… | |
… | |
380 | |
380 | |
381 | There is a slight catch to child watchers, however: you usually start |
381 | There is a slight catch to child watchers, however: you usually start |
382 | them *after* the child process was created, and this means the process |
382 | them *after* the child process was created, and this means the process |
383 | could have exited already (and no SIGCHLD will be sent anymore). |
383 | could have exited already (and no SIGCHLD will be sent anymore). |
384 | |
384 | |
385 | Not all event models handle this correctly (POE doesn't), but even for |
385 | Not all event models handle this correctly (neither POE nor IO::Async |
|
|
386 | do, see their AnyEvent::Impl manpages for details), but even for event |
386 | event models that *do* handle this correctly, they usually need to be |
387 | models that *do* handle this correctly, they usually need to be loaded |
387 | loaded before the process exits (i.e. before you fork in the first |
388 | before the process exits (i.e. before you fork in the first place). |
388 | place). |
389 | AnyEvent's pure perl event loop handles all cases correctly regardless |
|
|
390 | of when you start the watcher. |
389 | |
391 | |
390 | This means you cannot create a child watcher as the very first thing in |
392 | This means you cannot create a child watcher as the very first thing in |
391 | an AnyEvent program, you *have* to create at least one watcher before |
393 | an AnyEvent program, you *have* to create at least one watcher before |
392 | you "fork" the child (alternatively, you can call "AnyEvent::detect"). |
394 | you "fork" the child (alternatively, you can call "AnyEvent::detect"). |
393 | |
395 | |
… | |
… | |
575 | This can be used to signal any errors to the condition variable |
577 | This can be used to signal any errors to the condition variable |
576 | user/consumer. |
578 | user/consumer. |
577 | |
579 | |
578 | $cv->begin ([group callback]) |
580 | $cv->begin ([group callback]) |
579 | $cv->end |
581 | $cv->end |
580 | These two methods are EXPERIMENTAL and MIGHT CHANGE. |
|
|
581 | |
|
|
582 | These two methods can be used to combine many transactions/events |
582 | These two methods can be used to combine many transactions/events |
583 | into one. For example, a function that pings many hosts in parallel |
583 | into one. For example, a function that pings many hosts in parallel |
584 | might want to use a condition variable for the whole process. |
584 | might want to use a condition variable for the whole process. |
585 | |
585 | |
586 | Every call to "->begin" will increment a counter, and every call to |
586 | Every call to "->begin" will increment a counter, and every call to |
587 | "->end" will decrement it. If the counter reaches 0 in "->end", the |
587 | "->end" will decrement it. If the counter reaches 0 in "->end", the |
588 | (last) callback passed to "begin" will be executed. That callback is |
588 | (last) callback passed to "begin" will be executed. That callback is |
589 | *supposed* to call "->send", but that is not required. If no |
589 | *supposed* to call "->send", but that is not required. If no |
590 | callback was set, "send" will be called without any arguments. |
590 | callback was set, "send" will be called without any arguments. |
591 | |
591 | |
592 | Let's clarify this with the ping example: |
592 | You can think of "$cv->send" giving you an OR condition (one call |
|
|
593 | sends), while "$cv->begin" and "$cv->end" giving you an AND |
|
|
594 | condition (all "begin" calls must be "end"'ed before the condvar |
|
|
595 | sends). |
|
|
596 | |
|
|
597 | Let's start with a simple example: you have two I/O watchers (for |
|
|
598 | example, STDOUT and STDERR for a program), and you want to wait for |
|
|
599 | both streams to close before activating a condvar: |
|
|
600 | |
|
|
601 | my $cv = AnyEvent->condvar; |
|
|
