[ViewVC] Annotation of: cvs/AnyEvent/README

NAME
    AnyEvent - provide framework for multiple event loops

    EV, Event, Glib, Tk, Perl, Event::Lib, Qt and POE are various supported
    event loops.

SYNOPSIS
       use AnyEvent;

       # file descriptor readable
       my $w = AnyEvent->io (fh => $fh, poll => "r", cb => sub { ...  });

       # one-shot or repeating timers
       my $w = AnyEvent->timer (after => $seconds, cb => sub { ...  });
       my $w = AnyEvent->timer (after => $seconds, interval => $seconds, cb => ...

       print AnyEvent->now;  # prints current event loop time
       print AnyEvent->time; # think Time::HiRes::time or simply CORE::time.

       # POSIX signal
       my $w = AnyEvent->signal (signal => "TERM", cb => sub { ... });

       # child process exit
       my $w = AnyEvent->child (pid => $pid, cb => sub {
          my ($pid, $status) = @_;
          ...
       });

       # called when event loop idle (if applicable)
       my $w = AnyEvent->idle (cb => sub { ... });

       my $w = AnyEvent->condvar; # stores whether a condition was flagged
       $w->send; # wake up current and all future recv's
       $w->recv; # enters "main loop" till $condvar gets ->send
       # use a condvar in callback mode:
       $w->cb (sub { $_[0]->recv });

INTRODUCTION/TUTORIAL
    This manpage is mainly a reference manual. If you are interested in a
    tutorial or some gentle introduction, have a look at the AnyEvent::Intro
    manpage.

WHY YOU SHOULD USE THIS MODULE (OR NOT)
    Glib, POE, IO::Async, Event... CPAN offers event models by the dozen
    nowadays. So what is different about AnyEvent?

    Executive Summary: AnyEvent is *compatible*, AnyEvent is *free of
    policy* and AnyEvent is *small and efficient*.

    First and foremost, *AnyEvent is not an event model* itself, it only
    interfaces to whatever event model the main program happens to use, in a
    pragmatic way. For event models and certain classes of immortals alike,
    the statement "there can only be one" is a bitter reality: In general,
    only one event loop can be active at the same time in a process.
    AnyEvent cannot change this, but it can hide the differences between
    those event loops.

    The goal of AnyEvent is to offer module authors the ability to do event
    programming (waiting for I/O or timer events) without subscribing to a
    religion, a way of living, and most importantly: without forcing your
    module users into the same thing by forcing them to use the same event
    model you use.

    For modules like POE or IO::Async (which is a total misnomer as it is
    actually doing all I/O *synchronously*...), using them in your module is
    like joining a cult: After you joined, you are dependent on them and you
    cannot use anything else, as they are simply incompatible to everything
    that isn't them. What's worse, all the potential users of your module
    are *also* forced to use the same event loop you use.

    AnyEvent is different: AnyEvent + POE works fine. AnyEvent + Glib works
    fine. AnyEvent + Tk works fine etc. etc. but none of these work together
    with the rest: POE + IO::Async? No go. Tk + Event? No go. Again: if your
    module uses one of those, every user of your module has to use it, too.
    But if your module uses AnyEvent, it works transparently with all event
    models it supports (including stuff like IO::Async, as long as those use
    one of the supported event loops. It is trivial to add new event loops
    to AnyEvent, too, so it is future-proof).

    In addition to being free of having to use *the one and only true event
    model*, AnyEvent also is free of bloat and policy: with POE or similar
    modules, you get an enormous amount of code and strict rules you have to
    follow. AnyEvent, on the other hand, is lean and up to the point, by
    only offering the functionality that is necessary, in as thin as a
    wrapper as technically possible.

    Of course, AnyEvent comes with a big (and fully optional!) toolbox of
    useful functionality, such as an asynchronous DNS resolver, 100%
    non-blocking connects (even with TLS/SSL, IPv6 and on broken platforms
    such as Windows) and lots of real-world knowledge and workarounds for
    platform bugs and differences.

    Now, if you *do want* lots of policy (this can arguably be somewhat
    useful) and you want to force your users to use the one and only event
    model, you should *not* use this module.

DESCRIPTION
    AnyEvent provides an identical interface to multiple event loops. This
    allows module authors to utilise an event loop without forcing module
    users to use the same event loop (as only a single event loop can
    coexist peacefully at any one time).

    The interface itself is vaguely similar, but not identical to the Event
    module.

    During the first call of any watcher-creation method, the module tries
    to detect the currently loaded event loop by probing whether one of the
    following modules is already loaded: EV, Event, Glib,
    AnyEvent::Impl::Perl, Tk, Event::Lib, Qt, POE. The first one found is
    used. If none are found, the module tries to load these modules
    (excluding Tk, Event::Lib, Qt and POE as the pure perl adaptor should
    always succeed) in the order given. The first one that can be
    successfully loaded will be used. If, after this, still none could be
    found, AnyEvent will fall back to a pure-perl event loop, which is not
    very efficient, but should work everywhere.

    Because AnyEvent first checks for modules that are already loaded,
    loading an event model explicitly before first using AnyEvent will
    likely make that model the default. For example:

       use Tk;
       use AnyEvent;

       # .. AnyEvent will likely default to Tk

    The *likely* means that, if any module loads another event model and
    starts using it, all bets are off. Maybe you should tell their authors
    to use AnyEvent so their modules work together with others seamlessly...

    The pure-perl implementation of AnyEvent is called
    "AnyEvent::Impl::Perl". Like other event modules you can load it
    explicitly and enjoy the high availability of that event loop :)

WATCHERS
    AnyEvent has the central concept of a *watcher*, which is an object that
    stores relevant data for each kind of event you are waiting for, such as
    the callback to call, the file handle to watch, etc.

    These watchers are normal Perl objects with normal Perl lifetime. After
    creating a watcher it will immediately "watch" for events and invoke the
    callback when the event occurs (of course, only when the event model is
    in control).

    Note that callbacks must not permanently change global variables
    potentially in use by the event loop (such as $_ or $[) and that
    callbacks must not "die". The former is good programming practise in
    Perl and the latter stems from the fact that exception handling differs
    widely between event loops.

    To disable the watcher you have to destroy it (e.g. by setting the
    variable you store it in to "undef" or otherwise deleting all references
    to it).

    All watchers are created by calling a method on the "AnyEvent" class.

    Many watchers either are used with "recursion" (repeating timers for
    example), or need to refer to their watcher object in other ways.

    An any way to achieve that is this pattern:

       my $w; $w = AnyEvent->type (arg => value ..., cb => sub {
          # you can use $w here, for example to undef it
          undef $w;
       });

    Note that "my $w; $w =" combination. This is necessary because in Perl,
    my variables are only visible after the statement in which they are
    declared.

  I/O WATCHERS
    You can create an I/O watcher by calling the "AnyEvent->io" method with
    the following mandatory key-value pairs as arguments:

    "fh" is the Perl *file handle* (or a naked file descriptor) to watch for
    events (AnyEvent might or might not keep a reference to this file
    handle). Note that only file handles pointing to things for which
    non-blocking operation makes sense are allowed. This includes sockets,
    most character devices, pipes, fifos and so on, but not for example
    files or block devices.

    "poll" must be a string that is either "r" or "w", which creates a
    watcher waiting for "r"eadable or "w"ritable events, respectively.

    "cb" is the callback to invoke each time the file handle becomes ready.

    Although the callback might get passed parameters, their value and
    presence is undefined and you cannot rely on them. Portable AnyEvent
    callbacks cannot use arguments passed to I/O watcher callbacks.

    The I/O watcher might use the underlying file descriptor or a copy of
    it. You must not close a file handle as long as any watcher is active on
    the underlying file descriptor.

    Some event loops issue spurious readyness notifications, so you should
    always use non-blocking calls when reading/writing from/to your file
    handles.

