1 |
root |
1.1 |
=head1 NAME |
2 |
|
|
|
3 |
|
|
AnyEvent::Impl::POE - AnyEvent adaptor for POE |
4 |
|
|
|
5 |
root |
1.13 |
=encoding utf-8 |
6 |
|
|
|
7 |
root |
1.1 |
=head1 SYNOPSIS |
8 |
|
|
|
9 |
root |
1.18 |
use AnyEvent; |
10 |
|
|
use POE; |
11 |
|
|
|
12 |
|
|
# this module gets loaded automatically as required |
13 |
root |
1.1 |
|
14 |
|
|
=head1 DESCRIPTION |
15 |
|
|
|
16 |
|
|
This module provides transparent support for AnyEvent. You don't have to |
17 |
|
|
do anything to make POE work with AnyEvent except by loading POE before |
18 |
root |
1.22 |
creating the first AnyEvent watcher. There are some cases where POE will |
19 |
|
|
issue spurious (and non-suppressable) warnings. These can be avoided by |
20 |
|
|
loading AnyEvent::Impl::POE before loading any other modules using POE and |
21 |
|
|
AnyEvent, i.e. in your main program. |
22 |
|
|
|
23 |
|
|
AnyEvent::Impl::POE will output some spurious message how to work around |
24 |
|
|
POE's spurious messages when it detects these cases. |
25 |
root |
1.1 |
|
26 |
root |
1.3 |
Unfortunately, POE isn't generic enough to implement a fully working |
27 |
|
|
AnyEvent backend: POE is too badly designed, too badly documented and too |
28 |
|
|
badly implemented. |
29 |
root |
1.2 |
|
30 |
root |
1.3 |
Here are the details, and what it means to you if you want to be |
31 |
|
|
interoperable with POE: |
32 |
root |
1.2 |
|
33 |
|
|
=over 4 |
34 |
|
|
|
35 |
|
|
=item Weird messages |
36 |
|
|
|
37 |
root |
1.4 |
If you only use C<run_one_timeslice> (as AnyEvent has to for it's |
38 |
|
|
condition variables), POE will print an ugly, unsupressable, message at |
39 |
|
|
program exit: |
40 |
root |
1.2 |
|
41 |
|
|
Sessions were started, but POE::Kernel's run() method was never... |
42 |
|
|
|
43 |
|
|
The message is correct, the question is why POE prints it in the first |
44 |
|
|
place in a correct program (this is not a singular case though). |
45 |
|
|
|
46 |
|
|
The only way I found to work around this bug was to call C<< |
47 |
root |
1.5 |
->run >> at AnyEvent loading time and stop the kernel immediately |
48 |
root |
1.2 |
again. Unfortunately, due to another design bug in POE, this cannot be |
49 |
|
|
done (by documented means at least) without throwing away events in the |
50 |
|
|
event queue. |
51 |
|
|
|
52 |
root |
1.16 |
The author of POE verified that this is indeed true, and has no plans to |
53 |
|
|
change this. |
54 |
|
|
|
55 |
root |
1.2 |
This means that you will either have to live with lost events or you have |
56 |
|
|
to make sure to load AnyEvent early enough (this is usually not that |
57 |
|
|
difficult in a main program, but hard in a module). |
58 |
|
|
|
59 |
root |
1.19 |
POE has other weird messages, and sometimes weird behaviour, for example, |
60 |
|
|
it doesn't support overloaded code references as callbacks for no apparent |
61 |
|
|
reason. |
62 |
|
|
|
63 |
root |
1.7 |
=item One POE session per Event |
64 |
root |
1.2 |
|
65 |
|
|
AnyEvent has to create one POE::Session per event watcher, which is |
66 |
root |
1.3 |
immensely slow and makes watchers very large. The reason for this is |
67 |
|
|
lacking lifetime management (mostly undocumented, too). Without one |
68 |
|
|
session/watcher it is not possible to easily keep the kernel from running |
69 |
root |
