Coro(3pm)

📦
version 6.570 · 7 versions
, in debian trixie, ubuntu resolute

✨ Coro(3pm) ✨

User Contributed Perl Documentation, but make it iconic. This is the Coro(3pm) manual page, bestie, and we're diving in. 💖

🌸 NAME 🌸

Coro — the only real threads in perl 💅

🦋 SYNOPSIS 🦋

  use Coro;

  async {
     # some asynchronous thread of execution
     print "2\n";
     cede; # yield back to main
     print "4\n";
  };
  print "1\n";
  cede; # yield to coro
  print "3\n";
  cede; # and again

  # use locking
  my $lock = new Coro::Semaphore;
  my $locked;

  $lock->down;
  $locked = 1;
  $lock->up;

💖 DESCRIPTION 💖

Okay so, if you want a tutorial-style intro, bestie, go read the Coro::Intro manpage. This manpage? It's mainly the reference info, no cap. ✨

iykyk: Coro::Intro is the vibe tutorial. This page is the reference, periodt.

This module collection manages continuations in general, most often in the form of cooperative threads (also called coros, or just "coro" in the docs). They're similar to kernel threads, but they don't (in general) run in parallel at the same time, even on SMP machines. The specific flavor of thread this module gives you also guarantees you that it will not switch between threads unless necessary, at easily-identified points in your program. So locking and parallel access are rarely an issue, which makes thread programming much safer and easier than using other thread models. 🫶

Unlike the so-called "Perl threads" (which are not actually real threads but only the windows process emulation (see section of same name for more details) ported to UNIX, and as such act as processes), Coro provides a full shared address space, which makes communication between threads very easy. And coro threads are fast, too: disabling the Windows process emulation code in your perl and using Coro can easily result in a two to four times speed increase for your programs. A parallel matrix multiplication benchmark (very communication-intensive) runs over 300 times faster on a single core than perls pseudo-threads on a quad core using all four cores. 💀🔥

Coro achieves that by supporting multiple running interpreters that share data, which is especially useful to code pseudo-parallel processes and for event-based programming, like multiple HTTP-GET requests running concurrently. See Coro::AnyEvent to learn more on how to integrate Coro into an event-based environment. 🌈

In this module, a thread is defined as "callchain + lexical variables + some package variables + C stack), that is, a thread has its own callchain, its own set of lexicals and its own set of perls most important global variables (see Coro::State for more configuration and background info). 🧚

See also the "SEE ALSO" section at the end of this document — the Coro module family is quite large. 👑

💫 CORO THREAD LIFE CYCLE 💫

okay so bestie, during the long and exciting (or, like, not) life of a coro thread, it passes through a whole bunch of states:

1. Creation 🐣

the first thing in the life of a coro thread is its creation — obviously. the typical way to make a thread is calling the async BLOCK function:

async {
   # thread code goes here
};

you can also pass arguments, which get dropped into @_:

async {
   print $_[1]; # prints 2
} 1, 2, 3;

this creates a new coro thread and puts it into the ready queue, meaning it will run the second the CPU is free for it. slay.

async hands you back a Coro object — stash it for later reference or just ignore it, whatever — a thread that is running, ready to run or waiting for some event is alive all on its own. no babysitting.

another way to create a thread is calling the new constructor with a code-reference:

new Coro sub {
   # thread code goes here
}, @optional_arguments;

this is pretty similar to calling async, but here's the plot twist that matters: the new thread is not put into the ready queue, so it won't run until somebody puts it there. which means async is identical to this sequence:

my $coro = new Coro sub {
   # thread code goes here
};
$coro->ready;
return $coro;

2. Startup 🌱

when a new coro thread is created, only a copy of the code reference and the arguments get stored — no extra memory for stacks and so on is allocated, which keeps the coro thread in a low-memory state. frugal queen 💅

only when it actually starts executing will all the resources finally be allocated.

and the optional arguments you specified at coro creation? they're available in @_, just like with function calls.

3. Running / Blocking 🏃‍♀️💨

a LOT can happen after the coro thread has started running. quite usually it won't run all the way to the end in one go (bc honestly you could've used a function at that point), it'll give up the CPU regularly bc it's waiting for external events.

as long as a coro thread runs, its Coro object is available in the global variable $Coro::current.

the low-level way to give up the CPU is calling the scheduler, which selects a new coro thread to run:

Coro::schedule;

since running threads are not in the ready queue, calling the scheduler without doing anything else will block the coro thread forever 😭 — you need to arrange either for the coro to get woken up (readied) by some other event or some other thread, or you can put it into the ready queue before scheduling:

# this is exactly what Coro::cede does
$Coro::current->ready;
Coro::schedule;

all the higher-level synchronisation methods (Coro::Semaphore, Coro::rouse_*...) are actually implemented via ->ready and Coro::schedule. no cap.

while the coro thread is running it also might get assigned a C-level thread, or the C-level thread might be unassigned from it, as the Coro runtime wishes. a C-level thread needs to be assigned when your perl thread calls into some C-level function and that function in turn calls perl and perl then wants to switch coroutines. this happens most often when you run an event loop and block in the callback, or when perl itself calls some function such as AUTOLOAD or methods via the tie mechanism.