602 | |
|
|
603 | $cv->begin; # first watcher |
|
|
604 | my $w1 = AnyEvent->io (fh => $fh1, cb => sub { |
|
|
605 | defined sysread $fh1, my $buf, 4096 |
|
|
606 | or $cv->end; |
|
|
607 | }); |
|
|
608 | |
|
|
609 | $cv->begin; # second watcher |
|
|
610 | my $w2 = AnyEvent->io (fh => $fh2, cb => sub { |
|
|
611 | defined sysread $fh2, my $buf, 4096 |
|
|
612 | or $cv->end; |
|
|
613 | }); |
|
|
614 | |
|
|
615 | $cv->recv; |
|
|
616 | |
|
|
617 | This works because for every event source (EOF on file handle), |
|
|
618 | there is one call to "begin", so the condvar waits for all calls to |
|
|
619 | "end" before sending. |
|
|
620 | |
|
|
621 | The ping example mentioned above is slightly more complicated, as |
|
|
622 | the there are results to be passwd back, and the number of tasks |
|
|
623 | that are begung can potentially be zero: |
593 | |
624 | |
594 | my $cv = AnyEvent->condvar; |
625 | my $cv = AnyEvent->condvar; |
595 | |
626 | |
596 | my %result; |
627 | my %result; |
597 | $cv->begin (sub { $cv->send (\%result) }); |
628 | $cv->begin (sub { $cv->send (\%result) }); |
… | |
… | |
617 | the loop, which serves two important purposes: first, it sets the |
648 | the loop, which serves two important purposes: first, it sets the |
618 | callback to be called once the counter reaches 0, and second, it |
649 | callback to be called once the counter reaches 0, and second, it |
619 | ensures that "send" is called even when "no" hosts are being pinged |
650 | ensures that "send" is called even when "no" hosts are being pinged |
620 | (the loop doesn't execute once). |
651 | (the loop doesn't execute once). |
621 | |
652 | |
622 | This is the general pattern when you "fan out" into multiple |
653 | This is the general pattern when you "fan out" into multiple (but |
623 | subrequests: use an outer "begin"/"end" pair to set the callback and |
654 | potentially none) subrequests: use an outer "begin"/"end" pair to |
624 | ensure "end" is called at least once, and then, for each subrequest |
655 | set the callback and ensure "end" is called at least once, and then, |
625 | you start, call "begin" and for each subrequest you finish, call |
656 | for each subrequest you start, call "begin" and for each subrequest |
626 | "end". |
657 | you finish, call "end". |
627 | |
658 | |
628 | METHODS FOR CONSUMERS |
659 | METHODS FOR CONSUMERS |
629 | These methods should only be used by the consuming side, i.e. the code |
660 | These methods should only be used by the consuming side, i.e. the code |
630 | awaits the condition. |
661 | awaits the condition. |
631 | |
662 | |
… | |
… | |
697 | AnyEvent::Impl::Tk based on Tk, very bad choice. |
728 | AnyEvent::Impl::Tk based on Tk, very bad choice. |
698 | AnyEvent::Impl::Qt based on Qt, cannot be autoprobed (see its docs). |
729 | AnyEvent::Impl::Qt based on Qt, cannot be autoprobed (see its docs). |
699 | AnyEvent::Impl::EventLib based on Event::Lib, leaks memory and worse. |
730 | AnyEvent::Impl::EventLib based on Event::Lib, leaks memory and worse. |
700 | AnyEvent::Impl::POE based on POE, not generic enough for full support. |
731 | AnyEvent::Impl::POE based on POE, not generic enough for full support. |
701 | |
732 | |
|
|
733 | # warning, support for IO::Async is only partial, as it is too broken |
|
|
734 | # and limited toe ven support the AnyEvent API. See AnyEvent::Impl::Async. |
|
|
735 | AnyEvent::Impl::IOAsync based on IO::Async, cannot be autoprobed (see its docs). |
|
|
736 | |
702 | There is no support for WxWidgets, as WxWidgets has no support for |
737 | There is no support for WxWidgets, as WxWidgets has no support for |
703 | watching file handles. However, you can use WxWidgets through the |
738 | watching file handles. However, you can use WxWidgets through the |
704 | POE Adaptor, as POE has a Wx backend that simply polls 20 times per |
