: uncaught error: $_[0]." };
486 604
487 # default for on_event is to raise an error 605 # default for on_event is to raise an error
488 $on_event ||= sub { $on_error->("event received, but no on_event handler") }; 606 $on_event ||= sub { $on_error->("event received, but no on_event handler") };
489 607
490 my ($f, $t) = eval $serialiser; die $@ if $@; 608 my ($f, $t) = eval $serialiser; die $@ if $@;
511 }; 629 };
512 630
513 my $module = "AnyEvent::Fork::RPC::" . ($arg{async} ? "Async" : "Sync"); 631 my $module = "AnyEvent::Fork::RPC::" . ($arg{async} ? "Async" : "Sync");
514 632
515 $self->require ($module) 633 $self->require ($module)
516 ->send_arg ($function, $arg{init}, $serialiser) 634 ->send_arg ($function, $arg{init}, $serialiser, $arg{done} || "CORE::exit")
517 ->run ("$module\::run", sub { 635 ->run ("$module\::run", sub {
518 $fh = shift; 636 $fh = shift;
519 637
520 my ($id, $len); 638 my ($id, $len);
521 $rw = AE::io $fh, 0, sub { 639 $rw = AE::io $fh, 0, sub {
522 $rlen = $rlen * 2 + 16 if $rlen - 128 < length $rbuf; 640 $rlen = $rlen * 2 + 16 if $rlen - 128 < length $rbuf;
523 $len = sysread $fh, $rbuf, $rlen - length $rbuf, length $rbuf; 641 $len = sysread $fh, $rbuf, $rlen - length $rbuf, length $rbuf;
524 642
525 if ($len) { 643 if ($len) {
526 while (8 <= length $rbuf) { 644 while (8 <= length $rbuf) {
527 ($id, $len) = unpack "LL", $rbuf; 645 ($id, $len) = unpack "NN", $rbuf;
528 8 + $len <= length $rbuf 646 8 + $len <= length $rbuf
529 or last; 647 or last;
530 648
531 my @r = $t->(substr $rbuf, 8, $len); 649 my @r = $t->(substr $rbuf, 8, $len);
532 substr $rbuf, 0, 8 + $len, ""; 650 substr $rbuf, 0, 8 + $len, "";
548 undef $rw; undef $ww; # it ends here 666 undef $rw; undef $ww; # it ends here
549 667
550 if (@rcb || %rcb) { 668 if (@rcb || %rcb) {
551 $on_error->("unexpected eof"); 669 $on_error->("unexpected eof");
552 } else { 670 } else {
553 $on_destroy->(); 671 $on_destroy->()
672 if $on_destroy;
554 } 673 }
555 } elsif ($! != Errno::EAGAIN && $! != Errno::EWOULDBLOCK) { 674 } elsif ($! != Errno::EAGAIN && $! != Errno::EWOULDBLOCK) {
556 undef $rw; undef $ww; # it ends here 675 undef $rw; undef $ww; # it ends here
557 $on_error->("read: $!"); 676 $on_error->("read: $!");
558 } 677 }
561 $ww ||= AE::io $fh, 1, $wcb; 680 $ww ||= AE::io $fh, 1, $wcb;
562 }); 681 });
563 682
564 my $guard = Guard::guard { 683 my $guard = Guard::guard {
565 $shutdown = 1; 684 $shutdown = 1;
566 $ww ||= $fh && AE::io $fh, 1, $wcb; 685
686 shutdown $fh, 1 if $fh && !$ww;
567 }; 687 };
568 688
569 my $id; 689 my $id;
570 690
571 $arg{async} 691 $arg{async}
573 $id = ($id == 0xffffffff ? 0 : $id) + 1; 693 $id = ($id == 0xffffffff ? 0 : $id) + 1;
574 $id = ($id == 0xffffffff ? 0 : $id) + 1 while exists $rcb{$id}; # rarely loops 694 $id = ($id == 0xffffffff ? 0 : $id) + 1 while exists $rcb{$id}; # rarely loops
575 695
576 $rcb{$id} = pop; 696 $rcb{$id} = pop;
577 697
578 $guard; # keep it alive 698 $guard if 0; # keep it alive
579 699
580 $wbuf .= pack "LL/a*", $id, &$f; 700 $wbuf .= pack "NN/a*", $id, &$f;
581 $ww ||= $fh && AE::io $fh, 1, $wcb; 701 $ww ||= $fh && AE::io $fh, 1, $wcb;
582 } 702 }
583 : sub { 703 : sub {
584 push @rcb, pop; 704 push @rcb, pop;
585 705
586 $guard; # keep it alive 706 $guard; # keep it alive
587 707
588 $wbuf .= pack "L/a*", &$f; 708 $wbuf .= pack "N/a*", &$f;
589 $ww ||= $fh && AE::io $fh, 1, $wcb; 709 $ww ||= $fh && AE::io $fh, 1, $wcb;
590 } 710 }
591} 711}
592 712
593=item $rpc->(..., $cb->(...)) 713=item $rpc->(..., $cb->(...))