4. Termination 💀

many threads actually terminate after some time. there are a bunch of ways to terminate a coro thread, and the simplest is returning from the top-level code reference:

async {
   # after returning from here, the coro thread is terminated
};

async {
   return if 0.5 <  rand; # terminate a little earlier, maybe
   print "got a chance to print this\n";
   # or here
};

any values returned from the coroutine can be recovered using ->join:

my $coro = async {
   "hello, world\n" # return a string
};

my $hello_world = $coro->join;

print $hello_world;

another way to terminate is calling Coro::terminate, which works at any subroutine call nesting level:

async {
   Coro::terminate "return value 1", "return value 2";
};

yet another way is to ->cancel (or ->safe_cancel) the coro thread from another thread:

my $coro = async {
   exit 1;
};

$coro->cancel; # also accepts values for ->join to retrieve
⚠️ cancellation can be dangerous — it's a bit like calling exit without actually exiting, and it might leave C libraries and XS modules in a weird state. be careful bestie.

unlike other thread implementations, though, Coro is exceptionally safe with regards to cancellation, bc perl will always be in a consistent state. and for those cases where you want to do truly marvellous things with your coro while it's being cancelled — that is, make sure all cleanup code is executed from the thread being cancelled — there's even a ->safe_cancel method. iconic. 🎀

so, cancelling a thread that runs in an XS event loop might not be the best idea, but any other combination that deals with perl only (cancelling when a thread is in a tie method or an AUTOLOAD for example) is safe.

last not least, a coro thread object that isn't referenced is ->cancel'ed automatically — just like other objects in Perl. this is not such a common case, however — a running thread is referenced by $Coro::current, a thread ready to run is referenced by the ready queue, a thread waiting on a lock or semaphore is referenced by being in some wait list and so on. but a thread that isn't in any of those queues gets cancelled:

async {
   schedule; # cede to other coros, don't go into the ready queue
};

cede;
# now the async above is destroyed, as it is not referenced by anything.

a slightly embellished example might make it clearer:

async {
   my $guard = Guard::guard { print "destroyed\n" };
   schedule while 1;
};

cede;

superficially you might not expect any output — since the async implements an endless loop, the $guard will not be cleaned up. HOWEVER, since the thread object returned by async is not stored anywhere, the thread is initially referenced because it is in the ready queue, when it runs it is referenced by $Coro::current, but when it calls schedule, it gets canceled, causing the guard object to be destroyed (see the next section), and printing its message. 😳

if this seems a bit drastic, remember that this only happens when nothing references the thread anymore, which means there is no way to further execute it, ever. the only options at this point are leaking the thread, or cleaning it up, which brings us to...

5. Cleanup 🧹

threads will allocate various resources. most but not all will be returned when a thread terminates, during clean-up.

cleanup is quite similar to throwing an uncaught exception: perl will work its way up through all subroutine calls and blocks. on its way, it will release all my variables, undo all local's and free any other resources truly local to the thread.

so, a common way to free resources is to keep them referenced only by my variables:

async {
   my $big_cache = new Cache ...;
};

if there are no other references, then the $big_cache object will be freed when the thread terminates, regardless of how it does so.

what it does NOT do is unlock any Coro::Semaphores or similar resources, but that's where the guard methods come in handy:

my $sem = new Coro::Semaphore;

async {
   my $lock_guard = $sem->guard;
   # if we return, or die or get cancelled, here,
   # then the semaphore will be "up"ed.
};

the Guard::guard function comes in handy for any custom cleanup you might want to do (but you cannot switch to other coroutines from those code blocks):

async {
   my $window = new Gtk2::Window "toplevel";
   # The window will not be cleaned up automatically, even when $window
   # gets freed, so use a guard to ensure its destruction
   # in case of an error:
   my $window_guard = Guard::guard { $window->destroy };

   # we are safe here
};

last not least, local can often be handy, too, e.g. when temporarily replacing the coro thread description:

sub myfunction {
   local $Coro::current->{desc} = "inside myfunction(@_)";

   # if we return or die here, the description will be restored
}

6. Viva La Zombie Muerte 🧟‍♀️

even after a thread has terminated and cleaned up its resources, the Coro object is still there and stores the return values of the thread.

when there are no other references, it will simply be cleaned up and freed.