739 | POE Adaptor, as POE has a Wx backend that simply polls 20 times per |
705 | second, which was considered to be too horrible to even consider for |
740 | second, which was considered to be too horrible to even consider for |
706 | AnyEvent. Likewise, other POE backends can be used by AnyEvent by |
741 | AnyEvent. Likewise, other POE backends can be used by AnyEvent by |
… | |
… | |
906 | "PERL_ANYEVENT_STRICT" |
941 | "PERL_ANYEVENT_STRICT" |
907 | AnyEvent does not do much argument checking by default, as thorough |
942 | AnyEvent does not do much argument checking by default, as thorough |
908 | argument checking is very costly. Setting this variable to a true |
943 | argument checking is very costly. Setting this variable to a true |
909 | value will cause AnyEvent to load "AnyEvent::Strict" and then to |
944 | value will cause AnyEvent to load "AnyEvent::Strict" and then to |
910 | thoroughly check the arguments passed to most method calls. If it |
945 | thoroughly check the arguments passed to most method calls. If it |
911 | finds any problems it will croak. |
946 | finds any problems, it will croak. |
912 | |
947 | |
913 | In other words, enables "strict" mode. |
948 | In other words, enables "strict" mode. |
914 | |
949 | |
915 | Unlike "use strict", it is definitely recommended ot keep it off in |
950 | Unlike "use strict", it is definitely recommended to keep it off in |
916 | production. Keeping "PERL_ANYEVENT_STRICT=1" in your environment |
951 | production. Keeping "PERL_ANYEVENT_STRICT=1" in your environment |
917 | while developing programs can be very useful, however. |
952 | while developing programs can be very useful, however. |
918 | |
953 | |
919 | "PERL_ANYEVENT_MODEL" |
954 | "PERL_ANYEVENT_MODEL" |
920 | This can be used to specify the event model to be used by AnyEvent, |
955 | This can be used to specify the event model to be used by AnyEvent, |
… | |
… | |
1208 | EV/Any 100000 224 2.88 0.34 0.27 EV + AnyEvent watchers |
1243 | EV/Any 100000 224 2.88 0.34 0.27 EV + AnyEvent watchers |
1209 | CoroEV/Any 100000 224 2.85 0.35 0.28 coroutines + Coro::Signal |
1244 | CoroEV/Any 100000 224 2.85 0.35 0.28 coroutines + Coro::Signal |
1210 | Perl/Any 100000 452 4.13 0.73 0.95 pure perl implementation |
1245 | Perl/Any 100000 452 4.13 0.73 0.95 pure perl implementation |
1211 | Event/Event 16000 517 32.20 31.80 0.81 Event native interface |
1246 | Event/Event 16000 517 32.20 31.80 0.81 Event native interface |
1212 | Event/Any 16000 590 35.85 31.55 1.06 Event + AnyEvent watchers |
1247 | Event/Any 16000 590 35.85 31.55 1.06 Event + AnyEvent watchers |
|
|
1248 | IOAsync/Any 16000 989 38.10 32.77 11.13 via IO::Async::Loop::IO_Poll |
|
|
1249 | IOAsync/Any 16000 990 37.59 29.50 10.61 via IO::Async::Loop::Epoll |
1213 | Glib/Any 16000 1357 102.33 12.31 51.00 quadratic behaviour |
1250 | Glib/Any 16000 1357 102.33 12.31 51.00 quadratic behaviour |
1214 | Tk/Any 2000 1860 27.20 66.31 14.00 SEGV with >> 2000 watchers |
1251 | Tk/Any 2000 1860 27.20 66.31 14.00 SEGV with >> 2000 watchers |
1215 | POE/Event 2000 6328 109.99 751.67 14.02 via POE::Loop::Event |
1252 | POE/Event 2000 6328 109.99 751.67 14.02 via POE::Loop::Event |
1216 | POE/Select 2000 6027 94.54 809.13 579.80 via POE::Loop::Select |
1253 | POE/Select 2000 6027 94.54 809.13 579.80 via POE::Loop::Select |
1217 | |
1254 | |
… | |
… | |
1246 | few of them active), of course, but this was not subject of this |
1283 | few of them active), of course, but this was not subject of this |
1247 | benchmark. |
1284 | benchmark. |
1248 | |
1285 | |
1249 | The "Event" module has a relatively high setup and callback invocation |
1286 | The "Event" module has a relatively high setup and callback invocation |
1250 | cost, but overall scores in on the third place. |
1287 | cost, but overall scores in on the third place. |
|
|
1288 | |
|
|
1289 | "IO::Async" performs admirably well, about on par with "Event", even |