1.16 |
endlessly. |
70 |
|
|
|
71 |
|
|
This is not just a problem with the way AnyEvent has to interact with |
72 |
|
|
POE, but is a principal issue with POEs lifetime management (namely |
73 |
|
|
that stopping the kernel stops sessions, but AnyEvent has no control |
74 |
|
|
over who and when the kernel starts or stops w.r.t. AnyEvent watcher |
75 |
|
|
creation/destruction). |
76 |
|
|
|
77 |
|
|
From benchmark data it is not clear that session creation is that costly, |
78 |
|
|
though - the real inefficiencies with POE seem to come from other sources, |
79 |
|
|
such as event handling. |
80 |
root |
1.2 |
|
81 |
|
|
=item One watcher per fd/event combo |
82 |
|
|
|
83 |
|
|
POE, of course, suffers from the same bug as Tk and some other badly |
84 |
|
|
designed event models in that it doesn't support multiple watchers per |
85 |
root |
1.3 |
fd/poll combo. The workaround is the same as with Tk: AnyEvent::Impl::POE |
86 |
|
|
creates a separate file descriptor to hand to POE, which isn't fast and |
87 |
|
|
certainly not nice to your resources. |
88 |
root |
1.2 |
|
89 |
root |
1.7 |
Of course, without the workaround, POE also prints ugly messages again |
90 |
|
|
that say the program *might* be buggy. |
91 |
|
|
|
92 |
root |
1.16 |
While this is not good to performance, at least regarding speed, with a |
93 |
|
|
modern Linux kernel, the overhead is actually quite small. |
94 |
|
|
|
95 |
root |
1.3 |
=item Timing Deficiencies |
96 |
|
|
|
97 |
|
|
POE manages to not have a function that returns the current time. This is |
98 |
|
|
extremely problematic, as POE can use different time functions, which can |
99 |
root |
1.11 |
differ by more than a second - and user code is left guessing which one is |
100 |
|
|
used. |
101 |
|
|
|
102 |
|
|
In addition, most timer functions in POE want an absoltue timestamp, which |
103 |
|
|
is hard to create if all you have is a relative time and no function to |
104 |
root |
1.16 |
return the "current time". |
105 |
|
|
|
106 |
|
|
And of course POE doesn't handle time jumps at all (not even when using |
107 |
|
|
an event loop that happens to do that, such as L<EV>, as it does its own |
108 |
|
|
unoptimised timer management). |
109 |
root |
1.12 |
|
110 |
|
|
AnyEvent works around the unavailability of the current time using |
111 |
|
|
relative timers exclusively, in the hope that POE gets it right at least |
112 |
|
|
internally. |
113 |
root |
1.3 |
|
114 |
|
|
=item Event Non-Ordering |
115 |
|
|
|
116 |
|
|
POE cannot guarentee the order of callback invocation for timers, and |
117 |
|
|
usually gets it wrong. That is, if you have two timers, one timing out |
118 |
root |
1.25 |
after another (all else being equal), the callbacks might be called in |
119 |
root |
1.16 |
reverse order. |
120 |
root |
1.3 |
|
121 |
|
|
How one manages to even implement stuff that way escapes me. |
122 |
|
|
|
123 |
|
|
=item Child Watchers |
124 |
|
|
|
125 |
root |
1.7 |
POE offers child watchers - which is a laudable thing, as few event loops |
126 |
root |
1.3 |
do. Unfortunately, they cannot even implement AnyEvent's simple child |
127 |
root |
1.25 |
watchers: they are not generic enough (the POE implementation isn't even |
128 |
|
|
generic enough to let properly designed back-end use their native child |
129 |
|
|