if there are any references, the Coro object will stay around, and you can call ->join as many times as you wish to retrieve the result values:

async {
   print "hi\n";
   1
};

# run the async above, and free everything before returning
# from Coro::cede:
Coro::cede;

{
   my $coro = async {
      print "hi\n";
      1
   };

   # run the async above, and clean up, but do not free the coro
   # object:
   Coro::cede;

   # optionally retrieve the result values
   my @results = $coro->join;

   # now $coro goes out of scope, and presumably gets freed
};

🌸 GLOBAL VARIABLES 🌸

  • $Coro::main — this one stores the Coro object that reps the main program itself. You can ready it, and honestly you can do most of the coro things to it, but like… its main character job is letting you compare it against $Coro::current so you can figure out whether you're vibing in the main program or not. 💅
  • $Coro::current — the Coro object for the coro you're in right now (aka the last coro the Coro scheduler switched to). Its starting value is $Coro::main, obviously bestie, who else would it be. 🫶

This variable is strictly read-only. You can totally take copies of whatever value is chilling in there and use those copies like any other Coro object, but do not go modifying the variable itself. Hands off. 🙈

  • $Coro::idle — lowkey the glue for hooking Coro into event loops. Ngl it's usually way better to just lean on Coro::AnyEvent or Coro::EV, because this is pretty low-level functionality. You've been warned. 💀

It stores a Coro object that gets dropped into the ready queue when there are no other ready threads (and no ready hooks get invoked either).

The default implementation literally dies with "FATAL: deadlock detected.", followed by a thread listing, because the program genuinely has no other way to keep going. 😭 Not dramatic, just accurate.

This hook gets overwritten by modules like "Coro::EV" and "Coro::AnyEvent" so they can wait on an external event that will hopefully wake up a coro and let the scheduler run it.

Peep Coro::EV or Coro::AnyEvent for examples of this technique in action. ✨

🦋 SIMPLE CORO CREATION 🦋

  • async { ... } [@args...] — makes a brand new coro and hands you back its Coro object (which you'll usually just… ignore, tbh). The coro gets yeeted into the ready queue, so it starts running all by itself on the next scheduler run. Iconic behaviour. 👑

The first argument is a codeblock/closure that gets executed inside the coro. The moment it returns, the coro is automatically terminated. Slay.

Any remaining arguments get passed along as arguments to the closure.

For the tea on the coro environment that coros actually execute in, check the "Coro::State::new" constructor.

Calling "exit" inside a coro does exactly the same thing as calling exit outside of it. Same energy, same result. And likewise, if the coro dies, the program exits, just like it would in the main program.

Not feeling that? Then give it a default "die" handler, or just, like, avoid dying entirely (hello "eval").

Example: create a new coro that just prints its arguments. 🌈

async {
   print "@_\n";
} 1,2,3,4;
  • async_pool { ... } [@args...] — basically "async" but it pulls from a coro pool, so you shouldn't call terminate or join on it (though you are allowed to, no one's stopping you), and you get a coro that might have already run other code before (which can be a good or a bad thing :).

On the plus side, this function is about twice as fast as creating (and destroying) a completely brand-new coro, so if you need a whole bunch of generic coros in quick succession, use "async_pool", not "async". Fr fr. 🔥

The code block runs in an "eval" context, and a warning gets issued if there's an exception instead of the program just terminating like "async" does. And since the coro is being reused, stuff like "on_destroy" won't work the way you'd expect — unless you call terminate or cancel, which kinda defeats the whole point of pooling (but it's fine for the exceptional case).

After every run: the priority gets reset to 0, all "swap_sv" calls get undone, tracing gets disabled, the description gets reset, and the default output filehandle gets restored. So you can change all of those. Otherwise the coro gets reused "as-is" — most importantly, if you change other per-coro global stuff like $/, you must revert that change, and the easiest way is to use local, like: `"local $/".

The idle pool size is capped at 8 idle coros (you can tweak this by changing $Coro::POOL_SIZE), but there can be as many non-idle coros as you need.

Worried about pooled coros ballooning because one "async_pool" ate a ton of stackspace? You can do something like "async_pool { terminate }" once a second or so to slowly replenish the pool. And on top of that, when the stacks used by a handler grow larger than 32kb (adjustable via $Coro::POOL_RSS) it'll get destroyed too. 💅

💖 STATIC METHODS 💖

These are actually functions that implicitly operate on the current coro. Sneaky. 🧚

  • schedule — calls the scheduler. The scheduler finds the next coro to run from the ready queue and switches to it. The next coro to run is simply the one with the highest priority that's been longest in its ready queue. If there's no coro ready, it calls the $Coro::idle hook instead.

Heads up bestie: the current coro does not get put into the ready queue, so calling this usually means you'll never be called again unless something else (like an event handler) calls "->ready" and wakes you back up. 😴

Which is exactly what makes "schedule" the generic method for blocking the current coro and waiting for events: first you remember the current coro in a variable, then you arrange for some callback of yours to call "->ready" on it once some event happens, and last you call "schedule" to put yourself to sleep. And note — a lot of things can wake your coro up, so you need to actually check whether the event happened, like by storing the status in a variable.

See HOW TO WAIT FOR A CALLBACK, below, for some ways to wait for callbacks. 👀

  • cede — "cede" to other coros. This puts the current coro into the ready queue and calls "schedule", which has the effect of giving up the current "timeslice" to other coros of the same or higher priority. Once your coro gets its turn again, it'll automatically be resumed.

Other languages call this "yield", iykyk.

  • Coro::cede_notself — works like cede, but it's not exported by default and it'll cede to any coro, regardless of priority. Sometimes useful to make sure progress actually happens.
  • terminate [arg...] — terminates the current coro with the given status values (see cancel). Heads up: the values will not be copied, they'll be referenced directly. 🙈
  • Coro::on_enter BLOCK, Coro::on_leave BLOCK — these install enter and leave winders in the current scope. The enter block gets executed when on_enter is called, and whenever the current coro is re-entered by the scheduler. The leave block gets executed whenever the current coro is blocked by the scheduler, and also when the containing scope is exited (by whatever means — exit, die, last, whatever).

Neither invoking the scheduler, nor exceptions, are allowed inside those BLOCKs. So: do not even think about calling "die" without an eval, and do not even think about entering the scheduler in any way. Seriously. 💀

Since both BLOCKs are tied to the current scope, they'll automatically be removed when the current scope exits. Clean.

These functions implement the same concept as "dynamic-wind" in scheme, and they're useful when you want to localise some resource to a specific coro.

They slow down thread switching considerably for coros that use them (about 40% for a BLOCK with a single assignment, so thread switching is still reasonably fast if the handlers themselves are fast).

These are best understood by example: the following function changes the current timezone to "Antarctica/South_Pole", which requires a call to "tzset", but by using "on_enter" and "on_leave" — which remember/change the current timezone and restore the previous value, respectively — the timezone only changes for the coro that installed those handlers. 🐧

use POSIX qw(tzset);

async {
   my $old_tz; # store outside TZ value here

   Coro::on_enter {
      $old_tz = $ENV{TZ}; # remember the old value

      $ENV{TZ} = "Antarctica/South_Pole";
      tzset; # enable new value
   };

   Coro::on_leave {
      $ENV{TZ} = $old_tz;
      tzset; # restore old value
   };

   # at this place, the timezone is Antarctica/South_Pole,
   # without disturbing the TZ of any other coro.
};

This can be used to localise pretty much any resource (locale, uid, current working directory etc.) to a block, even with other coros running around everywhere.

Another interesting example implements time-sliced multitasking using interval timers (this could obviously be optimised, but it does the job):

# "timeslice" the given block
sub timeslice(&) {
   use Time::HiRes ();

   Coro::on_enter {
      # on entering the thread, we set an VTALRM handler to cede
      $SIG{VTALRM} = sub { cede };
      # and then start the interval timer
      Time::HiRes::setitimer &Time::HiRes::ITIMER_VIRTUAL, 0.01, 0.01;
   };
   Coro::on_leave {
      # on leaving the thread, we stop the interval timer again
      Time::HiRes::setitimer &Time::HiRes::ITIMER_VIRTUAL, 0, 0;
   };

   &{+shift};
}

# use like this:
timeslice {
   # The following is an endless loop that would normally
   # monopolise the process. Since it runs in a timesliced
   # environment, it will regularly cede to other threads.
   while () { }
};
  • killall — kills/terminates/cancels all coros except the one currently running. 💅

Note that while this will try to free some of the main interpreter resources if the calling coro isn't the main coro, you can't free all of them — so if a coro that isn't the main coro calls this function, there will be some one-time resource leak. Just so you know. 🫶

✨ CORO OBJECT METHODS ✨

These are the methods you can call on coro objects (or to create them). Let's gooo. 🔥

  • new Coro \&sub [, @args...] — creates a new coro and returns it. When the sub returns, the coro automatically terminates as if "terminate" with the returned values were called. To actually make the coro run, you have to first put it into the ready queue by calling the ready method.

See "async" and "Coro::State::new" for extra info about the coro environment. 🌈

  • $success = $coro->ready — puts the given coro at the end of its ready queue (there's one queue for each priority) and returns true. If the coro is already in the ready queue, do nothing and return false. Simple as that. 💖

...and this is what makes sure the scheduler will come back around and resume this coro automatically, but only once every coro with a higher priority — and every coro at the same priority that was already sitting in the ready queue before it — has been resumed. Queue etiquette, bestie. No cutting. 💅

  • $coro->suspend — Puts the coro you point at to sleep-sleep. 😴 A suspended coro is still a perfectly normal coro in every way, the only difference is that the scheduler straight up will not pick a suspended coro to run.