    Example: wait for readability of STDIN, then read a line and disable the
    watcher.

       my $w; $w = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub {
          chomp (my $input = <STDIN>);
          warn "read: $input\n";
          undef $w;
       });

  TIME WATCHERS
    You can create a time watcher by calling the "AnyEvent->timer" method
    with the following mandatory arguments:

    "after" specifies after how many seconds (fractional values are
    supported) the callback should be invoked. "cb" is the callback to
    invoke in that case.

    Although the callback might get passed parameters, their value and
    presence is undefined and you cannot rely on them. Portable AnyEvent
    callbacks cannot use arguments passed to time watcher callbacks.

    The callback will normally be invoked once only. If you specify another
    parameter, "interval", as a strictly positive number (> 0), then the
    callback will be invoked regularly at that interval (in fractional
    seconds) after the first invocation. If "interval" is specified with a
    false value, then it is treated as if it were missing.

    The callback will be rescheduled before invoking the callback, but no
    attempt is done to avoid timer drift in most backends, so the interval
    is only approximate.

    Example: fire an event after 7.7 seconds.

       my $w = AnyEvent->timer (after => 7.7, cb => sub {
          warn "timeout\n";
       });

       # to cancel the timer:
       undef $w;

    Example 2: fire an event after 0.5 seconds, then roughly every second.

       my $w = AnyEvent->timer (after => 0.5, interval => 1, cb => sub {
          warn "timeout\n";
       };

   TIMING ISSUES
    There are two ways to handle timers: based on real time (relative, "fire
    in 10 seconds") and based on wallclock time (absolute, "fire at 12
    o'clock").

    While most event loops expect timers to specified in a relative way,
    they use absolute time internally. This makes a difference when your
    clock "jumps", for example, when ntp decides to set your clock backwards
    from the wrong date of 2014-01-01 to 2008-01-01, a watcher that is
    supposed to fire "after" a second might actually take six years to
    finally fire.

    AnyEvent cannot compensate for this. The only event loop that is
    conscious about these issues is EV, which offers both relative
    (ev_timer, based on true relative time) and absolute (ev_periodic, based
    on wallclock time) timers.

    AnyEvent always prefers relative timers, if available, matching the
    AnyEvent API.

    AnyEvent has two additional methods that return the "current time":

    AnyEvent->time
        This returns the "current wallclock time" as a fractional number of
        seconds since the Epoch (the same thing as "time" or
        "Time::HiRes::time" return, and the result is guaranteed to be
        compatible with those).

        It progresses independently of any event loop processing, i.e. each
        call will check the system clock, which usually gets updated
        frequently.

    AnyEvent->now
        This also returns the "current wallclock time", but unlike "time",
        above, this value might change only once per event loop iteration,
        depending on the event loop (most return the same time as "time",
        above). This is the time that AnyEvent's timers get scheduled
        against.

        *In almost all cases (in all cases if you don't care), this is the
        function to call when you want to know the current time.*

        This function is also often faster then "AnyEvent->time", and thus
        the preferred method if you want some timestamp (for example,
        AnyEvent::Handle uses this to update it's activity timeouts).

        The rest of this section is only of relevance if you try to be very
        exact with your timing, you can skip it without bad conscience.

        For a practical example of when these times differ, consider
        Event::Lib and EV and the following set-up:

        The event loop is running and has just invoked one of your callback
        at time=500 (assume no other callbacks delay processing). In your
        callback, you wait a second by executing "sleep 1" (blocking the
        process for a second) and then (at time=501) you create a relative
        timer that fires after three seconds.

        With Event::Lib, "AnyEvent->time" and "AnyEvent->now" will both
        return 501, because that is the current time, and the timer will be
        scheduled to fire at time=504 (501 + 3).

        With EV, "AnyEvent->time" returns 501 (as that is the current time),
        but "AnyEvent->now" returns 500, as that is the time the last event
        processing phase started. With EV, your timer gets scheduled to run
        at time=503 (500 + 3).

        In one sense, Event::Lib is more exact, as it uses the current time
        regardless of any delays introduced by event processing. However,
        most callbacks do not expect large delays in processing, so this
        causes a higher drift (and a lot more system calls to get the
        current time).

        In another sense, EV is more exact, as your timer will be scheduled
        at the same time, regardless of how long event processing actually
        took.

        In either case, if you care (and in most cases, you don't), then you
        can get whatever behaviour you want with any event loop, by taking
        the difference between "AnyEvent->time" and "AnyEvent->now" into
        account.

    AnyEvent->now_update
        Some event loops (such as EV or AnyEvent::Impl::Perl) cache the
        current time for each loop iteration (see the discussion of
        AnyEvent->now, above).

        When a callback runs for a long time (or when the process sleeps),
        then this "current" time will differ substantially from the real
        time, which might affect timers and time-outs.

        When this is the case, you can call this method, which will update
        the event loop's idea of "current time".

        Note that updating the time *might* cause some events to be handled.

  SIGNAL WATCHERS
    You can watch for signals using a signal watcher, "signal" is the signal
    *name* in uppercase and without any "SIG" prefix, "cb" is the Perl
    callback to be invoked whenever a signal occurs.

    Although the callback might get passed parameters, their value and
    presence is undefined and you cannot rely on them. Portable AnyEvent
    callbacks cannot use arguments passed to signal watcher callbacks.

    Multiple signal occurrences can be clumped together into one callback
    invocation, and callback invocation will be synchronous. Synchronous
    means that it might take a while until the signal gets handled by the
    process, but it is guaranteed not to interrupt any other callbacks.

    The main advantage of using these watchers is that you can share a
    signal between multiple watchers.

    This watcher might use %SIG, so programs overwriting those signals
    directly will likely not work correctly.

    Example: exit on SIGINT

       my $w = AnyEvent->signal (signal => "INT", cb => sub { exit 1 });

  CHILD PROCESS WATCHERS
    You can also watch on a child process exit and catch its exit status.

    The child process is specified by the "pid" argument (if set to 0, it
    watches for any child process exit). The watcher will triggered only
    when the child process has finished and an exit status is available, not
    on any trace events (stopped/continued).

    The callback will be called with the pid and exit status (as returned by
    waitpid), so unlike other watcher types, you *can* rely on child watcher
    callback arguments.

    This watcher type works by installing a signal handler for "SIGCHLD",
    and since it cannot be shared, nothing else should use SIGCHLD or reap
    random child processes (waiting for specific child processes, e.g.
    inside "system", is just fine).

    There is a slight catch to child watchers, however: you usually start
    them *after* the child process was created, and this means the process
    could have exited already (and no SIGCHLD will be sent anymore).

    Not all event models handle this correctly (neither POE nor IO::Async
    do, see their AnyEvent::Impl manpages for details), but even for event
    models that *do* handle this correctly, they usually need to be loaded
    before the process exits (i.e. before you fork in the first place).
    AnyEvent's pure perl event loop handles all cases correctly regardless
    of when you start the watcher.

    This means you cannot create a child watcher as the very first thing in
    an AnyEvent program, you *have* to create at least one watcher before
    you "fork" the child (alternatively, you can call "AnyEvent::detect").

    Example: fork a process and wait for it

       my $done = AnyEvent->condvar;
  
       my $pid = fork or exit 5;
  
       my $w = AnyEvent->child (
          pid => $pid,
          cb  => sub {
             my ($pid, $status) = @_;
             warn "pid $pid exited with status $status";
             $done->send;
          },
       );
  
       # do something else, then wait for process exit
       $done->recv;

  IDLE WATCHERS
    Sometimes there is a need to do something, but it is not so important to
    do it instantly, but only when there is nothing better to do. This
    "nothing better to do" is usually defined to be "no other events need
    attention by the event loop".

    Idle watchers ideally get invoked when the event loop has nothing better
    to do, just before it would block the process to wait for new events.
    Instead of blocking, the idle watcher is invoked.