|
|
1290 | when using its pure perl backend. |
1251 | |
1291 | |
1252 | "Glib"'s memory usage is quite a bit higher, but it features a faster |
1292 | "Glib"'s memory usage is quite a bit higher, but it features a faster |
1253 | callback invocation and overall ends up in the same class as "Event". |
1293 | callback invocation and overall ends up in the same class as "Event". |
1254 | However, Glib scales extremely badly, doubling the number of watchers |
1294 | However, Glib scales extremely badly, doubling the number of watchers |
1255 | increases the processing time by more than a factor of four, making it |
1295 | increases the processing time by more than a factor of four, making it |
… | |
… | |
1326 | single "request", that is, reading the token from the pipe and |
1366 | single "request", that is, reading the token from the pipe and |
1327 | forwarding it to another server. This includes deleting the old timeout |
1367 | forwarding it to another server. This includes deleting the old timeout |
1328 | and creating a new one that moves the timeout into the future. |
1368 | and creating a new one that moves the timeout into the future. |
1329 | |
1369 | |
1330 | Results |
1370 | Results |
1331 | name sockets create request |
1371 | name sockets create request |
1332 | EV 20000 69.01 11.16 |
1372 | EV 20000 69.01 11.16 |
1333 | Perl 20000 73.32 35.87 |
1373 | Perl 20000 73.32 35.87 |
|
|
1374 | IOAsync 20000 157.00 98.14 epoll |
|
|
1375 | IOAsync 20000 159.31 616.06 poll |
1334 | Event 20000 212.62 257.32 |
1376 | Event 20000 212.62 257.32 |
1335 | Glib 20000 651.16 1896.30 |
1377 | Glib 20000 651.16 1896.30 |
1336 | POE 20000 349.67 12317.24 uses POE::Loop::Event |
1378 | POE 20000 349.67 12317.24 uses POE::Loop::Event |
1337 | |
1379 | |
1338 | Discussion |
1380 | Discussion |
1339 | This benchmark *does* measure scalability and overall performance of the |
1381 | This benchmark *does* measure scalability and overall performance of the |
1340 | particular event loop. |
1382 | particular event loop. |
1341 | |
1383 | |
1342 | EV is again fastest. Since it is using epoll on my system, the setup |
1384 | EV is again fastest. Since it is using epoll on my system, the setup |
1343 | time is relatively high, though. |
1385 | time is relatively high, though. |
1344 | |
1386 | |
1345 | Perl surprisingly comes second. It is much faster than the C-based event |
1387 | Perl surprisingly comes second. It is much faster than the C-based event |
1346 | loops Event and Glib. |
1388 | loops Event and Glib. |
|
|
1389 | |
|
|
1390 | IO::Async performs very well when using its epoll backend, and still |
|
|
1391 | quite good compared to Glib when using its pure perl backend. |
1347 | |
1392 | |
1348 | Event suffers from high setup time as well (look at its code and you |
1393 | Event suffers from high setup time as well (look at its code and you |
1349 | will understand why). Callback invocation also has a high overhead |
1394 | will understand why). Callback invocation also has a high overhead |
1350 | compared to the "$_->() for .."-style loop that the Perl event loop |
1395 | compared to the "$_->() for .."-style loop that the Perl event loop |
1351 | uses. Event uses select or poll in basically all documented |
1396 | uses. Event uses select or poll in basically all documented |
… | |
… | |
1407 | THE IO::Lambda BENCHMARK |
1452 | THE IO::Lambda BENCHMARK |
1408 | Recently I was told about the benchmark in the IO::Lambda manpage, which |
1453 | Recently I was told about the benchmark in the IO::Lambda manpage, which |
1409 | could be misinterpreted to make AnyEvent look bad. In fact, the |
1454 | could be misinterpreted to make AnyEvent look bad. In fact, the |
1410 | benchmark simply compares IO::Lambda with POE, and IO::Lambda looks |
1455 | benchmark simply compares IO::Lambda with POE, and IO::Lambda looks |