watcher instead - it insist on doing it itself the broken way). |
130 |
|
|
|
131 |
|
|
Unfortunately, POE's child handling is inherently racy: if the child |
132 |
|
|
exits before the handler is created (which is impossible to avoid in |
133 |
|
|
general, imagine the forked program to exit immediately because of a |
134 |
|
|
bug, or imagine the POE kernel being busy for a second), one has to |
135 |
|
|
wait for another event to occur, which can take an indefinite amount of |
136 |
|
|
time. Apparently POE implements a busy-waiting loop every second, but this |
137 |
|
|
is not guaranteed or documented, so in practise child status events can be |
138 |
|
|
delayed for up to a second "only". |
139 |
|
|
|
140 |
|
|
Of course, whenever POE reaps an unrelated child it will also output a |
141 |
|
|
message for it that you cannot suppress (which shouldn't be too surprising |
142 |
|
|
at this point). Very professional. |
143 |
root |
1.6 |
|
144 |
root |
1.7 |
As a workaround, AnyEvent::Impl::POE will take advantage of undocumented |
145 |
root |
1.25 |
behaviour in POE::Kernel to catch the status of all child processes, but |
146 |
|
|
it cannot guarantee delivery. |
147 |
root |
1.7 |
|
148 |
|
|
How one manages to have such a glaring bug in an event loop after ten |
149 |
|
|
years of development escapes me. |
150 |
root |
1.3 |
|
151 |
root |
1.29 |
(There are more annoying bugs, for example, POE runs C<waitpid> |
152 |
|
|
unconditionally on finalizing, so your program will hang until all child |
153 |
|
|
processes have exited.) |
154 |
|
|
|
155 |
root |
1.3 |
=item Documentation Quality |
156 |
|
|
|
157 |
|
|
At the time of this writing, POE was in its tenth year. Still, its |
158 |
|
|
documentation is extremely lacking, making it impossible to implement |
159 |
root |
1.10 |
stuff as trivial as AnyEvent watchers without having to resort to |
160 |
root |
1.3 |
undocumented behaviour or features. |
161 |
|
|
|
162 |
root |
1.16 |
For example, the POE::Kernel manpage has nine occurances of the word TODO |
163 |
root |
1.10 |
with an explanation of whats missing. In general, the POE manpages are |
164 |
|
|
littered with comments like "section not yet written". |
165 |
|
|
|
166 |
|
|
Some other gems: |
167 |
root |
1.3 |
|
168 |
|
|
This allows many object methods to also be package methods. |
169 |
|
|
|
170 |
|
|
This is nice, but since it doesn't document I<which> methods these are, |
171 |
|
|
this is utterly useless information. |
172 |
|
|
|
173 |
|
|
Terminal signals will kill sessions if they are not handled by a |
174 |
root |
1.10 |
"sig_handled"() call. The OS signals that usually kill or dump a |
175 |
|
|
process are considered terminal in POE, but they never trigger a |
176 |
root |
1.3 |
coredump. These are: HUP, INT, QUIT and TERM. |
177 |
|
|
|
178 |
root |
1.10 |
Although AnyEvent calls C<sig_handled>, removing it has no apparent |
179 |
|
|
effects on POE handling SIGINT. |
180 |
root |
1.3 |
|
181 |
root |
1.11 |
refcount_increment SESSION_ID, COUNTER_NAME |
182 |
|
|
|
183 |
|
|
Nowhere is explained which COUNTER_NAMEs are valid and which aren't - not |
184 |
|
|
all scalars (or even strings) are valid counter names. Take your guess, |
185 |
root |
1.16 |
failure is of course completely silent. I found this out the hard way, as |
186 |
|
|
the first name I came up with was silently ignored. |
187 |
root |
1.11 |
|
188 |
|
|
get_next_event_time() returns the time the next event is due, in a form |
189 |
|
|
compatible with the UNIX time() function. |
190 |
|
|
|
191 |
|
|
And surely, one would hope that POE supports subsecond accuracy as |
192 |
|
|
documented elsewhere, unlike the explanation above implies. Yet: |
193 |
|
|
|
194 |
|
|
POE::Kernel timers support subsecond accuracy, but don’t expect too |
195 |
|
|
much here. Perl is not the right language for realtime programming. |
196 |
|
|
|
197 |
|
|
... of course, Perl is not the right language to expect subsecond accuray |
198 |
|
|
- the manpage author must hate Perl to spread so much FUD in so little |
199 |
|
|
space. The Deliantra game server logs with 100µs-accuracy because Perl is |
200 |
|
|
fast enough to require this, and is still able to deliver map updates with |
201 |
|
|
little jitter at exactly the right time. It does not, however, use POE. |
202 |
|
|
|
203 |
root |
1.3 |
Furthermore, since the Kernel keeps track of everything sessions do, it |
204 |
|
|
knows when a session has run out of tasks to perform. |
205 |
|
|
|
206 |
root |
1.11 |
This is impossible - how does the kernel know that a session is no longer |
207 |
root |
1.3 |
watching for some (external) event (e.g. by some other session)? It |
208 |
root |
1.10 |
cannot, and therefore this is wrong - but you would be hard pressed to |
209 |
|
|
find out how to work around this and tell the kernel manually about such |
210 |
|
|
events. |
211 |
root |
1.3 |
|
212 |
|
|
It gets worse, though - the notion of "task" or "resource", although used |
213 |
|
|
throughout the documentation, is not defined in a usable way. For example, |
214 |
|
|
waiting for a timeout is considered to be a task, waiting for a signal is |
215 |
root |
1.10 |
not (a session that only waits for a signal is considered finished and |
216 |
|
|
gets removed). The user is left guessing when waiting for an event counts |
217 |
root |
1.15 |
as task and when not (in fact, the issue with signals is mentioned in |
218 |
|
|
passing in a section about child watchers and directly contradicts earlier |
219 |
|
|
parts in that document). |
220 |
root |
1.3 |
|
221 |
root |
1.10 |
One could go on endlessly - ten years, no usable documentation. |
222 |
root |
1.3 |
|
223 |
root |
1.8 |
It is likely that differences between documentation, or the one or two |
224 |
root |
1.10 |
things I had to guess, cause unanticipated problems with this adaptor. |
225 |
root |
1.3 |
|
226 |
root |
1.16 |
=item Fragile and inconsistent API |
227 |
root |
1.3 |
|
228 |
|
|
The POE API is extremely inconsistent - sometimes you have to pass a |
229 |
|
|
session argument, sometimes it gets ignored, sometimes a session-specific |
230 |
|
|
method must not use a session argument. |
231 |
root |
1.2 |
|
232 |
root |
1.10 |
Error handling is sub-standard as well: even for programming mistakes, |
233 |
|
|
POE does not C<croak> but, in most cases, just sets C<$!> or simply does |
234 |
|
|
nothing at all, leading to fragile programs. |
235 |
|
|
|
236 |
|
|
Sometimes registering a handler uses the "eventname, parameter" ordering |
237 |
|
|
(timeouts), sometimes it is "parameter, eventname" (signals). There is |
238 |
|
|
little consistency overall. |