    Most event loops unfortunately do not really support idle watchers (only
    EV, Event and Glib do it in a usable fashion) - for the rest, AnyEvent
    will simply call the callback "from time to time".

    Example: read lines from STDIN, but only process them when the program
    is otherwise idle:

       my @lines; # read data
       my $idle_w;
       my $io_w = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub {
          push @lines, scalar <STDIN>;

          # start an idle watcher, if not already done
          $idle_w ||= AnyEvent->idle (cb => sub {
             # handle only one line, when there are lines left
             if (my $line = shift @lines) {
                print "handled when idle: $line";
             } else {
                # otherwise disable the idle watcher again
                undef $idle_w;
             }
          });
       });

  CONDITION VARIABLES
    If you are familiar with some event loops you will know that all of them
    require you to run some blocking "loop", "run" or similar function that
    will actively watch for new events and call your callbacks.

    AnyEvent is slightly different: it expects somebody else to run the
    event loop and will only block when necessary (usually when told by the
    user).

    The instrument to do that is called a "condition variable", so called
    because they represent a condition that must become true.

    Now is probably a good time to look at the examples further below.

    Condition variables can be created by calling the "AnyEvent->condvar"
    method, usually without arguments. The only argument pair allowed is
    "cb", which specifies a callback to be called when the condition
    variable becomes true, with the condition variable as the first argument
    (but not the results).

    After creation, the condition variable is "false" until it becomes
    "true" by calling the "send" method (or calling the condition variable
    as if it were a callback, read about the caveats in the description for
    the "->send" method).

    Condition variables are similar to callbacks, except that you can
    optionally wait for them. They can also be called merge points - points
    in time where multiple outstanding events have been processed. And yet
    another way to call them is transactions - each condition variable can
    be used to represent a transaction, which finishes at some point and
    delivers a result.

    Condition variables are very useful to signal that something has
    finished, for example, if you write a module that does asynchronous http
    requests, then a condition variable would be the ideal candidate to
    signal the availability of results. The user can either act when the
    callback is called or can synchronously "->recv" for the results.

    You can also use them to simulate traditional event loops - for example,
    you can block your main program until an event occurs - for example, you
    could "->recv" in your main program until the user clicks the Quit
    button of your app, which would "->send" the "quit" event.

    Note that condition variables recurse into the event loop - if you have
    two pieces of code that call "->recv" in a round-robin fashion, you
    lose. Therefore, condition variables are good to export to your caller,
    but you should avoid making a blocking wait yourself, at least in
    callbacks, as this asks for trouble.

    Condition variables are represented by hash refs in perl, and the keys
    used by AnyEvent itself are all named "_ae_XXX" to make subclassing easy
    (it is often useful to build your own transaction class on top of
    AnyEvent). To subclass, use "AnyEvent::CondVar" as base class and call
    it's "new" method in your own "new" method.

    There are two "sides" to a condition variable - the "producer side"
    which eventually calls "-> send", and the "consumer side", which waits
    for the send to occur.

    Example: wait for a timer.

       # wait till the result is ready
       my $result_ready = AnyEvent->condvar;

       # do something such as adding a timer
       # or socket watcher the calls $result_ready->send
       # when the "result" is ready.
       # in this case, we simply use a timer:
       my $w = AnyEvent->timer (
          after => 1,
          cb    => sub { $result_ready->send },
       );

       # this "blocks" (while handling events) till the callback
       # calls -<send
       $result_ready->recv;

    Example: wait for a timer, but take advantage of the fact that condition
    variables are also callable directly.

       my $done = AnyEvent->condvar;
       my $delay = AnyEvent->timer (after => 5, cb => $done);
       $done->recv;

    Example: Imagine an API that returns a condvar and doesn't support
    callbacks. This is how you make a synchronous call, for example from the
    main program:

       use AnyEvent::CouchDB;

       ...

       my @info = $couchdb->info->recv;

    And this is how you would just set a callback to be called whenever the
    results are available:

       $couchdb->info->cb (sub {
          my @info = $_[0]->recv;
       });

   METHODS FOR PRODUCERS
    These methods should only be used by the producing side, i.e. the
    code/module that eventually sends the signal. Note that it is also the
    producer side which creates the condvar in most cases, but it isn't
    uncommon for the consumer to create it as well.

    $cv->send (...)
        Flag the condition as ready - a running "->recv" and all further
        calls to "recv" will (eventually) return after this method has been
        called. If nobody is waiting the send will be remembered.

        If a callback has been set on the condition variable, it is called
        immediately from within send.

        Any arguments passed to the "send" call will be returned by all
        future "->recv" calls.

        Condition variables are overloaded so one can call them directly (as
        if they were a code reference). Calling them directly is the same as
        calling "send".

    $cv->croak ($error)
        Similar to send, but causes all call's to "->recv" to invoke
        "Carp::croak" with the given error message/object/scalar.

        This can be used to signal any errors to the condition variable
        user/consumer. Doing it this way instead of calling "croak" directly
        delays the error detetcion, but has the overwhelmign advantage that
        it diagnoses the error at the place where the result is expected,
        and not deep in some event clalback without connection to the actual
        code causing the problem.

    $cv->begin ([group callback])
    $cv->end
        These two methods can be used to combine many transactions/events
        into one. For example, a function that pings many hosts in parallel
        might want to use a condition variable for the whole process.

        Every call to "->begin" will increment a counter, and every call to
        "->end" will decrement it. If the counter reaches 0 in "->end", the
        (last) callback passed to "begin" will be executed. That callback is
        *supposed* to call "->send", but that is not required. If no
        callback was set, "send" will be called without any arguments.

        You can think of "$cv->send" giving you an OR condition (one call
        sends), while "$cv->begin" and "$cv->end" giving you an AND
        condition (all "begin" calls must be "end"'ed before the condvar
        sends).

        Let's start with a simple example: you have two I/O watchers (for
        example, STDOUT and STDERR for a program), and you want to wait for
        both streams to close before activating a condvar:

           my $cv = AnyEvent->condvar;

           $cv->begin; # first watcher
           my $w1 = AnyEvent->io (fh => $fh1, cb => sub {
              defined sysread $fh1, my $buf, 4096
                 or $cv->end;
           });

           $cv->begin; # second watcher
           my $w2 = AnyEvent->io (fh => $fh2, cb => sub {
              defined sysread $fh2, my $buf, 4096
                 or $cv->end;
           });

           $cv->recv;

        This works because for every event source (EOF on file handle),
        there is one call to "begin", so the condvar waits for all calls to
        "end" before sending.

        The ping example mentioned above is slightly more complicated, as
        the there are results to be passwd back, and the number of tasks
        that are begung can potentially be zero:

           my $cv = AnyEvent->condvar;

           my %result;
           $cv->begin (sub { $cv->send (\%result) });

           for my $host (@list_of_hosts) {
              $cv->begin;
              ping_host_then_call_callback $host, sub {
                 $result{$host} = ...;
                 $cv->end;
              };
           }

           $cv->end;

        This code fragment supposedly pings a number of hosts and calls
        "send" after results for all then have have been gathered - in any
        order. To achieve this, the code issues a call to "begin" when it
        starts each ping request and calls "end" when it has received some
        result for it. Since "begin" and "end" only maintain a counter, the
        order in which results arrive is not relevant.

        There is an additional bracketing call to "begin" and "end" outside
        the loop, which serves two important purposes: first, it sets the
        callback to be called once the counter reaches 0, and second, it
        ensures that "send" is called even when "no" hosts are being pinged
        (the loop doesn't execute once).

        This is the general pattern when you "fan out" into multiple (but
        potentially none) subrequests: use an outer "begin"/"end" pair to
        set the callback and ensure "end" is called at least once, and then,
        for each subrequest you start, call "begin" and for each subrequest
        you finish, call "end".

   METHODS FOR CONSUMERS
    These methods should only be used by the consuming side, i.e. the code
    awaits the condition.