So when is that useful? When you want to keep a coro around but stop it from running, without actually destroying it — or when you want to freeze a coro for a sec (debugging, looking at you 👀) and then bring it back later.

A real scenario for the "keep it alive" flavour: after a fork, suspend all the other coros and keep them alive, so their destructors never get called, while still being able to create new coros. Chef's kiss. 🎀

  • $coro->resume — If that coro was suspended, boom, it's resumed now. ⭐ But heads up: if the coro happened to be in the ready queue at the moment it got suspended, the scheduler might have un-readied it, which means an activation might have been lost. Rude, but that's the situation.

To dodge that whole mess, the move is to put a suspended coro into the ready queue unconditionally — because literally every synchronisation mechanism has to protect itself against spurious wakeups, and the ones in the Coro family absolutely do. ✨

  • $state->is_new — Returns true iff this Coro object is "new", meaning it has never been run yet. A new state is basically just the code reference to call plus the arguments — it uses up almost no other resources. 🦋 New states will automatically get assigned a perl interpreter when they are transferred to.
  • $state->is_zombie — Returns true iff the Coro object has been cancelled, i.e. its resources got freed because it was "cancel"'ed, "terminate"'d, "safe_cancel"'ed, or it just simply went out of scope.

The name "zombie" comes straight from UNIX culture, where a process that has exited and only stores an exit status and no other resources is called a "zombie". 💀 Heritage.

  • $is_ready = $coro->is_ready — Returns true iff the Coro object is in the ready queue. And unless the Coro object gets destroyed, the scheduler will eventually get to it. It's in the queue, it's getting its turn. periodt.
  • $is_running = $coro->is_running — Returns true iff the Coro object is currently running. Only one Coro object can ever be in the running state at a time (though currently it is possible to have multiple running Coro::States — yeah, we know 🙈).
  • $is_suspended = $coro->is_suspended — Returns true iff this Coro object has been suspended. Suspended Coros will not ever be scheduled. Ever. Not once.
  • $coro->cancel ($arg...) — Terminates the given Coro thread and makes it return the given arguments as its status (default: an empty list). Never returns if the Coro is the current Coro. 😭

This is a pretty brutal way to free a coro, and it comes with limitations — if the thread is inside a C callback that isn't expecting to be cancelled, bad things can happen; and if the cancelled thread insists on running complicated cleanup handlers that rely on its thread context, things will not work.

Any cleanup code that does run (from "guard" blocks, destructors, whatever) runs without a thread context, and is not allowed to switch to other threads. A super common mistake: calling ->cancel from a destructor that got triggered by die'ing inside the thread you're trying to cancel. Don't do that. 🙈

On the plus side: ->cancel will always clean up the thread, no matter what. But if your cleanup code is complex, or you want to avoid cancelling a C-thread that has no idea how to clean up after itself, it can be better to ->throw an exception, or use ->safe_cancel.

Also, the arguments to ->cancel are not copied — they're referenced directly. So if you pass $var and then change that variable after the call, you might change the return values that get passed to e.g. join. So don't do that. 🫶

The resources of the Coro are usually freed (or destructed) before this call returns, but that can be delayed for an indefinite amount of time, since in some cases the manager thread has to run first to actually destruct the Coro object.

  • $coro->safe_cancel ($arg...) — Works mostly like ->cancel, but it's inherently "safer" — and as a consequence, it can fail with an exception in cases where the thread isn't in a cancellable state. Essentially, ->safe_cancel is a ->cancel with extra checks before cancelling. 🔍

It works a bit like throwing an exception that cannot be caught — specifically, it cleans up the thread from within itself, so all cleanup handlers (e.g. "guard" blocks) run with full thread context and are allowed to block if they want. The downside: there's no guarantee the thread can be cancelled when you call this method, so it might fail. It's also considerably slower than cancel or terminate.

A thread is in a safe-cancellable state if it either has never been run yet, has already been canceled/terminated or otherwise destroyed, or has no C context attached and is inside an SLF function.

The first two states are trivial — a thread that hasn't started, or has already finished, is safe to cancel.

That last state basically means the thread isn't currently inside a perl callback called from some C function (usually via some XS modules), and isn't currently executing inside some C function itself (via Coro's XS API).

This call returns true when it could cancel the thread, or croaks with an error otherwise — so it either returns true, or it doesn't return at all.