1411 | better (which shouldn't come as a surprise to anybody). As such, the |
1456 | better (which shouldn't come as a surprise to anybody). As such, the |
1412 | benchmark is fine, and shows that the AnyEvent backend from IO::Lambda |
1457 | benchmark is fine, and mostly shows that the AnyEvent backend from |
1413 | isn't very optimal. But how would AnyEvent compare when used without the |
1458 | IO::Lambda isn't very optimal. But how would AnyEvent compare when used |
1414 | extra baggage? To explore this, I wrote the equivalent benchmark for |
1459 | without the extra baggage? To explore this, I wrote the equivalent |
1415 | AnyEvent. |
1460 | benchmark for AnyEvent. |
1416 | |
1461 | |
1417 | The benchmark itself creates an echo-server, and then, for 500 times, |
1462 | The benchmark itself creates an echo-server, and then, for 500 times, |
1418 | connects to the echo server, sends a line, waits for the reply, and then |
1463 | connects to the echo server, sends a line, waits for the reply, and then |
1419 | creates the next connection. This is a rather bad benchmark, as it |
1464 | creates the next connection. This is a rather bad benchmark, as it |
1420 | doesn't test the efficiency of the framework, but it is a benchmark |
1465 | doesn't test the efficiency of the framework or much non-blocking I/O, |
1421 | nevertheless. |
1466 | but it is a benchmark nevertheless. |
1422 | |
1467 | |
1423 | name runtime |
1468 | name runtime |
1424 | Lambda/select 0.330 sec |
1469 | Lambda/select 0.330 sec |
1425 | + optimized 0.122 sec |
1470 | + optimized 0.122 sec |
1426 | Lambda/AnyEvent 0.327 sec |
1471 | Lambda/AnyEvent 0.327 sec |
… | |
… | |
1432 | |
1477 | |
1433 | AnyEvent/select/nb 0.085 sec |
1478 | AnyEvent/select/nb 0.085 sec |
1434 | AnyEvent/EV/nb 0.068 sec |
1479 | AnyEvent/EV/nb 0.068 sec |
1435 | +state machine 0.134 sec |
1480 | +state machine 0.134 sec |
1436 | |
1481 | |
1437 | The benchmark is also a bit unfair (my fault) - the IO::Lambda |
1482 | The benchmark is also a bit unfair (my fault): the IO::Lambda/POE |
1438 | benchmarks actually make blocking connects and use 100% blocking I/O, |
1483 | benchmarks actually make blocking connects and use 100% blocking I/O, |
1439 | defeating the purpose of an event-based solution. All of the newly |
1484 | defeating the purpose of an event-based solution. All of the newly |
1440 | written AnyEvent benchmarks use 100% non-blocking connects (using |
1485 | written AnyEvent benchmarks use 100% non-blocking connects (using |
1441 | AnyEvent::Socket::tcp_connect and the asynchronous pure perl DNS |
1486 | AnyEvent::Socket::tcp_connect and the asynchronous pure perl DNS |
1442 | resolver), so AnyEvent is at a disadvantage here as non-blocking |
1487 | resolver), so AnyEvent is at a disadvantage here, as non-blocking |
1443 | connects generally require a lot more bookkeeping and event handling |
1488 | connects generally require a lot more bookkeeping and event handling |
1444 | than blocking connects (which involve a single syscall only). |
1489 | than blocking connects (which involve a single syscall only). |
1445 | |
1490 | |
1446 | The last AnyEvent benchmark additionally uses AnyEvent::Handle, which |
1491 | The last AnyEvent benchmark additionally uses AnyEvent::Handle, which |
1447 | offers similar expressive power as POE and IO::Lambda (using |
1492 | offers similar expressive power as POE and IO::Lambda, using |
1448 | conventional Perl syntax), which means both the echo server and the |
1493 | conventional Perl syntax. This means that both the echo server and the |
1449 | client are 100% non-blocking w.r.t. I/O, further placing it at a |
1494 | client are 100% non-blocking, further placing it at a disadvantage. |
1450 | disadvantage. |
|
|
1451 | |
1495 | |
1452 | As you can see, AnyEvent + EV even beats the hand-optimised "raw sockets |