    $cv->recv
        Wait (blocking if necessary) until the "->send" or "->croak" methods
        have been called on c<$cv>, while servicing other watchers normally.

        You can only wait once on a condition - additional calls are valid
        but will return immediately.

        If an error condition has been set by calling "->croak", then this
        function will call "croak".

        In list context, all parameters passed to "send" will be returned,
        in scalar context only the first one will be returned.

        Note that doing a blocking wait in a callback is not supported by
        any event loop, that is, recursive invocation of a blocking "->recv"
        is not allowed, and the "recv" call will "croak" if such a condition
        is detected. This condition can be slightly loosened by using
        Coro::AnyEvent, which allows you to do a blocking "->recv" from any
        thread that doesn't run the event loop itself.

        Not all event models support a blocking wait - some die in that case
        (programs might want to do that to stay interactive), so *if you are
        using this from a module, never require a blocking wait*. Instead,
        let the caller decide whether the call will block or not (for
        example, by coupling condition variables with some kind of request
        results and supporting callbacks so the caller knows that getting
        the result will not block, while still supporting blocking waits if
        the caller so desires).

        You can ensure that "-recv" never blocks by setting a callback and
        only calling "->recv" from within that callback (or at a later
        time). This will work even when the event loop does not support
        blocking waits otherwise.

    $bool = $cv->ready
        Returns true when the condition is "true", i.e. whether "send" or
        "croak" have been called.

    $cb = $cv->cb ($cb->($cv))
        This is a mutator function that returns the callback set and
        optionally replaces it before doing so.

        The callback will be called when the condition becomes "true", i.e.
        when "send" or "croak" are called, with the only argument being the
        condition variable itself. Calling "recv" inside the callback or at
        any later time is guaranteed not to block.

SUPPORTED EVENT LOOPS/BACKENDS
    The available backend classes are (every class has its own manpage):

    Backends that are autoprobed when no other event loop can be found.
        EV is the preferred backend when no other event loop seems to be in
        use. If EV is not installed, then AnyEvent will try Event, and,
        failing that, will fall back to its own pure-perl implementation,
        which is available everywhere as it comes with AnyEvent itself.

           AnyEvent::Impl::EV        based on EV (interface to libev, best choice).
           AnyEvent::Impl::Event     based on Event, very stable, few glitches.
           AnyEvent::Impl::Perl      pure-perl implementation, fast and portable.

    Backends that are transparently being picked up when they are used.
        These will be used when they are currently loaded when the first
        watcher is created, in which case it is assumed that the application
        is using them. This means that AnyEvent will automatically pick the
        right backend when the main program loads an event module before
        anything starts to create watchers. Nothing special needs to be done
        by the main program.

           AnyEvent::Impl::Glib      based on Glib, slow but very stable.
           AnyEvent::Impl::Tk        based on Tk, very broken.
           AnyEvent::Impl::EventLib  based on Event::Lib, leaks memory and worse.
           AnyEvent::Impl::POE       based on POE, very slow, some limitations.

    Backends with special needs.
        Qt requires the Qt::Application to be instantiated first, but will
        otherwise be picked up automatically. As long as the main program
        instantiates the application before any AnyEvent watchers are
        created, everything should just work.

           AnyEvent::Impl::Qt        based on Qt.

        Support for IO::Async can only be partial, as it is too broken and
        architecturally limited to even support the AnyEvent API. It also is
        the only event loop that needs the loop to be set explicitly, so it
        can only be used by a main program knowing about AnyEvent. See
        AnyEvent::Impl::Async for the gory details.

           AnyEvent::Impl::IOAsync   based on IO::Async, cannot be autoprobed.

    Event loops that are indirectly supported via other backends.
        Some event loops can be supported via other modules:

        There is no direct support for WxWidgets (Wx) or Prima.

        WxWidgets has no support for watching file handles. However, you can
        use WxWidgets through the POE adaptor, as POE has a Wx backend that
        simply polls 20 times per second, which was considered to be too
        horrible to even consider for AnyEvent.

        Prima is not supported as nobody seems to be using it, but it has a
        POE backend, so it can be supported through POE.

        AnyEvent knows about both Prima and Wx, however, and will try to
        load POE when detecting them, in the hope that POE will pick them
        up, in which case everything will be automatic.

GLOBAL VARIABLES AND FUNCTIONS
    These are not normally required to use AnyEvent, but can be useful to
    write AnyEvent extension modules.

    $AnyEvent::MODEL
        Contains "undef" until the first watcher is being created, before
        the backend has been autodetected.

        Afterwards it contains the event model that is being used, which is
        the name of the Perl class implementing the model. This class is
        usually one of the "AnyEvent::Impl:xxx" modules, but can be any
        other class in the case AnyEvent has been extended at runtime (e.g.
        in *rxvt-unicode* it will be "urxvt::anyevent").

    AnyEvent::detect
        Returns $AnyEvent::MODEL, forcing autodetection of the event model
        if necessary. You should only call this function right before you
        would have created an AnyEvent watcher anyway, that is, as late as
        possible at runtime, and not e.g. while initialising of your module.

        If you need to do some initialisation before AnyEvent watchers are
        created, use "post_detect".

    $guard = AnyEvent::post_detect { BLOCK }
        Arranges for the code block to be executed as soon as the event
        model is autodetected (or immediately if this has already happened).

        The block will be executed *after* the actual backend has been
        detected ($AnyEvent::MODEL is set), but *before* any watchers have
        been created, so it is possible to e.g. patch @AnyEvent::ISA or do
        other initialisations - see the sources of AnyEvent::Strict or
        AnyEvent::AIO to see how this is used.

        The most common usage is to create some global watchers, without
        forcing event module detection too early, for example, AnyEvent::AIO
        creates and installs the global IO::AIO watcher in a "post_detect"
        block to avoid autodetecting the event module at load time.

        If called in scalar or list context, then it creates and returns an
        object that automatically removes the callback again when it is
        destroyed. See Coro::BDB for a case where this is useful.

    @AnyEvent::post_detect
        If there are any code references in this array (you can "push" to it
        before or after loading AnyEvent), then they will called directly
        after the event loop has been chosen.

        You should check $AnyEvent::MODEL before adding to this array,
        though: if it is defined then the event loop has already been
        detected, and the array will be ignored.

        Best use "AnyEvent::post_detect { BLOCK }" when your application
        allows it,as it takes care of these details.

        This variable is mainly useful for modules that can do something
        useful when AnyEvent is used and thus want to know when it is
        initialised, but do not need to even load it by default. This array
        provides the means to hook into AnyEvent passively, without loading
        it.

WHAT TO DO IN A MODULE
    As a module author, you should "use AnyEvent" and call AnyEvent methods
    freely, but you should not load a specific event module or rely on it.

    Be careful when you create watchers in the module body - AnyEvent will
    decide which event module to use as soon as the first method is called,
    so by calling AnyEvent in your module body you force the user of your
    module to load the event module first.

    Never call "->recv" on a condition variable unless you *know* that the
    "->send" method has been called on it already. This is because it will
    stall the whole program, and the whole point of using events is to stay
    interactive.

    It is fine, however, to call "->recv" when the user of your module
    requests it (i.e. if you create a http request object ad have a method
    called "results" that returns the results, it should call "->recv"
    freely, as the user of your module knows what she is doing. always).

WHAT TO DO IN THE MAIN PROGRAM
    There will always be a single main program - the only place that should
    dictate which event model to use.

    If it doesn't care, it can just "use AnyEvent" and use it itself, or not
    do anything special (it does not need to be event-based) and let
    AnyEvent decide which implementation to chose if some module relies on
    it.

    If the main program relies on a specific event model - for example, in
    Gtk2 programs you have to rely on the Glib module - you should load the
    event module before loading AnyEvent or any module that uses it:
    generally speaking, you should load it as early as possible. The reason
    is that modules might create watchers when they are loaded, and AnyEvent
    will decide on the event model to use as soon as it creates watchers,
    and it might chose the wrong one unless you load the correct one
    yourself.

    You can chose to use a pure-perl implementation by loading the
    "AnyEvent::Impl::Perl" module, which gives you similar behaviour
    everywhere, but letting AnyEvent chose the model is generally better.

  MAINLOOP EMULATION
    Sometimes (often for short test scripts, or even standalone programs who
    only want to use AnyEvent), you do not want to run a specific event
    loop.

    In that case, you can use a condition variable like this:

       AnyEvent->condvar->recv;

    This has the effect of entering the event loop and looping forever.

    Note that usually your program has some exit condition, in which case it
    is better to use the "traditional" approach of storing a condition
    variable somewhere, waiting for it, and sending it when the program
    should exit cleanly.

OTHER MODULES
    The following is a non-exhaustive list of additional modules that use
    AnyEvent as a client and can therefore be mixed easily with other
    AnyEvent modules and other event loops in the same program. Some of the
    modules come with AnyEvent, most are available via CPAN.

    AnyEvent::Util
        Contains various utility functions that replace often-used but
        blocking functions such as "inet_aton" by event-/callback-based
        versions.

    AnyEvent::Socket
        Provides various utility functions for (internet protocol) sockets,
        addresses and name resolution. Also functions to create non-blocking
        tcp connections or tcp servers, with IPv6 and SRV record support and
        more.

    AnyEvent::Handle
        Provide read and write buffers, manages watchers for reads and
        writes, supports raw and formatted I/O, I/O queued and fully
        transparent and non-blocking SSL/TLS (via AnyEvent::TLS.

    AnyEvent::DNS
        Provides rich asynchronous DNS resolver capabilities.

    AnyEvent::HTTP
        A simple-to-use HTTP library that is capable of making a lot of
        concurrent HTTP requests.

    AnyEvent::HTTPD
        Provides a simple web application server framework.

    AnyEvent::FastPing
        The fastest ping in the west.

    AnyEvent::DBI
        Executes DBI requests asynchronously in a proxy process.

    AnyEvent::AIO
        Truly asynchronous I/O, should be in the toolbox of every event
        programmer. AnyEvent::AIO transparently fuses IO::AIO and AnyEvent
        together.

    AnyEvent::BDB
        Truly asynchronous Berkeley DB access. AnyEvent::BDB transparently
        fuses BDB and AnyEvent together.

    AnyEvent::GPSD
        A non-blocking interface to gpsd, a daemon delivering GPS
        information.

    AnyEvent::IRC
        AnyEvent based IRC client module family (replacing the older
        Net::IRC3).

    AnyEvent::XMPP
        AnyEvent based XMPP (Jabber protocol) module family (replacing the
        older Net::XMPP2>.

    AnyEvent::IGS
        A non-blocking interface to the Internet Go Server protocol (used by
        App::IGS).

    Net::FCP
        AnyEvent-based implementation of the Freenet Client Protocol,
        birthplace of AnyEvent.

    Event::ExecFlow
        High level API for event-based execution flow control.

    Coro
        Has special support for AnyEvent via Coro::AnyEvent.

ERROR AND EXCEPTION HANDLING
    In general, AnyEvent does not do any error handling - it relies on the
    caller to do that if required. The AnyEvent::Strict module (see also the
    "PERL_ANYEVENT_STRICT" environment variable, below) provides strict
    checking of all AnyEvent methods, however, which is highly useful during
    development.

    As for exception handling (i.e. runtime errors and exceptions thrown
    while executing a callback), this is not only highly event-loop
    specific, but also not in any way wrapped by this module, as this is the
    job of the main program.

    The pure perl event loop simply re-throws the exception (usually within
    "condvar->recv"), the Event and EV modules call "$Event/EV::DIED->()",
    Glib uses "install_exception_handler" and so on.

ENVIRONMENT VARIABLES
    The following environment variables are used by this module or its
    submodules.

    Note that AnyEvent will remove *all* environment variables starting with
    "PERL_ANYEVENT_" from %ENV when it is loaded while taint mode is
    enabled.

    "PERL_ANYEVENT_VERBOSE"
        By default, AnyEvent will be completely silent except in fatal
        conditions. You can set this environment variable to make AnyEvent
        more talkative.

        When set to 1 or higher, causes AnyEvent to warn about unexpected
        conditions, such as not being able to load the event model specified
        by "PERL_ANYEVENT_MODEL".

        When set to 2 or higher, cause AnyEvent to report to STDERR which
        event model it chooses.

    "PERL_ANYEVENT_STRICT"
        AnyEvent does not do much argument checking by default, as thorough
        argument checking is very costly. Setting this variable to a true
        value will cause AnyEvent to load "AnyEvent::Strict" and then to
        thoroughly check the arguments passed to most method calls. If it
        finds any problems, it will croak.

        In other words, enables "strict" mode.

        Unlike "use strict", it is definitely recommended to keep it off in
        production. Keeping "PERL_ANYEVENT_STRICT=1" in your environment
        while developing programs can be very useful, however.

    "PERL_ANYEVENT_MODEL"
        This can be used to specify the event model to be used by AnyEvent,
        before auto detection and -probing kicks in. It must be a string
        consisting entirely of ASCII letters. The string "AnyEvent::Impl::"
        gets prepended and the resulting module name is loaded and if the
        load was successful, used as event model. If it fails to load
        AnyEvent will proceed with auto detection and -probing.

        This functionality might change in future versions.

        For example, to force the pure perl model (AnyEvent::Impl::Perl) you
        could start your program like this:

           PERL_ANYEVENT_MODEL=Perl perl ...

    "PERL_ANYEVENT_PROTOCOLS"
        Used by both AnyEvent::DNS and AnyEvent::Socket to determine
        preferences for IPv4 or IPv6. The default is unspecified (and might
        change, or be the result of auto probing).

        Must be set to a comma-separated list of protocols or address
        families, current supported: "ipv4" and "ipv6". Only protocols
        mentioned will be used, and preference will be given to protocols
        mentioned earlier in the list.

        This variable can effectively be used for denial-of-service attacks
        against local programs (e.g. when setuid), although the impact is
        likely small, as the program has to handle conenction and other
        failures anyways.

        Examples: "PERL_ANYEVENT_PROTOCOLS=ipv4,ipv6" - prefer IPv4 over
        IPv6, but support both and try to use both.
        "PERL_ANYEVENT_PROTOCOLS=ipv4" - only support IPv4, never try to
        resolve or contact IPv6 addresses.
        "PERL_ANYEVENT_PROTOCOLS=ipv6,ipv4" support either IPv4 or IPv6, but
        prefer IPv6 over IPv4.

    "PERL_ANYEVENT_EDNS0"
        Used by AnyEvent::DNS to decide whether to use the EDNS0 extension
        for DNS. This extension is generally useful to reduce DNS traffic,
        but some (broken) firewalls drop such DNS packets, which is why it
        is off by default.

        Setting this variable to 1 will cause AnyEvent::DNS to announce
        EDNS0 in its DNS requests.

    "PERL_ANYEVENT_MAX_FORKS"
        The maximum number of child processes that
        "AnyEvent::Util::fork_call" will create in parallel.

    "PERL_ANYEVENT_MAX_OUTSTANDING_DNS"
        The default value for the "max_outstanding" parameter for the
        default DNS resolver - this is the maximum number of parallel DNS
        requests that are sent to the DNS server.

    "PERL_ANYEVENT_RESOLV_CONF"
        The file to use instead of /etc/resolv.conf (or OS-specific
        configuration) in the default resolver. When set to the empty
        string, no default config will be used.

    "PERL_ANYEVENT_CA_FILE", "PERL_ANYEVENT_CA_PATH".
        When neither "ca_file" nor "ca_path" was specified during
        AnyEvent::TLS context creation, and either of these environment
        variables exist, they will be used to specify CA certificate
        locations instead of a system-dependent default.

SUPPLYING YOUR OWN EVENT MODEL INTERFACE
    This is an advanced topic that you do not normally need to use AnyEvent
    in a module. This section is only of use to event loop authors who want
    to provide AnyEvent compatibility.

    If you need to support another event library which isn't directly
    supported by AnyEvent, you can supply your own interface to it by
    pushing, before the first watcher gets created, the package name of the
    event module and the package name of the interface to use onto
    @AnyEvent::REGISTRY. You can do that before and even without loading
    AnyEvent, so it is reasonably cheap.

    Example:

       push @AnyEvent::REGISTRY, [urxvt => urxvt::anyevent::];

    This tells AnyEvent to (literally) use the "urxvt::anyevent::"
    package/class when it finds the "urxvt" package/module is already
    loaded.

    When AnyEvent is loaded and asked to find a suitable event model, it
    will first check for the presence of urxvt by trying to "use" the
    "urxvt::anyevent" module.

    The class should provide implementations for all watcher types. See
    AnyEvent::Impl::EV (source code), AnyEvent::Impl::Glib (Source code) and
    so on for actual examples. Use "perldoc -m AnyEvent::Impl::Glib" to see
    the sources.

    If you don't provide "signal" and "child" watchers than AnyEvent will
    provide suitable (hopefully) replacements.

    The above example isn't fictitious, the *rxvt-unicode* (a.k.a. urxvt)
    terminal emulator uses the above line as-is. An interface isn't included
    in AnyEvent because it doesn't make sense outside the embedded
    interpreter inside *rxvt-unicode*, and it is updated and maintained as
    part of the *rxvt-unicode* distribution.

    *rxvt-unicode* also cheats a bit by not providing blocking access to
    condition variables: code blocking while waiting for a condition will
    "die". This still works with most modules/usages, and blocking calls
    must not be done in an interactive application, so it makes sense.

EXAMPLE PROGRAM
    The following program uses an I/O watcher to read data from STDIN, a
    timer to display a message once per second, and a condition variable to
    quit the program when the user enters quit:

       use AnyEvent;

       my $cv = AnyEvent->condvar;

       my $io_watcher = AnyEvent->io (
          fh   => \*STDIN,
          poll => 'r',
          cb   => sub {
             warn "io event <$_[0]>\n";   # will always output <r>
             chomp (my $input = <STDIN>); # read a line
             warn "read: $input\n";       # output what has been read
             $cv->send if $input =~ /^q/i; # quit program if /^q/i
          },
       );

       my $time_watcher; # can only be used once

       sub new_timer {
          $timer = AnyEvent->timer (after => 1, cb => sub {
             warn "timeout\n"; # print 'timeout' about every second
             &new_timer; # and restart the time
          });
       }

       new_timer; # create first timer

       $cv->recv; # wait until user enters /^q/i

REAL-WORLD EXAMPLE
    Consider the Net::FCP module. It features (among others) the following
    API calls, which are to freenet what HTTP GET requests are to http:

       my $data = $fcp->client_get ($url); # blocks

       my $transaction = $fcp->txn_client_get ($url); # does not block
       $transaction->cb ( sub { ... } ); # set optional result callback
       my $data = $transaction->result; # possibly blocks

    The "client_get" method works like "LWP::Simple::get": it requests the
    given URL and waits till the data has arrived. It is defined to be:

       sub client_get { $_[0]->txn_client_get ($_[1])->result }

    And in fact is automatically generated. This is the blocking API of
    Net::FCP, and it works as simple as in any other, similar, module.

    More complicated is "txn_client_get": It only creates a transaction
    (completion, result, ...) object and initiates the transaction.

       my $txn = bless { }, Net::FCP::Txn::;

    It also creates a condition variable that is used to signal the
    completion of the request:

       $txn->{finished} = AnyAvent->condvar;

    It then creates a socket in non-blocking mode.

       socket $txn->{fh}, ...;
       fcntl $txn->{fh}, F_SETFL, O_NONBLOCK;
       connect $txn->{fh}, ...
          and !$!{EWOULDBLOCK}
          and !$!{EINPROGRESS}
          and Carp::croak "unable to connect: $!\n";

    Then it creates a write-watcher which gets called whenever an error
    occurs or the connection succeeds:

       $txn->{w} = AnyEvent->io (fh => $txn->{fh}, poll => 'w', cb => sub { $txn->fh_ready_w });

    And returns this transaction object. The "fh_ready_w" callback gets
    called as soon as the event loop detects that the socket is ready for
    writing.

    The "fh_ready_w" method makes the socket blocking again, writes the
    request data and replaces the watcher by a read watcher (waiting for
    reply data). The actual code is more complicated, but that doesn't
    matter for this example:

       fcntl $txn->{fh}, F_SETFL, 0;
       syswrite $txn->{fh}, $txn->{request}
          or die "connection or write error";
       $txn->{w} = AnyEvent->io (fh => $txn->{fh}, poll => 'r', cb => sub { $txn->fh_ready_r });

    Again, "fh_ready_r" waits till all data has arrived, and then stores the
    result and signals any possible waiters that the request has finished:

       sysread $txn->{fh}, $txn->{buf}, length $txn->{$buf};

       if (end-of-file or data complete) {
         $txn->{result} = $txn->{buf};
         $txn->{finished}->send;
         $txb->{cb}->($txn) of $txn->{cb}; # also call callback
       }

    The "result" method, finally, just waits for the finished signal (if the
    request was already finished, it doesn't wait, of course, and returns
    the data:

       $txn->{finished}->recv;
       return $txn->{result};

    The actual code goes further and collects all errors ("die"s,
    exceptions) that occurred during request processing. The "result" method
    detects whether an exception as thrown (it is stored inside the $txn
    object) and just throws the exception, which means connection errors and
    other problems get reported tot he code that tries to use the result,
    not in a random callback.

    All of this enables the following usage styles:

    1. Blocking:

       my $data = $fcp->client_get ($url);

    2. Blocking, but running in parallel:

       my @datas = map $_->result,
                      map $fcp->txn_client_get ($_),
                         @urls;

    Both blocking examples work without the module user having to know
    anything about events.

    3a. Event-based in a main program, using any supported event module:

       use EV;

       $fcp->txn_client_get ($url)->cb (sub {
          my $txn = shift;
          my $data = $txn->result;
          ...
       });

       EV::loop;

    3b. The module user could use AnyEvent, too:

       use AnyEvent;

       my $quit = AnyEvent->condvar;

       $fcp->txn_client_get ($url)->cb (sub {
          ...
          $quit->send;
       });

       $quit->recv;

BENCHMARKS
    To give you an idea of the performance and overheads that AnyEvent adds
    over the event loops themselves and to give you an impression of the
    speed of various event loops I prepared some benchmarks.

  BENCHMARKING ANYEVENT OVERHEAD
    Here is a benchmark of various supported event models used natively and
    through AnyEvent. The benchmark creates a lot of timers (with a zero
    timeout) and I/O watchers (watching STDOUT, a pty, to become writable,
    which it is), lets them fire exactly once and destroys them again.

    Source code for this benchmark is found as eg/bench in the AnyEvent
    distribution.

   Explanation of the columns
    *watcher* is the number of event watchers created/destroyed. Since
    different event models feature vastly different performances, each event
    loop was given a number of watchers so that overall runtime is
    acceptable and similar between tested event loop (and keep them from
    crashing): Glib would probably take thousands of years if asked to
    process the same number of watchers as EV in this benchmark.

    *bytes* is the number of bytes (as measured by the resident set size,
    RSS) consumed by each watcher. This method of measuring captures both C
    and Perl-based overheads.

    *create* is the time, in microseconds (millionths of seconds), that it
    takes to create a single watcher. The callback is a closure shared
    between all watchers, to avoid adding memory overhead. That means
    closure creation and memory usage is not included in the figures.

    *invoke* is the time, in microseconds, used to invoke a simple callback.
    The callback simply counts down a Perl variable and after it was invoked
    "watcher" times, it would "->send" a condvar once to signal the end of
    this phase.

    *destroy* is the time, in microseconds, that it takes to destroy a
    single watcher.

   Results
              name watchers bytes create invoke destroy comment
             EV/EV   400000   224   0.47   0.35    0.27 EV native interface
            EV/Any   100000   224   2.88   0.34    0.27 EV + AnyEvent watchers
        CoroEV/Any   100000   224   2.85   0.35    0.28 coroutines + Coro::Signal
          Perl/Any   100000   452   4.13   0.73    0.95 pure perl implementation
       Event/Event    16000   517  32.20  31.80    0.81 Event native interface
         Event/Any    16000   590  35.85  31.55    1.06 Event + AnyEvent watchers
       IOAsync/Any    16000   989  38.10  32.77   11.13 via IO::Async::Loop::IO_Poll
       IOAsync/Any    16000   990  37.59  29.50   10.61 via IO::Async::Loop::Epoll
          Glib/Any    16000  1357 102.33  12.31   51.00 quadratic behaviour
            Tk/Any     2000  1860  27.20  66.31   14.00 SEGV with >> 2000 watchers
         POE/Event     2000  6328 109.99 751.67   14.02 via POE::Loop::Event
        POE/Select     2000  6027  94.54 809.13  579.80 via POE::Loop::Select

   Discussion
    The benchmark does *not* measure scalability of the event loop very
    well. For example, a select-based event loop (such as the pure perl one)
    can never compete with an event loop that uses epoll when the number of
    file descriptors grows high. In this benchmark, all events become ready
    at the same time, so select/poll-based implementations get an unnatural
    speed boost.

    Also, note that the number of watchers usually has a nonlinear effect on
    overall speed, that is, creating twice as many watchers doesn't take
    twice the time - usually it takes longer. This puts event loops tested
    with a higher number of watchers at a disadvantage.

    To put the range of results into perspective, consider that on the
    benchmark machine, handling an event takes roughly 1600 CPU cycles with
    EV, 3100 CPU cycles with AnyEvent's pure perl loop and almost 3000000
    CPU cycles with POE.

    "EV" is the sole leader regarding speed and memory use, which are both
    maximal/minimal, respectively. Even when going through AnyEvent, it uses
    far less memory than any other event loop and is still faster than Event
    natively.

    The pure perl implementation is hit in a few sweet spots (both the
    constant timeout and the use of a single fd hit optimisations in the
    perl interpreter and the backend itself). Nevertheless this shows that
    it adds very little overhead in itself. Like any select-based backend
    its performance becomes really bad with lots of file descriptors (and
    few of them active), of course, but this was not subject of this
    benchmark.

    The "Event" module has a relatively high setup and callback invocation
    cost, but overall scores in on the third place.

    "IO::Async" performs admirably well, about on par with "Event", even
    when using its pure perl backend.

    "Glib"'s memory usage is quite a bit higher, but it features a faster
    callback invocation and overall ends up in the same class as "Event".
    However, Glib scales extremely badly, doubling the number of watchers
    increases the processing time by more than a factor of four, making it
    completely unusable when using larger numbers of watchers (note that
    only a single file descriptor was used in the benchmark, so
    inefficiencies of "poll" do not account for this).

    The "Tk" adaptor works relatively well. The fact that it crashes with
    more than 2000 watchers is a big setback, however, as correctness takes
    precedence over speed. Nevertheless, its performance is surprising, as
    the file descriptor is dup()ed for each watcher. This shows that the
    dup() employed by some adaptors is not a big performance issue (it does
    incur a hidden memory cost inside the kernel which is not reflected in
    the figures above).

    "POE", regardless of underlying event loop (whether using its pure perl
    select-based backend or the Event module, the POE-EV backend couldn't be
    tested because it wasn't working) shows abysmal performance and memory
    usage with AnyEvent: Watchers use almost 30 times as much memory as EV
    watchers, and 10 times as much memory as Event (the high memory
    requirements are caused by requiring a session for each watcher).
    Watcher invocation speed is almost 900 times slower than with AnyEvent's
    pure perl implementation.

    The design of the POE adaptor class in AnyEvent can not really account
    for the performance issues, though, as session creation overhead is
    small compared to execution of the state machine, which is coded pretty
    optimally within AnyEvent::Impl::POE (and while everybody agrees that
    using multiple sessions is not a good approach, especially regarding
    memory usage, even the author of POE could not come up with a faster
    design).

   Summary
    *   Using EV through AnyEvent is faster than any other event loop (even
        when used without AnyEvent), but most event loops have acceptable
        performance with or without AnyEvent.

    *   The overhead AnyEvent adds is usually much smaller than the overhead
        of the actual event loop, only with extremely fast event loops such
        as EV adds AnyEvent significant overhead.

    *   You should avoid POE like the plague if you want performance or
        reasonable memory usage.

  BENCHMARKING THE LARGE SERVER CASE
    This benchmark actually benchmarks the event loop itself. It works by
    creating a number of "servers": each server consists of a socket pair, a
    timeout watcher that gets reset on activity (but never fires), and an
    I/O watcher waiting for input on one side of the socket. Each time the
    socket watcher reads a byte it will write that byte to a random other
    "server".

    The effect is that there will be a lot of I/O watchers, only part of
    which are active at any one point (so there is a constant number of
    active fds for each loop iteration, but which fds these are is random).
    The timeout is reset each time something is read because that reflects
    how most timeouts work (and puts extra pressure on the event loops).

    In this benchmark, we use 10000 socket pairs (20000 sockets), of which
    100 (1%) are active. This mirrors the activity of large servers with
    many connections, most of which are idle at any one point in time.

    Source code for this benchmark is found as eg/bench2 in the AnyEvent
    distribution.

   Explanation of the columns
    *sockets* is the number of sockets, and twice the number of "servers"
    (as each server has a read and write socket end).

    *create* is the time it takes to create a socket pair (which is
    nontrivial) and two watchers: an I/O watcher and a timeout watcher.

    *request*, the most important value, is the time it takes to handle a
    single "request", that is, reading the token from the pipe and
    forwarding it to another server. This includes deleting the old timeout
    and creating a new one that moves the timeout into the future.

   Results
         name sockets create  request 
           EV   20000  69.01    11.16 
         Perl   20000  73.32    35.87 
      IOAsync   20000 157.00    98.14 epoll
      IOAsync   20000 159.31   616.06 poll
        Event   20000 212.62   257.32 
         Glib   20000 651.16  1896.30 
          POE   20000 349.67 12317.24 uses POE::Loop::Event

   Discussion
    This benchmark *does* measure scalability and overall performance of the
    particular event loop.

    EV is again fastest. Since it is using epoll on my system, the setup
    time is relatively high, though.

    Perl surprisingly comes second. It is much faster than the C-based event
    loops Event and Glib.

    IO::Async performs very well when using its epoll backend, and still
    quite good compared to Glib when using its pure perl backend.

    Event suffers from high setup time as well (look at its code and you
    will understand why). Callback invocation also has a high overhead
    compared to the "$_->() for .."-style loop that the Perl event loop
    uses. Event uses select or poll in basically all documented
    configurations.

    Glib is hit hard by its quadratic behaviour w.r.t. many watchers. It
    clearly fails to perform with many filehandles or in busy servers.

    POE is still completely out of the picture, taking over 1000 times as
    long as EV, and over 100 times as long as the Perl implementation, even
    though it uses a C-based event loop in this case.

   Summary
    *   The pure perl implementation performs extremely well.

    *   Avoid Glib or POE in large projects where performance matters.

  BENCHMARKING SMALL SERVERS
    While event loops should scale (and select-based ones do not...) even to
    large servers, most programs we (or I :) actually write have only a few
    I/O watchers.