239 |
root |
1.1 |
|
240 |
root |
1.17 |
=item Lack of knowledge |
241 |
|
|
|
242 |
root |
1.18 |
The IO::Poll event loop provides an alternative that theoretically |
243 |
|
|
scales better than select(). |
244 |
root |
1.17 |
|
245 |
|
|
The IO::Poll "event loop" (who in his right mind would call that an event |
246 |
|
|
loop) of course scales about identically (sometimes it is a bit faster, |
247 |
|
|
sometimes a bit slower) to select in theory, and also in practise, of |
248 |
|
|
course, as both are O(n) in the number of file descriptors, which is |
249 |
|
|
rather bad. |
250 |
|
|
|
251 |
|
|
This is just one place where it gets obvious how little the author of the |
252 |
|
|
POE manpage understands. |
253 |
|
|
|
254 |
root |
1.23 |
=item No idle events |
255 |
|
|
|
256 |
|
|
The POE-recommended workaround to this is apparently to use |
257 |
|
|
C<fork>. Consequently, idle watchera will have to be emulated by AnyEvent. |
258 |
|
|
|
259 |
root |
1.3 |
=back |
260 |
root |
1.1 |
|
261 |
root |
1.9 |
On the good side, AnyEvent allows you to write your modules in a 100% |
262 |
|
|
POE-compatible way (bug-for-bug compatible even), without forcing your |
263 |
root |
1.3 |
module to use POE - it is still open to better event models, of which |
264 |
|
|
there are plenty. |
265 |
root |
1.1 |
|
266 |
root |
1.28 |
Oh, and one other positive thing: |
267 |
|
|
|
268 |
|
|
RUNNING_IN_HELL |
269 |
|
|
|
270 |
|
|
POE knows about the nature of the beast! |
271 |
|
|
|
272 |
root |
1.1 |
=cut |
273 |
|
|
|
274 |
|
|
package AnyEvent::Impl::POE; |
275 |
|
|
|
276 |
root |
1.30 |
use AnyEvent (); BEGIN { AnyEvent::common_sense } |
277 |
root |
1.1 |
use POE; |
278 |
|
|
|
279 |
root |
1.22 |
# if POE is already running |
280 |
|
|
if (${ $poe_kernel->[POE::Kernel::KR_RUN] } && POE::Kernel::KR_RUN_CALLED) { |
281 |
|
|
print STDERR <<EOF; |
282 |
root |
1.28 |
|
283 |
|
|
*** |
284 |
|
|
*** POE is going to complain about: |
285 |
|
|
*** |
286 |
|
|
*** Sessions were started, but POE::Kernel's run() method was never... |
287 |
|
|
*** |
288 |
|
|
*** Try putting: |
289 |
|
|
*** |
290 |
|
|
*** use AnyEvent::Impl::POE; |
291 |
|
|
*** |
292 |
|
|
*** at the very top of your main program to suppress these spurious warnings. |
293 |
|
|
*** |
294 |
|
|
|
295 |
root |
1.22 |
EOF |
296 |
|
|
} else { |
297 |
|
|
# workaround to suppress noise |
298 |
|
|
POE::Session->create (inline_states => { _start => sub { @_[KERNEL]->stop } }); |
299 |
|
|
POE::Kernel->run; |
300 |
|
|
} |
301 |
root |
1.1 |
|
302 |
|
|
sub io { |
303 |
|
|
my ($class, %arg) = @_; |
304 |
root |
1.2 |
|
305 |
root |
1.26 |
# POE itself might do the right thing, but some POE backends don't, |
306 |
|
|
# so do the safe thing, it's not as if this will slow us down |
307 |
|
|
# any further *g* |
308 |
root |
1.21 |
my ($fh, $pee) = AnyEvent::_dupfh $arg{poll}, $arg{fh}, "select_read", "select_write"; |
309 |
root |
1.2 |
|
310 |
root |
1.27 |
my $cb = delete $arg{cb}; my $cb = sub { &$cb }; # POE doesn't like callable objects |
311 |
root |
1.19 |
|
312 |
root |
1.1 |
my $session = POE::Session->create ( |
313 |
|
|
inline_states => { |
314 |
root |
1.26 |
_start => sub { $_[KERNEL]->$pee ($fh => "ready") }, |
315 |
|
|
ready => sub { $cb->() }, |
316 |
|
|
stop => sub { $_[KERNEL]->$pee ($fh) }, |
317 |
root |
1.1 |
}, |
318 |
|
|
); |
319 |
root |
1.23 |
bless \\$session, "AnyEvent::Impl::POE" |
320 |
root |
1.1 |
} |
321 |
|
|
|
322 |
|
|
sub timer { |
323 |
|
|
my ($class, %arg) = @_; |
324 |
root |
1.19 |
|
325 |
root |
1.27 |
my $after = delete $arg{after}; |
326 |
|
|
my $ival = delete $arg{interval}; |
327 |
|
|
my $cb = delete $arg{cb}; my $cb = sub { &$cb }; # POE doesn't like callable objects |
328 |
root |
1.19 |
|
329 |
root |
1.1 |
my $session = POE::Session->create ( |
330 |
|
|
inline_states => { |
331 |
|
|
_start => sub { |
332 |
|
|
$_[KERNEL]->delay_set (timeout => $after); |
333 |
|
|
}, |
334 |
root |
1.20 |
timeout => $ival ? sub { $_[KERNEL]->delay_set (timeout => $ival); $cb->() } : $cb, |
335 |
root |
1.1 |
stop => sub { |
336 |
|
|
$_[KERNEL]->alarm_remove_all; |
337 |
|
|
}, |
338 |
|
|
}, |
339 |
|
|
); |
340 |
root |
1.23 |
bless \\$session, "AnyEvent::Impl::POE" |
341 |
root |
1.1 |
} |
342 |
|
|
|
343 |
|
|
sub signal { |
344 |
|
|
my ($class, %arg) = @_; |
345 |
root |
1.31 |
my $signal = AnyEvent::Base::sig2name delete $arg{signal}; |
346 |
root |
1.27 |
my $cb = delete $arg{cb}; my $cb = sub { &$cb }; # POE doesn't like callable objects |
347 |
root |
1.1 |
my $session = POE::Session->create ( |
348 |
|
|
inline_states => { |
349 |
|
|
_start => sub { |
350 |
|
|
$_[KERNEL]->sig ($signal => "catch"); |
351 |
|
|
$_[KERNEL]->refcount_increment ($_[SESSION]->ID => "poe"); |
352 |
|
|
}, |
353 |
|
|
catch => sub { |
354 |
|
|
$cb->(); |
355 |
|
|
$_[KERNEL]->sig_handled; |
356 |
|
|
}, |
357 |
|
|
stop => sub { |
358 |
|
|
$_[KERNEL]->refcount_decrement ($_[SESSION]->ID => "poe"); |
359 |
|
|
$_[KERNEL]->sig ($signal); |
360 |
|
|
}, |
361 |
|
|
}, |
362 |
|
|
); |
363 |
root |
1.23 |
bless \\$session, "AnyEvent::Impl::POE" |
364 |
root |
1.1 |
} |
365 |
|
|
|
366 |
root |
1.7 |
sub child { |
367 |
|
|
my ($class, %arg) = @_; |
368 |
|
|
my $pid = delete $arg{pid}; |
369 |
root |
1.27 |
my $cb = delete $arg{cb}; my $cb = sub { &$cb }; # POE doesn't like callable objects |
370 |
root |
1.7 |
my $session = POE::Session->create ( |
371 |
|
|
inline_states => { |
372 |
|
|
_start => sub { |
373 |
|
|
$_[KERNEL]->sig (CHLD => "child"); |
374 |
|
|
$_[KERNEL]->refcount_increment ($_[SESSION]->ID => "poe"); |
375 |
|
|
}, |
376 |
|
|
child => sub { |
377 |
|
|
my ($rpid, $status) = @_[ARG1, ARG2]; |
378 |
|
|
|
379 |
|
|
$cb->($rpid, $status) if $rpid == $pid || $pid == 0; |
380 |
|
|
}, |
381 |
|
|
stop => sub { |
382 |
|
|
$_[KERNEL]->refcount_decrement ($_[SESSION]->ID => "poe"); |
383 |
|
|
$_[KERNEL]->sig ("CHLD"); |
384 |
|
|
}, |
385 |
|
|
}, |
386 |
|
|
); |
387 |
root |
1.23 |
bless \\$session, "AnyEvent::Impl::POE" |
388 |
root |
1.7 |
} |
389 |
|
|
|
390 |
root |
1.1 |
sub DESTROY { |
391 |
|
|
POE::Kernel->post (${${$_[0]}}, "stop"); |
392 |
|
|
} |
393 |
|
|
|
394 |
|
|
sub one_event { |
395 |
|
|
POE::Kernel->loop_do_timeslice; |
396 |
|
|
} |
397 |
|
|
|
398 |
root |
1.24 |
sub loop { |
399 |
|
|
POE::Kernel->run; |
400 |
|
|
} |
401 |
|
|
|
402 |
root |
1.1 |
1; |
403 |
|
|
|
404 |
|
|
=head1 SEE ALSO |
405 |
|
|
|
406 |
|
|
L<AnyEvent>, L<POE>. |
407 |
|
|
|
408 |
|
|
=head1 AUTHOR |
409 |
|
|
|
410 |
|
|
Marc Lehmann <schmorp@schmorp.de> |
411 |
|
|
http://home.schmorp.de/ |
412 |
|
|
|
413 |
|
|
=cut |
414 |
|
|
|