1496 | As you can see, the AnyEvent + EV combination even beats the |
1453 | benchmark", while AnyEvent + its pure perl backend easily beats |
1497 | hand-optimised "raw sockets benchmark", while AnyEvent + its pure perl |
1454 | IO::Lambda and POE. |
1498 | backend easily beats IO::Lambda and POE. |
1455 | |
1499 | |
1456 | And even the 100% non-blocking version written using the high-level (and |
1500 | And even the 100% non-blocking version written using the high-level (and |
1457 | slow :) AnyEvent::Handle abstraction beats both POE and IO::Lambda, even |
1501 | slow :) AnyEvent::Handle abstraction beats both POE and IO::Lambda by a |
1458 | thought it does all of DNS, tcp-connect and socket I/O in a non-blocking |
1502 | large margin, even though it does all of DNS, tcp-connect and socket I/O |
1459 | way. |
1503 | in a non-blocking way. |
1460 | |
1504 | |
1461 | The two AnyEvent benchmarks can be found as eg/ae0.pl and eg/ae2.pl in |
1505 | The two AnyEvent benchmarks programs can be found as eg/ae0.pl and |
1462 | the AnyEvent distribution, the remaining benchmarks are part of the |
1506 | eg/ae2.pl in the AnyEvent distribution, the remaining benchmarks are |
1463 | IO::lambda distribution and were used without any changes. |
1507 | part of the IO::lambda distribution and were used without any changes. |
1464 | |
1508 | |
1465 | SIGNALS |
1509 | SIGNALS |
1466 | AnyEvent currently installs handlers for these signals: |
1510 | AnyEvent currently installs handlers for these signals: |
1467 | |
1511 | |
1468 | SIGCHLD |
1512 | SIGCHLD |
1469 | A handler for "SIGCHLD" is installed by AnyEvent's child watcher |
1513 | A handler for "SIGCHLD" is installed by AnyEvent's child watcher |
1470 | emulation for event loops that do not support them natively. Also, |
1514 | emulation for event loops that do not support them natively. Also, |
1471 | some event loops install a similar handler. |
1515 | some event loops install a similar handler. |
|
|
1516 | |
|
|
1517 | If, when AnyEvent is loaded, SIGCHLD is set to IGNORE, then AnyEvent |
|
|
1518 | will reset it to default, to avoid losing child exit statuses. |
1472 | |
1519 | |
1473 | SIGPIPE |
1520 | SIGPIPE |
1474 | A no-op handler is installed for "SIGPIPE" when $SIG{PIPE} is |
1521 | A no-op handler is installed for "SIGPIPE" when $SIG{PIPE} is |
1475 | "undef" when AnyEvent gets loaded. |
1522 | "undef" when AnyEvent gets loaded. |
1476 | |
1523 | |
… | |
… | |
1512 | Similar considerations apply to $ENV{PERL_ANYEVENT_VERBOSE}, as that can |
1559 | Similar considerations apply to $ENV{PERL_ANYEVENT_VERBOSE}, as that can |
1513 | be used to probe what backend is used and gain other information (which |
1560 | be used to probe what backend is used and gain other information (which |
1514 | is probably even less useful to an attacker than PERL_ANYEVENT_MODEL), |
1561 | is probably even less useful to an attacker than PERL_ANYEVENT_MODEL), |
1515 | and $ENV{PERL_ANYEVENT_STRICT}. |
1562 | and $ENV{PERL_ANYEVENT_STRICT}. |
1516 | |
1563 | |
|
|
1564 | Note that AnyEvent will remove *all* environment variables starting with |
|
|
1565 | "PERL_ANYEVENT_" from %ENV when it is loaded while taint mode is |
|
|
1566 | enabled. |
|
|
1567 | |
1517 | BUGS |
1568 | BUGS |
1518 | Perl 5.8 has numerous memleaks that sometimes hit this module and are |
1569 | Perl 5.8 has numerous memleaks that sometimes hit this module and are |
1519 | hard to work around. If you suffer from memleaks, first upgrade to Perl |
1570 | hard to work around. If you suffer from memleaks, first upgrade to Perl |
1520 | 5.10 and check wether the leaks still show up. (Perl 5.10.0 has other |
1571 | 5.10 and check wether the leaks still show up. (Perl 5.10.0 has other |
1521 | annoying memleaks, such as leaking on "map" and "grep" but it is usually |
1572 | annoying memleaks, such as leaking on "map" and "grep" but it is usually |