632child process to the parent, except that there is no notion of return 752child process to the parent, except that there is no notion of return
633values. 753values.
634 754
635See the examples section earlier in this document for some actual 755See the examples section earlier in this document for some actual
636examples. 756examples.
757
758=back
759
760=head2 PROCESS EXIT
761
762If and when the child process exits depends on the backend and
763configuration. Apart from explicit exits (e.g. by calling C<exit>) or
764runtime conditions (uncaught exceptions, signals etc.), the backends exit
765under these conditions:
766
767=over 4
768
769=item Synchronous Backend
770
771The synchronous backend is very simple: when the process waits for another
772request to arrive and the writing side (usually in the parent) is closed,
773it will exit normally, i.e. as if your main program reached the end of the
774file.
775
776That means that if your parent process exits, the RPC process will usually
777exit as well, either because it is idle anyway, or because it executes a
778request. In the latter case, you will likely get an error when the RPc
779process tries to send the results to the parent (because agruably, you
780shouldn't exit your parent while there are still outstanding requests).
781
782The process is usually quiescent when it happens, so it should rarely be a
783problem, and C<END> handlers can be used to clean up.
784
785=item Asynchronous Backend
786
787For the asynchronous backend, things are more complicated: Whenever it
788listens for another request by the parent, it might detect that the socket
789was closed (e.g. because the parent exited). It will sotp listening for
790new requests and instead try to write out any remaining data (if any) or
791simply check whether the socket cna be written to. After this, the RPC
792process is effectively done - no new requests are incoming, no outstanding
793request data can be written back.
794
795Since chances are high that there are event watchers that the RPC server
796knows nothing about (why else would one use the async backend if not for
797the ability to register watchers?), the event loop would often happily
798continue.
799
800This is why the asynchronous backend explicitly calls C<CORE::exit> when
801it is done (it will raise an exception under other circumstances, which
802might lead to the process not exiting on it's own).
803
804You can override this by specifying a function name to call via the C<done>
805parameter instead.
637 806
638=back 807=back
639 808
640=head1 ADVANCED TOPICS 809=head1 ADVANCED TOPICS
641 810
724half it has passed earlier. 893half it has passed earlier.
725 894
726Here is some (untested) pseudocode to that effect: 895Here is some (untested) pseudocode to that effect:
727 896
728 use AnyEvent::Util; 897 use AnyEvent::Util;
898 use AnyEvent::Fork;
729 use AnyEvent::Fork::RPC; 899 use AnyEvent::Fork::RPC;
730 use IO::FDPass; 900 use IO::FDPass;
731 901
732 my ($s1, $s2) = AnyEvent::Util::portable_socketpair; 902 my ($s1, $s2) = AnyEvent::Util::portable_socketpair;
733 903
769 939
770Of course, this might be blocking if you pass a lot of file descriptors, 940Of course, this might be blocking if you pass a lot of file descriptors,
771so you might want to look into L<AnyEvent::FDpasser> which can handle the 941so you might want to look into L<AnyEvent::FDpasser> which can handle the
772gory details. 942gory details.
773 943
944=head1 EXCEPTIONS
945
946There are no provisions whatsoever for catching exceptions at this time -
947in the child, exeptions might kill the process, causing calls to be lost
948and the parent encountering a fatal error. In the parent, exceptions in
949the result callback will not be caught and cause undefined behaviour.
950
774=head1 SEE ALSO 951=head1 SEE ALSO
775 952
776L<AnyEvent::Fork>, to create the processes in the first place. 953L<AnyEvent::Fork>, to create the processes in the first place.
954
955L<AnyEvent::Fork::Remote>, likewise, but helpful for remote processes.
777 956
778L<AnyEvent::Fork::Pool>, to manage whole pools of processes. 957L<AnyEvent::Fork::Pool>, to manage whole pools of processes.
779 958
780=head1 AUTHOR AND CONTACT INFORMATION 959=head1 AUTHOR AND CONTACT INFORMATION
781 960

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines