1 # Permission is granted to copy, distribute and/or modify this document
2 # under the terms of the GNU Free Documentation License, Version 1.3
3 # or any later version published by the Free Software Foundation;
4 # with no Invariant Sections, no Front-Cover Texts, and no Back-Cover Texts.
5 # A copy of the license is included in the section entitled ``GNU
6 # Free Documentation License''.
8 # A copy of the license is also available from the Free Software
9 # Foundation Web site at http://www.gnu.org/licenses/fdl.html
11 # Altenately, this document is also available under the Lesser General
12 # Public License, version 3 or later, as published by the Free Software
15 # A copy of the license is also available from the Free Software
16 # Foundation Web site at http://www.gnu.org/licenses/lgpl.html
20 Welcome to 8sync's documentation!
21 8sync is an asynchronous programming environment for GNU Guile.
22 (Get it? 8sync? Async??? Quiet your groans, it's a great name!)
24 8sync has some nice properties:
26 - 8sync uses the actor model as its fundamental concurrency
27 synchronization mechanism.
28 Since the actor model is a "shared nothing" asynchronous
29 environment, you don't need to worry about deadlocks or other
30 tricky problems common to other asynchronous models.
31 Actors are modular units of code and state which communicate
32 by sending messages to each other.
33 - If you've done enough asynchronous programming, you're probably
34 familiar with the dreaded term "callback hell".
35 Getting around callback hell usually involves a tradeoff of other,
36 still rather difficult to wrap your brain around programming
38 8sync uses some clever tricks involving "delimited continuations"
39 under the hood to make the code you write look familiar and
41 When you need to send a request to another actor and get some
42 information back from it without blocking, there's no need
43 to write a separate procedure... 8sync's scheduler will suspend
44 your procedure and wake it back up when a response is ready.
45 - Even nonblocking I/O code is straightforward to write.
46 Thanks to the "suspendable ports" code introduced in Guile 2.2,
47 writing asynchronous, nonblocking networked code looks mostly
48 like writing the same synchronous code.
49 8sync's scheduler handles suspending and resuming networked
50 code that would otherwise block.
51 - 8sync aims to be "batteries included".
52 Useful subsystems for IRC bots, HTTP servers, and so on are
53 included out of the box.
54 - 8sync prioritizes live hacking.
55 If using an editor like Emacs with a nice mode like Geiser,
56 an 8sync-using developer can change and fine-tune the behavior
57 of code /while it runs/.
58 This makes both debugging and development much more natural,
59 allowing the right designs to evolve under your fingertips.
60 A productive hacker is a happy hacker, after all!
62 In the future, 8sync will also provide the ability to spawn and
63 communicate with actors on different threads, processes, and machines,
64 with most code running the same as if actors were running in the same
65 execution environment.
67 But as a caution, 8sync is still very young.
68 The API is stabilizing, but not yet stable, and it is not yet well
71 But, consider this as much an opportunity as a warning.
72 8sync is in a state where there is much room for feedback and
76 And now, into the wild, beautiful frontier.
81 ** A silly little IRC bot
83 IRC! Internet Relay Chat!
84 The classic chat protocol of the Internet.
85 And it turns out, one of the best places to learn about networked
86 programming.[fn:irc-hacking]
87 We ourselves are going to explore chat bots as a basis for getting our
90 First of all, we're going to need to import some modules. Put this at
94 (use-modules (8sync) ; 8sync's agenda and actors
95 (8sync systems irc) ; the irc bot subsystem
96 (oop goops) ; 8sync's actors use GOOPS
97 (ice-9 format) ; basic string formatting
98 (ice-9 match)) ; pattern matching
101 Now we need to add our bot. Initially, it won't do much.
104 (define-class <my-irc-bot> (<irc-bot>))
106 (define-method (handle-line (irc-bot <my-irc-bot>) speaker channel
109 (format #t "~a emoted ~s in channel ~a\n"
110 speaker line channel)
111 (format #t "~a said ~s in channel ~a\n"
112 speaker line channel)))
115 We've just defined our own IRC bot!
116 This is an 8sync actor.
117 (8sync uses GOOPS to define actors.)
118 We extended the handle-line generic method, so this is the code that
119 will be called whenever the IRC bot "hears" anything.
120 For now the code is pretty basic: it just outputs whatever it "hears"
121 from a user in a channel to the current output port.
123 But it should help us make sure we have things working when we kick
126 Speaking of, even though we've defined our actor, it's not running
127 yet. Time to fix that!
130 (define* (run-bot #:key (username "examplebot")
131 (server "irc.freenode.net")
132 (channels '("##botchat")))
133 (define hive (make-hive))
135 (bootstrap-actor* hive <my-irc-bot> "irc-bot"
138 #:channels channels))
142 Actors are connected to something called a "hive", which is a
143 special kind of actor that runs and manages all the other actors.
144 Actors can spawn other actors, but before we start the hive we use
145 this special "bootstrap-actor*" method.
146 It takes the hive as its first argument, the actor class as the second
147 argument, a decorative "cookie" as the third argument (this is
148 optional, but it helps with debugging... you can skip it by setting it
149 to #f if you prefer), and the rest are initialization arguments to the
150 actor. bootstrap-actor* passes back not the actor itself (we don't
151 get access to that usually) but the *id* of the actor.
152 (More on this later.)
153 Finally we run the hive with run-hive and pass it a list of
154 "bootstrapped" messages.
155 Normally actors send messages to each other (and sometimes themselves),
156 but we need to send a message or messages to start things or else
157 nothing is going to happen.
162 (run-bot #:username "some-bot-username") ; be creative!
165 Assuming all the tubes on the internet are properly connected, you
166 should be able to join the "##botchat" channel on irc.freenode.net and
167 see your bot join as well.
168 Now, as you probably guessed, you can't really /do/ much yet.
169 If you talk to the bot, it'll send messages to the terminal informing
170 you as such, but it's hardly a chat bot if it's not chatting yet.
172 So let's do the most boring (and annoying) thing possible.
173 Let's get it to echo whatever we say back to us.
174 Change handle-line to this:
177 (define-method (handle-line (irc-bot <my-irc-bot>) speaker channel
179 (<- (actor-id irc-bot) 'send-line channel
180 (format #f "Bawwwwk! ~a says: ~a" speaker line)))
183 This will do exactly what it looks like: repeat back whatever anyone
184 says like an obnoxious parrot.
185 Give it a try, but don't keep it running for too long... this
186 bot is so annoying it's likely to get banned from whatever channel
189 This method handler does have the advantage of being simple though.
190 It introduces a new concept simply... sending a message!
191 Whenever you see "<-", you can think of that as saying "send this
193 The arguments to "<-" are as follows: the actor sending the message,
194 the id of the actor the message is being sent to, the "action" we
195 want to invoke (a symbol), and the rest are arguments to the
196 "action handler" which is in this case send-line (with itself takes
197 two arguments: the channel our bot should send a message to, and
198 the line we want it to spit out to the channel).
200 (Footnote: 8sync's name for sending a message, "<-", comes from older,
201 early lisp object oriented systems which were, as it turned out,
202 inspired by the actor model!
203 Eventually message passing was dropped in favor of something called
204 "generic functions" or "generic methods"
205 (you may observe we made use of such a thing in extending
207 Many lispers believe that there is no need for message passing
208 with generic methods and some advanced functional techniques,
209 but in a concurrent environment message passing becomes useful
210 again, especially when the communicating objects / actors are not
211 in the same address space.)
213 Normally in the actor model, we don't have direct references to
214 an actor, only an identifier.
215 This is for two reasons: to quasi-enforce the "shared nothing"
216 environment (actors absolutely control their own resources, and
217 "all you can do is send a message" to request that they modify
218 them) and because... well, you don't even know where that actor is!
219 Actors can be anything, and anywhere.
220 It's possible in 8sync to have an actor on a remote hive, which means
221 the actor could be on a remote process or even remote machine, and
222 in most cases message passing will look exactly the same.
223 (There are some exceptions; it's possible for two actors on the same
224 hive to "hand off" some special types of data that can't be serialized
225 across processes or the network, eg a socket or a closure, perhaps even
226 one with mutable state.
227 This must be done with care, and the actors should be careful both
228 to ensure that they are both local and that the actor handing things
229 off no longer accesses that value to preserve the actor model.
230 But this is an advanced topic, and we are getting ahead of ourselves.)
231 We have to supply the id of the receiving actor, and usually we'd have
233 But since in this case, since the actor we're sending this to is
234 ourselves, we have to pass in our identifier, since the Hive won't
235 deliver to anything other than an address.
237 Astute readers may observe, since this is a case where we are just
238 referencing our own object, couldn't we just call "sending a line"
239 as a method of our own object without all the message passing?
240 Indeed, we do have such a method, so we /could/ rewrite handle-line
244 (define-method (handle-line (irc-bot <my-irc-bot>) speaker channel
246 (irc-bot-send-line irc-bot channel
247 (format #f "Bawwwwk! ~a says: ~a" speaker line)))
250 ... but we want to get you comfortable and familiar with message
251 passing, and we'll be making use of this same message passing shortly
252 so that /other/ actors may participate in communicating with IRC
255 Anyway, our current message handler is simply too annoying.
256 What would be much more interesting is if we could recognize
257 when an actor could repeat messages /only/ when someone is speaking
259 Luckily this is an easy adjustment to make.
262 (define-method (handle-line (irc-bot <my-irc-bot>) speaker channel
264 (define my-name (irc-bot-username irc-bot))
265 (define (looks-like-me? str)
266 (or (equal? str my-name)
267 (equal? str (string-concatenate (list my-name ":")))))
268 (when (looks-like-me?)
269 (<- (actor-id irc-bot) 'send-line channel
270 (format #f "Bawwwwk! ~a says: ~a" speaker line))))
273 This is relatively straightforward, but it isn't very interesting.
274 What we would really like to do is have our bot respond to individual
275 "commands" like this:
278 <foo-user> examplebot: hi!
279 <examplebot> Oh hi foo-user!
280 <foo-user> examplebot: botsnack
281 <examplebot> Yippie! *does a dance!*
282 <foo-user> examplebot: echo I'm a very silly bot
283 <examplebot> I'm a very silly bot
286 Whee, that looks like fun!
287 To implement it, we're going to pull out Guile's pattern matcher.
290 (define-method (handle-line (irc-bot <my-irc-bot>) speaker channel
292 (define my-name (irc-bot-username irc-bot))
293 (define (looks-like-me? str)
294 (or (equal? str my-name)
295 (equal? str (string-concatenate (list my-name ":")))))
296 (match (string-split line #\space)
297 (((? looks-like-me? _) action action-args ...)
299 ;; The classic botsnack!
301 (<- (actor-id irc-bot) 'send-line channel
302 "Yippie! *does a dance!*"))
304 ((or "hello" "hello!" "hello." "greetings" "greetings." "greetings!"
305 "hei" "hei." "hei!" "hi" "hi!")
306 (<- (actor-id irc-bot) 'send-line channel
307 (format #f "Oh hi ~a!" speaker)))
309 (<- (actor-id irc-bot) 'send-line channel
310 (string-join action-args " ")))
312 ;; ---> Add yours here <---
316 (<- (actor-id irc-bot) 'send-line channel
317 "*stupid puppy look*"))))))
320 Parsing the pattern matcher syntax is left as an exercise for the
323 If you're getting the sense that we could make this a bit less wordy,
327 (define-method (handle-line (irc-bot <my-irc-bot>) speaker channel
329 (define my-name (irc-bot-username irc-bot))
330 (define (looks-like-me? str)
331 (or (equal? str my-name)
332 (equal? str (string-concatenate (list my-name ":")))))
333 (define (respond respond-line)
334 (<- (actor-id irc-bot) 'send-line channel
336 (match (string-split line #\space)
337 (((? looks-like-me? _) action action-args ...)
339 ;; The classic botsnack!
341 (respond "Yippie! *does a dance!*"))
343 ((or "hello" "hello!" "hello." "greetings" "greetings." "greetings!"
344 "hei" "hei." "hei!" "hi" "hi." "hi!")
345 (respond (format #f "Oh hi ~a!" speaker)))
347 (respond (string-join action-args " ")))
349 ;; ---> Add yours here <---
353 (respond "*stupid puppy look*"))))))
356 Okay, that looks pretty good!
357 Now we have enough information to build an IRC bot that can do a lot
359 Take some time to experiment with extending the bot a bit before
360 moving on to the next section!
361 What cool commands can you add?
364 In the 1990s I remember stumbling into some funky IRC chat rooms and
365 being astounded that people there had what they called "bots" hanging
367 From then until now, I've always enjoyed encountering bots whose range
368 of functionality has spanned from saying absurd things, to taking
369 messages when their "owners" were offline, to reporting the weather,
370 to logging meetings for participants.
371 And it turns out, IRC bots are a great way to cut your teeth on
372 networked programming; since IRC is a fairly simple line-delineated
373 protocol, it's a great way to learn to interact with sockets.
374 (My first IRC bot helped my team pick a place to go to lunch, previously
375 a source of significant dispute!)
376 At the time of writing, venture capital awash startups are trying to
377 turn chatbots into "big business"... a strange (and perhaps absurd)
378 thing given chat bots being a fairly mundane novelty amongst hackers
379 and teenagers everywhere in the 1990s.
381 ** An intermission: about live hacking
383 This section is optional, but highly recommended.
384 It requires that you're a user of GNU Emacs.
385 If you aren't, don't worry... you can forge ahead and come back in case
386 you ever do become an Emacs user.
387 (If you're more familiar with Vi/Vim style editing, I hear good things
390 So you may have noticed while updating the last section that the
391 start/stop cycle of hacking isn't really ideal.
392 You might either edit a file in your editor, then run it, or
393 type the whole program into the REPL, but then you'll have to spend
394 extra time copying it to a file.
395 Wouldn't it be nice if it were possible to both write code in a
396 file and try it as you go?
397 And wouldn't it be even better if you could live edit a program
400 Luckily, there's a great Emacs mode called Geiser which makes
401 editing and hacking and experimenting all happen in harmony.
402 And even better, 8sync is optimized for this experience.
403 8sync provides easy drop-in "cooperative REPL" support, and
404 most code can be simply redefined on the fly in 8sync through Geiser
405 and actors will immediately update their behavior, so you can test
406 and tweak things as you go.
408 Okay, enough talking. Let's add it!
409 Redefine run-bot like so:
412 (define* (run-bot #:key (username "examplebot")
413 (server "irc.freenode.net")
414 (channels '("##botchat"))
415 (repl-path "/tmp/8sync-repl"))
416 (define hive (make-hive))
418 (bootstrap-actor* hive <my-irc-bot> "irc-bot"
421 #:channels channels))
423 (bootstrap-actor* hive <repl-manager> "repl"
429 If we put a call to run-bot at the bottom of our file we can call it,
430 and the repl-manager will start something we can connect to automatically.
432 Now when we run this it'll start up a REPL with a unix domain socket at
434 We can connect to it in emacs like so:
436 : M-x geiser-connect-local <RET> guile <RET> /tmp/8sync-repl <RET>
438 Okay, so what does this get us?
439 Well, we can now live edit our program.
440 Let's change how our bot behaves a bit.
441 Let's change handle-line and tweak how the bot responds to a botsnack.
447 (respond "Yippie! *does a dance!*"))
451 (respond "Yippie! *catches botsnack in midair!*"))
454 Okay, now let's evaluate the change of the definition.
455 You can hit "C-M-x" anywhere in the definition to re-evaluate.
456 (You can also position your cursor at the end of the definition and press
457 "C-x C-e", but I've come to like "C-M-x" better because I can evaluate as soon
458 as I'm done writing.)
459 Now, on IRC, ask your bot for a botsnack.
460 The bot should give the new message... with no need to stop and start the
463 Let's fix a bug live.
464 Our current program works great if you talk to your bot in the same
465 IRC channel, but what if you try to talk to them over private message?
468 IRC> /query examplebot
469 <foo-user> examplebot: hi!
472 Hm, we aren't seeing any response on IRC!
473 Huh? What's going on?
474 It's time to do some debugging.
475 There are plenty of debugging tools in Guile, but sometimes the simplest
476 is the nicest, and the simplest debugging route around is good old
477 fashioned print debugging.
479 It turns out Guile has an under-advertised feature which makes print
480 debugging really easy called "pk", pronounced "peek".
481 What pk accepts a list of arguments, prints out the whole thing,
482 but returns the last argument.
483 This makes wrapping bits of our code pretty easy to see what's
485 So let's peek into our program with pk.
486 Edit the respond section to see what channel it's really sending
490 (define-method (handle-line (irc-bot <my-irc-bot>) speaker channel
493 (define (respond respond-line)
494 (<- (actor-id irc-bot) 'send-line (pk 'channel channel)
501 Now let's ping our bot in both the channel and over PM.
504 ;;; (channel "##botchat")
506 ;;; (channel "sinkbot")
509 Oh okay, this makes sense.
510 When we're talking in a normal multi-user channel, the channel we see
511 the message coming from is the same one we send to.
512 But over PM, the channel is a username, and in this case the username
513 we're sending our line of text to is ourselves.
514 That isn't what we want.
515 Let's edit our code so that if we see that the channel we're sending
516 to looks like our own username that we respond back to the sender.
517 (We can remove the pk now that we know what's going on.)
520 (define-method (handle-line (irc-bot <my-irc-bot>) speaker channel
523 (define (respond respond-line)
524 (<- (actor-id irc-bot) 'send-line
525 (if (looks-like-me? channel)
527 channel) ; normal IRC channel
533 Re-evaluate and test.
536 IRC> /query examplebot
537 <foo-user> examplebot: hi!
538 <examplebot> Oh hi foo-user!
543 ** Writing our own actors
545 Let's write the most basic, boring actor possible.
546 How about an actor that start sleeping, and keeps sleeping?
549 (use-modules (oop goops)
552 (define-class <sleeper> (<actor>)
553 (actions #:allocation #:each-subclass
554 #:init-value (build-actions
555 (*init* sleeper-loop))))
557 (define (sleeper-loop actor message)
558 (while (actor-alive? actor)
559 (display "Zzzzzzzz....\n")
560 ;; Sleep for one second
561 (8sleep (sleeper-sleep-secs actor))))
563 (let* ((hive (make-hive))
564 (sleeper (bootstrap-actor hive <sleeper>)))
568 We see some particular things in this example.
569 One thing is that our <sleeper> actor has an actions slot.
570 This is used to look up what the "action handler" for a message is.
571 We have to set the #:allocation to either #:each-subclass or #:class.
572 (#:class should be fine, except there is [[https://debbugs.gnu.org/cgi/bugreport.cgi?bug=25211][a bug in Guile]] which keeps
573 us from using it for now.)
575 The only action handler we've added is for =*init*=, which is called
576 implicitly when the actor first starts up.
577 (This will be true whether we bootstrap the actor before the hive
578 starts or create it during the hive's execution.)
580 In our sleeper-loop we also see a call to "8sleep".
581 "8sleep" is like Guile's "sleep" method, except it is non-blocking
582 and will always yield to the scheduler.
584 Our while loop also checks "actor-alive?" to see whether or not
585 it is still registered.
586 In general, if you keep a loop in your actor that regularly yields
587 to the scheduler, you should check this.
588 (An alternate way to handle it would be to not use a while loop at all
589 but simply send a message to ourselves with "<-" to call the
590 sleeper-loop handler again.
591 If the actor was dead, the message simply would not be delivered and
592 thus the loop would stop.)
594 It turns out we could have written the class for the actor much more
598 ;; You could do this instead of the define-class above.
599 (define-actor <sleeper> (<actor>)
600 ((*init* sleeper-loop)))
603 This is sugar, and expands into exactly the same thing as the
605 The third argument is an argument list, the same as what's passed
607 Everything after that is a slot.
608 So for example, if we had added an optional slot to specify
609 how many seconds to sleep, we could have done it like so:
612 (define-actor <sleeper> (<actor>)
613 ((*init* sleeper-loop))
614 (sleep-secs #:init-value 1
615 #:getter sleeper-sleep-secs))
618 This actor is pretty lazy though.
619 Time to get back to work!
620 Let's build a worker / manager type system.
626 (define-actor <manager> (<actor>)
627 ((assign-task manager-assign-task))
628 (direct-report #:init-keyword #:direct-report
629 #:getter manager-direct-report))
631 (define (manager-assign-task manager message difficulty)
632 "Delegate a task to our direct report"
633 (display "manager> Work on this task for me!\n")
634 (<- (manager-direct-report manager)
635 'work-on-this difficulty))
638 This manager keeps track of a direct report and tells them to start
639 working on a task... simple delegation.
640 Nothing here is really new, but note that our friend "<-" (which means
641 "send message") is back.
642 There's one difference this time... the first time we saw "<-" was in
643 the handle-line procedure of the irc-bot, and in that case we explicitly
644 pulled the actor-id after the actor we were sending the message to
645 (ourselves), which we aren't doing here.
646 But that was an unusual case, because the actor was ourself.
647 In this case, and in general, actors don't have direct references to
648 other actors; instead, all they have is access to identifiers which
649 reference other actors.
652 (define-actor <worker> (<actor>)
653 ((work-on-this worker-work-on-this))
654 (task-left #:init-keyword #:task-left
655 #:accessor worker-task-left))
657 (define (worker-work-on-this worker message difficulty)
658 "Work on one task until done."
659 (set! (worker-task-left worker) difficulty)
660 (display "worker> Whatever you say, boss!\n")
661 (while (and (actor-alive? worker)
662 (> (worker-task-left worker) 0))
663 (display "worker> *huff puff*\n")
664 (set! (worker-task-left worker)
665 (- (worker-task-left worker) 1))
669 The worker also contains familiar code, but we now see that we can
670 call 8sleep with non-integer real numbers.
672 Looks like there's nothing left to do but run it.
675 (let* ((hive (make-hive))
676 (worker (bootstrap-actor hive <worker>))
677 (manager (bootstrap-actor hive <manager>
678 #:direct-report worker)))
679 (run-hive hive (list (bootstrap-message hive manager 'assign-task 5))))
682 Unlike the =<sleeper>=, our =<manager>= doesn't have an implicit
683 =*init*= method, so we've bootstrapped the calling =assign-task= action.
686 manager> Work on this task for me!
687 worker> Whatever you say, boss!
695 "<-" pays no attention to what happens with the messages it has sent
697 This is useful in many cases... we can blast off many messages and
698 continue along without holding anything back.
700 But sometimes we want to make sure that something completes before
701 we do something else, or we want to send a message and get some sort
703 Luckily 8sync comes with an answer to that with "<-wait", which will
704 suspend the caller until the callee gives some sort of response, but
705 which does not block the rest of the program from running.
706 Let's try applying that to our own code by turning our manager
711 ;;; Update this method
712 (define (manager-assign-task manager message difficulty)
713 "Delegate a task to our direct report"
714 (display "manager> Work on this task for me!\n")
715 (<- (manager-direct-report manager)
716 'work-on-this difficulty)
718 ;; Wait a moment, then call the micromanagement loop
720 (manager-micromanage-loop manager))
722 ;;; And add the following
723 ;;; (... Note: do not model actual employee management off this)
724 (define (manager-micromanage-loop manager)
725 "Pester direct report until they're done with their task."
726 (display "manager> Are you done yet???\n")
727 (let ((worker-is-done
728 (mbody-val (<-wait (manager-direct-report manager)
731 (begin (display "manager> Oh! I guess you can go home then.\n")
732 (<- (manager-direct-report manager) 'go-home))
733 (begin (display "manager> Harumph!\n")
735 (when (actor-alive? manager)
736 (manager-micromanage-loop manager))))))
739 We've appended a micromanagement loop here... but what's going on?
740 "<-wait", as it sounds, waits for a reply, and returns a reply
742 In this case there's a value in the body of the message we want,
743 so we pull it out with mbody-val.
744 (It's possible for a remote actor to return multiple values, in which
745 case we'd want to use mbody-receive, but that's a bit more
748 Of course, we need to update our worker accordingly as well.
751 ;;; Update the worker to add the following new actions:
752 (define-actor <worker> (<actor>)
753 ((work-on-this worker-work-on-this)
755 (done-yet? worker-done-yet?)
756 (go-home worker-go-home))
757 (task-left #:init-keyword #:task-left
758 #:accessor worker-task-left))
761 (define (worker-done-yet? worker message)
762 "Reply with whether or not we're done yet."
763 (let ((am-i-done? (= (worker-task-left worker) 0)))
765 (display "worker> Yes, I finished up!\n")
766 (display "worker> No... I'm still working on it...\n"))
767 (<-reply message am-i-done?)))
769 (define (worker-go-home worker message)
770 "It's off of work for us!"
771 (display "worker> Whew! Free at last.\n")
772 (self-destruct worker))
775 (As you've probably guessed, you wouldn't normally call =display=
776 everywhere as we are in this program... that's just to make the
777 examples more illustrative.)
779 Running it is the same as before:
782 (let* ((hive (make-hive))
783 (worker (bootstrap-actor hive <worker>))
784 (manager (bootstrap-actor hive <manager>
785 #:direct-report worker)))
786 (run-hive hive (list (bootstrap-message hive manager 'assign-task 5))))
789 But the output is a bit different:
792 manager> Work on this task for me!
793 worker> Whatever you say, boss!
796 manager> Are you done yet???
797 worker> No... I'm still working on it...
800 manager> Are you done yet???
802 worker> No... I'm still working on it...
805 manager> Are you done yet???
806 worker> Yes, I finished up!
807 manager> Oh! I guess you can go home then.
808 worker> Whew! Free at last.
811 "<-reply" is what actually returns the information to the actor
812 waiting on the reply.
813 It takes as an argument the actor sending the message, the message
814 it is in reply to, and the rest of the arguments are the "body" of
816 (If an actor handles a message that is being "waited on" but does not
817 explicitly reply to it, an auto-reply with an empty body will be
818 triggered so that the waiting actor is not left waiting around.)
820 The last thing to note is the call to "self-destruct".
821 This does what you might expect: it removes the actor from the hive.
822 No new messages will be sent to it.
825 ** Writing our own network-enabled actor
827 So, you want to write a networked actor!
828 Well, luckily that's pretty easy, especially with all you know so far.
831 (use-modules (oop goops)
833 (ice-9 rdelim) ; line delineated i/o
834 (ice-9 match)) ; pattern matching
836 (define-actor <telcmd> (<actor>)
837 ((*init* telcmd-init)
838 (*cleanup* telcmd-cleanup)
839 (new-client telcmd-new-client))
840 (socket #:accessor telcmd-socket
844 Nothing surprising about the actor definition, though we do see that
845 it has a slot for a socket.
846 Unsurprisingly, that will be set up in the =*init*= handler.
849 (define (set-port-nonblocking! port)
850 (let ((flags (fcntl port F_GETFL)))
851 (fcntl port F_SETFL (logior O_NONBLOCK flags))))
853 (define (setup-socket)
856 (socket PF_INET SOCK_STREAM 0))
857 ;; reuse port even if busy
858 (setsockopt s SOL_SOCKET SO_REUSEADDR 1)
859 ;; connect to port 8889 on localhost
860 (bind s AF_INET INADDR_LOOPBACK 8889)
861 ;; make it nonblocking and start listening
862 (set-port-nonblocking! s)
866 (define (telcmd-init telcmd message)
867 (set! (telcmd-socket telcmd) (setup-socket))
868 (display "Connect like: telnet localhost 8889\n")
869 (while (actor-alive? telcmd)
870 (let ((client-connection (accept (telcmd-socket telcmd))))
871 (<- (actor-id telcmd) 'new-client client-connection))))
873 (define (telcmd-cleanup telcmd message)
874 (display "Closing socket!\n")
875 (when (telcmd-socket telcmd)
876 (close (telcmd-socket telcmd))))
879 That =setup-socket= code looks pretty hard to read!
880 But that's pretty standard code for setting up a socket.
881 One special thing is done though... the call to
882 =set-port-nonblocking!= sets flags on the socket port so that,
883 you guessed it, will be a nonblocking port.
885 This is put to immediate use in the telcmd-init method.
886 This code looks suspiciously like it /should/ block... after
887 all, it just keeps looping forever.
888 But since 8sync is using Guile's suspendable ports code feature,
889 so every time this loop hits the =accept= call, if that call
890 /would have/ blocked, instead this whole procedure suspends
891 to the scheduler... automatically!... allowing other code to run.
893 So, as soon as we do accept a connection, we send a message to
894 ourselves with the =new-client= action.
896 Aren't actors only supposed to handle one message at a time?
897 If the telcmd-init loop just keeps on looping and looping,
898 when will the =new-client= message ever be handled?
899 8sync actors only receive one message at a time, but by default if an
900 actor's message handler suspends to the agenda for some reason (such
901 as to send a message or on handling I/O), that actor may continue to
902 accept other messages, but always in the same thread.[fn:queued-handler]
904 We also see that we've established a =*cleanup*= handler.
905 This is run any time either the actor dies, either through self
906 destructing, because the hive completes its work, or because
907 a signal was sent to interrupt or terminate our program.