Why the weird interface? Well, there are two common models for how and when to cancel things. In the first one, you just expect that your coro thread can be cancelled whenever you feel like cancelling it — so if the thread isn't cancellable, that's a bug somewhere, and ->safe_cancel croaks to let you know about the bug. 📣

In the second model, you sometimes just want to ask nicely whether you can cancel a thread, and if it's not a good time, well, then don't cancel. That one's pretty easy to do:

   if (! eval { $coro->safe_cancel }) {
      warn "unable to cancel thread: $@";
   }

However — what you should never do is try to cancel "safely" first and then, if that fails, cancel the "hard" way with ->cancel. That makes no sense at all: either you rely on being able to execute cleanup code in your thread context, or you don't. If you do, then ->safe_cancel is the only way. And if you don't, then ->cancel is always faster and more direct. ✨

  • $coro->schedule_to — Puts the current coro to sleep (like "Coro::schedule"), but instead of continuing with the next coro from the ready queue, it always switches to the given coro object — regardless of priority, regardless of anything. The readyness state of that coro isn't changed.

This is an advanced method for special cases — and honestly, the author would love to hear about any uses for this one. 👀

  • $coro->cede_to — Like schedule_to, but it also puts the current coro into the ready queue. The effect: you temporarily switch to the given coro, and then continue some time later. 🔁

Also an advanced method for special cases — and again, the author would love to hear about any uses for this one. 🫶

  • $coro->throw ([$scalar]) — If $throw is specified and defined, it gets thrown as an exception inside the coro at the next convenient point in time. Otherwise it clears the exception object.

Coro checks for the exception every time a schedule-like function returns — so after each "schedule", "cede", "Coro::Semaphore->down", "Coro::Handle->readable" and so on. Most of those functions (all the ones that are part of Coro itself) detect this case and return early when an exception is pending.

The exception object gets thrown "as is", with the specified scalar sitting in $@ — meaning if it's a string, no line number or newline gets appended (unlike with "die").

You can use this as a softer way than either "cancel" or "safe_cancel" to ask a coro to end itself — though there's no guarantee the exception will lead to termination, and if the exception isn't caught it might very well end the whole program. 😭

You might also think of "throw" as the moral equivalent of killing a coro with a signal (in this case, a scalar). 💅

  • $coro->join — Waits until the coro terminates, then returns any values that were given to the "terminate" or "cancel" functions. "join" can be called concurrently from multiple threads, and all of them will be resumed and given the status return once the $coro terminates. 👑
  • $coro->on_destroy (\&cb) — Registers a callback that gets called when this coro thread gets destroyed — that is, after its resources have been freed but before it is joined. The callback gets passed the terminate/cancel arguments, if there were any, and it must not die, under any circumstances. 🚫💀

There can be any number of "on_destroy" callbacks per coro, and there is currently no way to remove a callback once it's been added.

  • $oldprio = $coro->prio ($newprio) — Sets (or gets, if the argument is missing) the priority of the coro thread. Higher priority coros get run before lower priority coros. Priorities are small signed integers (currently -4 .. +3), and you can refer to them using the PRIO_xxx constants (use the import tag :prio to get them):
   PRIO_MAX > PRIO_HIGH > PRIO_NORMAL > PRIO_LOW > PRIO_IDLE > PRIO_MIN
       3    >     1     >      0      >    -1    >    -3     >    -4

   # set priority to HIGH
   current->prio (PRIO_HIGH);

The idle coro thread ($Coro::idle) always has a lower priority than any existing coro.

Changing the priority of the current coro takes effect immediately, but changing the priority of a coro that's in the ready queue (but not running) will only take effect after the next schedule (of that coro). That's a bug, and it will be fixed in some future version. 🐛

  • $newprio = $coro->nice ($change) — Similar to "prio", but it subtracts the given value from the priority — i.e. higher values mean lower priority, exactly like UNIX's nice command. 🎀
  • $olddesc = $coro->desc ($newdesc) — Sets (or gets, if the argument is missing) the description for this coro thread. It's just a free-form string you can associate with a coro. Whatever vibes you want.

This method simply sets the $coro->{desc} member to the given string. You can modify that member directly if you feel like it — and in fact, that's often preferred, so you can flag major processing states that then show up for example in a Coro::Debug session:

sub my_long_function {
   local $Coro::current->{desc} = "now in my_long_function";
   ...
   $Coro::current->{desc} = "my_long_function: phase 1";
   ...
   $Coro::current->{desc} = "my_long_function: phase 2";
   ...
}

✨ GLOBAL FUNCTIONS ✨

Coro::nready

Hands you back the number of coro that are currently in the ready state — like, the ones you could actually switch to by calling schedule directly or indirectly. 🫶

And a value of 0? That means the only runnable coro is the one running right now, so cede would do literally nothing, and schedule would deadlock unless there's an idle handler somewhere that wakes up some coro. bestie that's your warning sign. 💀

my $guard = Coro::guard { ... }

okay so this function still exists, but it's deprecated. 🙈 Please use the Guard::guard function instead, fr fr.

unblock_sub { ... }

This lil utility takes a BLOCK or a code reference and "unblocks" it, returning a shiny new coderef. ✨ And what does unblocking mean exactly? It means calling the new coderef will return immediately without blocking, returning nothing — while the original code ref gets called (with parameters) from within another coro. iconic.

The reason this function exists at all is that a lot of event libraries (such as the venerable Event module) are not thread-safe (a weaker form of reentrancy). Which means: you must not block inside event callbacks, or you might suffer from crashes or worse. 😭 The only event library currently known to be safe without unblock_sub is EV (but you might still run into deadlocks if all event loops are blocked).

Coro will try to catch you when you block in the event loop — it'll yell FATAL: $Coro::idle blocked itself — but that's just best effort and only works when you do not run your own event loop. 💅

So this function lets your callbacks block anyway, by executing them in another coro where blocking is actually safe. One example where blocking is handy: when you use the Coro::AIO functions to save results to disk, for example. 🍓

In short: simply use unblock_sub { ... } instead of sub { ... } when creating event callbacks that want to block.

If your handler does not plan to block (e.g. it simply sends a message to another coro, or puts some other coro into the ready queue), there is no reason to use unblock_sub. no cap.

Also note, bestie: you need unblock_sub for any other callbacks that are indirectly executed by any C-based event loop too. For example, when you use a module that uses AnyEvent (and you use Coro::AnyEvent) and it provides callbacks that are the result of some event callback, then you must not block either — or use unblock_sub. 👑

$cb = rouse_cb

Create and return a "rouse callback". That's a code reference that, when called, will remember a copy of its arguments and notify the owner coro of the callback. 🎀

Only the first invocation will store arguments and signal any waiter — further calls will effectively be ignored, but it is ok to try. 💜

Also see the next function. 👇

@args = rouse_wait [$cb]

Wait for the specified rouse callback to be invoked (or, if the argument is missing, use the most recently created callback in the current coro).

As soon as the callback is invoked (or when the callback was invoked before rouse_wait), it will return the arguments originally passed to the rouse callback. In scalar context, that means you get the last argument — just as if rouse_wait had a return ($a1, $a2, $a3...) statement at the end. 🌸

You are only allowed to wait once for a given rouse callback. one shot, no takebacks.

See the section HOW TO WAIT FOR A CALLBACK for an actual usage example. 🔥

As of Coro 6.57, you can reliably wait for a rouse callback in a different thread than from where it was created. slay.

🦋 HOW TO WAIT FOR A CALLBACK 🦋

It is very common for a coro to wait for some callback to be called. This happens naturally when you use coro in an otherwise event-based program, or when you use event-based libraries. 💫

These typically register a callback for some event, and call that callback when the event occurred. But in a coro, you typically just want to wait for the event — simplifying things. main character energy.

For example, AnyEvent->child registers a callback to be called when a specific child has exited:

my $child_watcher = AnyEvent->child (pid => $pid, cb => sub { ... });

But from within a coro, you often just want to write this:

my $status = wait_for_child $pid;

Coro offers two functions specifically designed to make this easy: rouse_cb and rouse_wait. 🫶

The first function, rouse_cb, generates and returns a callback that, when invoked, will save its arguments and notify the coro that created the callback.

The second function, rouse_wait, waits for the callback to be called (by calling schedule to go to sleep) and returns the arguments originally passed to the callback.

Using these functions, it becomes easy to write the wait_for_child function mentioned above:

sub wait_for_child($) {
   my ($pid) = @_;

   my $watcher = AnyEvent->child (pid => $pid, cb => rouse_cb);

   my ($rpid, $rstatus) = rouse_wait;
   $rstatus
}

In the case where rouse_cb and rouse_wait are not flexible enough, you can roll your own, using schedule and ready:

sub wait_for_child($) {
   my ($pid) = @_;

   # store the current coro in $current,
   # and provide result variables for the closure passed to ->child
   my $current = $Coro::current;
   my ($done, $rstatus);

   # pass a closure to ->child
   my $watcher = AnyEvent->child (pid => $pid, cb => sub {
      $rstatus = $_[1]; # remember rstatus
      $done = 1;        # mark $rstatus as valid
      $current->ready;  # wake up the waiting thread
   });

   # wait until the closure has been called
   schedule while !$done;

   $rstatus
}

💀 BUGS/LIMITATIONS 💀

fork with pthread backend

When Coro is compiled using the pthread backend (which isn't recommended but required on many BSDs, as their libcs are completely broken 😭), then coro will not survive a fork. There is no known workaround except to fix your libc and use a saner backend. periodt.

perl process emulation ("threads")

This module is not perl-pseudo-thread-safe. You should only ever use this module from the first thread (this requirement might be removed in the future to allow per-thread schedulers, but Coro::State does not yet allow this). I recommend disabling thread support and using processes, as having the windows process emulation enabled under unix roughly halves perl performance, even when not used. 💀

Attempts to use threads created in another emulated process will crash ("cleanly", with a null pointer exception). 🙈

coro switching is not signal safe

You must not switch to another coro from within a signal handler (only relevant with %SIG — most event libraries provide safe signals), unless you are sure you are not interrupting a Coro function.