    In this benchmark, I use the same benchmark program as in the large
    server case, but it uses only eight "servers", of which three are active
    at any one time. This should reflect performance for a small server
    relatively well.

    The columns are identical to the previous table.

   Results
        name sockets create request 
          EV      16  20.00    6.54 
        Perl      16  25.75   12.62 
       Event      16  81.27   35.86 
        Glib      16  32.63   15.48 
         POE      16 261.87  276.28 uses POE::Loop::Event

   Discussion
    The benchmark tries to test the performance of a typical small server.
    While knowing how various event loops perform is interesting, keep in
    mind that their overhead in this case is usually not as important, due
    to the small absolute number of watchers (that is, you need efficiency
    and speed most when you have lots of watchers, not when you only have a
    few of them).

    EV is again fastest.

    Perl again comes second. It is noticeably faster than the C-based event
    loops Event and Glib, although the difference is too small to really
    matter.

    POE also performs much better in this case, but is is still far behind
    the others.

   Summary
    *   C-based event loops perform very well with small number of watchers,
        as the management overhead dominates.

  THE IO::Lambda BENCHMARK
    Recently I was told about the benchmark in the IO::Lambda manpage, which
    could be misinterpreted to make AnyEvent look bad. In fact, the
    benchmark simply compares IO::Lambda with POE, and IO::Lambda looks
    better (which shouldn't come as a surprise to anybody). As such, the
    benchmark is fine, and mostly shows that the AnyEvent backend from
    IO::Lambda isn't very optimal. But how would AnyEvent compare when used
    without the extra baggage? To explore this, I wrote the equivalent
    benchmark for AnyEvent.

    The benchmark itself creates an echo-server, and then, for 500 times,
    connects to the echo server, sends a line, waits for the reply, and then
    creates the next connection. This is a rather bad benchmark, as it
    doesn't test the efficiency of the framework or much non-blocking I/O,
    but it is a benchmark nevertheless.

       name                    runtime
       Lambda/select           0.330 sec
          + optimized          0.122 sec
       Lambda/AnyEvent         0.327 sec
          + optimized          0.138 sec
       Raw sockets/select      0.077 sec
       POE/select, components  0.662 sec
       POE/select, raw sockets 0.226 sec
       POE/select, optimized   0.404 sec

       AnyEvent/select/nb      0.085 sec
       AnyEvent/EV/nb          0.068 sec
          +state machine       0.134 sec

    The benchmark is also a bit unfair (my fault): the IO::Lambda/POE
    benchmarks actually make blocking connects and use 100% blocking I/O,
    defeating the purpose of an event-based solution. All of the newly
    written AnyEvent benchmarks use 100% non-blocking connects (using
    AnyEvent::Socket::tcp_connect and the asynchronous pure perl DNS
    resolver), so AnyEvent is at a disadvantage here, as non-blocking
    connects generally require a lot more bookkeeping and event handling
    than blocking connects (which involve a single syscall only).

    The last AnyEvent benchmark additionally uses AnyEvent::Handle, which
    offers similar expressive power as POE and IO::Lambda, using
    conventional Perl syntax. This means that both the echo server and the
    client are 100% non-blocking, further placing it at a disadvantage.

    As you can see, the AnyEvent + EV combination even beats the
    hand-optimised "raw sockets benchmark", while AnyEvent + its pure perl
    backend easily beats IO::Lambda and POE.

    And even the 100% non-blocking version written using the high-level (and
    slow :) AnyEvent::Handle abstraction beats both POE and IO::Lambda by a
    large margin, even though it does all of DNS, tcp-connect and socket I/O
    in a non-blocking way.

    The two AnyEvent benchmarks programs can be found as eg/ae0.pl and
    eg/ae2.pl in the AnyEvent distribution, the remaining benchmarks are
    part of the IO::lambda distribution and were used without any changes.

SIGNALS
    AnyEvent currently installs handlers for these signals:

    SIGCHLD
        A handler for "SIGCHLD" is installed by AnyEvent's child watcher
        emulation for event loops that do not support them natively. Also,
        some event loops install a similar handler.

        Additionally, when AnyEvent is loaded and SIGCHLD is set to IGNORE,
        then AnyEvent will reset it to default, to avoid losing child exit
        statuses.

    SIGPIPE
        A no-op handler is installed for "SIGPIPE" when $SIG{PIPE} is
        "undef" when AnyEvent gets loaded.

        The rationale for this is that AnyEvent users usually do not really
        depend on SIGPIPE delivery (which is purely an optimisation for
        shell use, or badly-written programs), but "SIGPIPE" can cause
        spurious and rare program exits as a lot of people do not expect
        "SIGPIPE" when writing to some random socket.

        The rationale for installing a no-op handler as opposed to ignoring
        it is that this way, the handler will be restored to defaults on
        exec.

        Feel free to install your own handler, or reset it to defaults.

FORK
    Most event libraries are not fork-safe. The ones who are usually are
    because they rely on inefficient but fork-safe "select" or "poll" calls.
    Only EV is fully fork-aware.

    If you have to fork, you must either do so *before* creating your first
    watcher OR you must not use AnyEvent at all in the child.

SECURITY CONSIDERATIONS
    AnyEvent can be forced to load any event model via
    $ENV{PERL_ANYEVENT_MODEL}. While this cannot (to my knowledge) be used
    to execute arbitrary code or directly gain access, it can easily be used
    to make the program hang or malfunction in subtle ways, as AnyEvent
    watchers will not be active when the program uses a different event
    model than specified in the variable.

    You can make AnyEvent completely ignore this variable by deleting it
    before the first watcher gets created, e.g. with a "BEGIN" block:

       BEGIN { delete $ENV{PERL_ANYEVENT_MODEL} }
  
       use AnyEvent;

    Similar considerations apply to $ENV{PERL_ANYEVENT_VERBOSE}, as that can
    be used to probe what backend is used and gain other information (which
    is probably even less useful to an attacker than PERL_ANYEVENT_MODEL),
    and $ENV{PERL_ANYEVENT_STRICT}.

    Note that AnyEvent will remove *all* environment variables starting with
    "PERL_ANYEVENT_" from %ENV when it is loaded while taint mode is
    enabled.

BUGS
    Perl 5.8 has numerous memleaks that sometimes hit this module and are
    hard to work around. If you suffer from memleaks, first upgrade to Perl
    5.10 and check wether the leaks still show up. (Perl 5.10.0 has other
    annoying memleaks, such as leaking on "map" and "grep" but it is usually
    not as pronounced).

SEE ALSO
    Utility functions: AnyEvent::Util.

    Event modules: EV, EV::Glib, Glib::EV, Event, Glib::Event, Glib, Tk,
    Event::Lib, Qt, POE.

    Implementations: AnyEvent::Impl::EV, AnyEvent::Impl::Event,
    AnyEvent::Impl::Glib, AnyEvent::Impl::Tk, AnyEvent::Impl::Perl,
    AnyEvent::Impl::EventLib, AnyEvent::Impl::Qt, AnyEvent::Impl::POE,
    AnyEvent::Impl::IOAsync.

    Non-blocking file handles, sockets, TCP clients and servers:
    AnyEvent::Handle, AnyEvent::Socket, AnyEvent::TLS.

    Asynchronous DNS: AnyEvent::DNS.

    Coroutine support: Coro, Coro::AnyEvent, Coro::EV, Coro::Event,

    Nontrivial usage examples: AnyEvent::GPSD, AnyEvent::XMPP,
    AnyEvent::HTTP.

AUTHOR
       Marc Lehmann <schmorp@schmorp.de>
       http://home.schmorp.de/

Revision:	1.45
Committed:	Fri Jul 17 14:57:03 2009 UTC (14 years, 10 months ago) by root
Branch:	MAIN
CVS Tags:	rel-4_83
Changes since 1.44:	+28 -31 lines
Log Message:	4.83