908 In our case, we politely close the socket when =<telcmd>= dies.
911 (define (telcmd-new-client telcmd message client-connection)
912 (define client (car client-connection))
913 (set-port-nonblocking! client)
915 (let ((line (read-line client)))
916 (cond ((eof-object? line)
919 (telcmd-handle-line telcmd client
920 (string-trim-right line #\return))
921 (when (actor-alive? telcmd)
924 (define (telcmd-handle-line telcmd client line)
925 (match (string-split line #\space)
926 (("") #f) ; ignore empty lines
929 (strftime "The time is: %c\n" (localtime (current-time)))
932 (format client "~a\n" (string-join rest " ")))
934 (_ (display "Sorry, I don't know that command.\n" client))))
937 Okay, we have a client, so we handle it!
938 And once again... we see this goes off on a loop of its own!
939 (Also once again, we have to do the =set-port-nonblocking!= song and
941 This loop also automatically suspends when it would otherwise block...
942 as long as read-line has information to process, it'll keep going, but
943 if it would have blocked waiting for input, then it would suspend the
946 The actual method called whenever we have a "line" of input is pretty
947 straightforward... in fact it looks an awful lot like the IRC bot
948 handle-line procedure we used earlier.
954 (let* ((hive (make-hive))
955 (telcmd (bootstrap-actor hive <telcmd>)))
959 Open up another terminal... you can connect via telnet:
962 $ telnet localhost 8889
964 Connected to localhost.
965 Escape character is '^]'.
967 The time is: Thu Jan 5 03:20:17 2017
971 Sorry, I don't know that command.
975 Type =Ctrl+] Ctrl+d= to exit telnet.
978 There's more that could be optimized, but we'll consider that to be
979 advanced topics of discussion.
981 So that's a pretty solid intro to how 8sync works!
982 Now that you've gone through this introduction, we hope you'll have fun
983 writing and hooking together your own actors.
984 Since actors are so modular, it's easy to have a program that has
985 multiple subystems working together.
986 You could build a worker queue system that displayed a web interface
987 and spat out notifications about when tasks finish to IRC, and making
988 all those actors talk to each other should be a piece of cake.
994 If there's a lot of data coming in and you don't want your I/O loop
995 to become too "greedy", take a look at =setvbuf=.
998 This is customizable: an actor can be set up to queue messages so
999 that absolutely no messages are handled until the actor completely
1000 finishes handling one message.
1001 Our loop couldn't look quite like this though!
1009 ** COMMENT Websockets
1012 ** Recommended .emacs additions
1014 In order for =mbody-receive= to indent properly, put this in your
1017 #+BEGIN_SRC emacs-lisp
1018 (put 'mbody-receive 'scheme-indent-function 2)
1023 One other major library for asynchronous communication in Guile-land
1024 is [[https://github.com/wingo/fibers/][Fibers]].
1025 There's a lot of overlap:
1027 - Both use Guile's suspendable-ports facility
1028 - Both communicate between asynchronous processes using message passing;
1029 you don't have to squint hard to see the relationship between Fibers'
1030 channels and 8sync's actor inboxes.
1032 However, there are clearly differences too.
1033 There's a one to one relationship between 8sync actors and an actor inbox,
1034 whereas each Fibers fiber may read from multiple channels, for example.
1036 Luckily, it turns out there's a clear relationship, based on real,
1038 8sync is based on the [[https://en.wikipedia.org/wiki/Actor_model][actor model]] whereas fibers follows
1039 [[http://usingcsp.com/][Communicating Sequential Processes (CSP)]], which is a form of
1040 [[https://en.wikipedia.org/wiki/Process_calculus][process calculi]].
1041 And it turns out, the
1042 [[https://en.wikipedia.org/wiki/Actor_model_and_process_calculi][relationship between the actor model and process calculi]] is well documented,
1043 and even more precisely, the
1044 [[https://en.wikipedia.org/wiki/Communicating_sequential_processes#Comparison_with_the_Actor_Model][relationship between CSP and the actor model]] is well understood too.
1046 So, 8sync and Fibers do take somewhat different approaches, but both
1047 have a solid theoretical backing... and their theories are well
1048 understood in terms of each other.
1049 Good news for theory nerds!
1051 (Since the actors and CSP are [[https://en.wikipedia.org/wiki/Dual_%28mathematics%29][dual]], maybe eventually 8sync will be
1052 implemented on top of Fibers... that remains to be seen!)