That means you MUST NOT call any function that might "block" the current coro — cede, schedule, Coro::Semaphore->down, or anything that calls those. Everything else, including calling ready, works. ✨

🌈 WINDOWS PROCESS EMULATION 🌈

A great many people seem to be confused about ithreads (for example, Chip Salzenberg called me unintelligent, incapable, stupid and gullible, while in the same mail making rather confused statements about perl ithreads — for example, that memory or files would be shared — showing his lack of understanding of this area; if it is hard to understand for Chip, it is probably not obvious to everybody). 😭

What follows is an ultra-condensed version of my talk about threads in scripting languages, given on the perl workshop 2009:

The so-called "ithreads" were originally implemented for two reasons: first, to (badly) emulate unix processes on native win32 perls, and secondly, to replace the older, real thread model (5.005-threads).

It does that by using threads instead of OS processes. The difference between processes and threads is that threads share memory (and other state, such as files) between threads within a single process, while processes do not share anything (at least not semantically). That means that modifications done by one thread are seen by others, while modifications by one process are not seen by other processes.

The "ithreads" work exactly like that: when creating a new ithreads process, all state is copied (memory is copied physically, files and code is copied logically). Afterwards, it isolates all modifications. On UNIX, the same behaviour can be achieved by using operating system processes — except that UNIX typically uses hardware built into the system to do this efficiently, while the windows process emulation emulates this hardware in software (rather efficiently, but of course it is still much slower than dedicated hardware). 🐢

As mentioned before, loading code, modifying code, modifying data structures and so on is only visible in the ithreads process doing the modification, not in other ithread processes within the same OS process.

This is why "ithreads" do not implement threads for perl at all — only processes. What makes it so bad is that on non-windows platforms, you can actually take advantage of custom hardware for this purpose (as evidenced by the forks module, which gives you the (i-) threads API, just much faster). 🔥

Sharing data in the i-threads model is done by transferring data structures between threads using copying semantics, which is very slow — shared data simply does not exist. Benchmarks using i-threads which are communication-intensive show extremely bad behaviour with i-threads (in fact, so bad that Coro, which cannot take direct advantage of multiple CPUs, is often orders of magnitude faster because it shares data using real threads — refer to my talk for details). 👑

As summary, i-threads use threads to implement processes, while the compatible forks module uses processes to emulate, uhm, processes. I-threads slow down every perl program when enabled, and outside of windows, serve no (or little) practical purpose, but disadvantages every single-threaded Perl program. 💅

This is the reason that I try to avoid the name "ithreads", as it is misleading — it implies that it implements some kind of thread model for perl — and prefer the name windows process emulation, which describes the actual use and behaviour of it much better.

💖 SEE ALSO 💖

okay bestie, this is the friend group. every single one of these is a real module, and they all hang out with Coro:

  • Event-Loop integration: Coro::AnyEvent, Coro::EV, Coro::Event. ✨
  • Debugging: Coro::Debug. 🔍
  • Support/Utility: Coro::Specific, Coro::Util. 🫶
  • Locking and IPC: Coro::Signal, Coro::Channel, Coro::Semaphore, Coro::SemaphoreSet, Coro::RWLock. 🔒
  • I/O and Timers: Coro::Timer, Coro::Handle, Coro::Socket, Coro::AIO. ⏰
  • Compatibility with other modules: Coro::LWP — but lowkey also check out AnyEvent::HTTP for a better-working alternative 💅 — plus Coro::BDB, Coro::Storable, Coro::Select.
  • XS API: Coro::MakeMaker. 🔧
  • Low level Configuration, Thread Environment, Continuations: Coro::State. 🧠

👑 AUTHOR/SUPPORT/CONTACT 👑

Marc A. Lehmann — <schmorp@schmorp.de> http://software.schmorp.de/pkg/Coro.html

okay that's literally the whole page, you made it to the end, main character behavior fr fr 🎀 the source is perl v5.40.1 and this version is dated 2025-03-21 (the page itself is Coro(3pm)), so if you're reading a copy from somewhere else, that's the tea 🦋 go write some coroutines bestie, ily, slay 💖✨

📦 where this page comes from