ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/Coro/Changes
(Generate patch)

Comparing Coro/Changes (file contents):
Revision 1.424 by root, Mon Nov 17 07:03:11 2008 UTC vs.
Revision 1.595 by root, Sat Oct 6 21:04:59 2012 UTC

1Revision history for Perl extension Coro. 1Revision history for Perl extension Coro.
2 2
3TODO: should explore PerlIO::coroaio (perl leaks like hell). 3TODO: should explore PerlIO::coroaio (perl leaks like hell).
4TODO: maybe implement a default message channel, very much like Erlang's 4TODO: channel->maxsize(newsize)?
5 actor model (which is cool in a lot of important aspects (failures!), 5TODO: http://www.microsoft.com/msj/archive/s2ce.aspx
6 but very lacking in others (higher level ipc)). 6TODO: __GCC_HAVE_DWARF2_CFI_ASM
7 7
8TODO: testcancel 8 - Coro::EV I/O watchers were not interruptible by exceptions
9TODO: aio 9 (Coro::State::throw) (testcase by sten).
10TODO: Coro::Semaphore 10 - use fd -1 in mmap.
11TODO: trow disturbs cleanup path inside pp_slf? 11 - cast I32 to int in error message printf.
12TODO: throw special-case in transfer? 12 - warn about broken so-called "hardened" kernels.
135.0 13
146.08 Fri Apr 13 12:05:47 CEST 2012
15 - be more aggressive about exiting like perl does - formerly,
16 exiting from the non-main thread would not execute END blocks.
17
186.07 Fri Nov 11 21:21:48 CET 2011
19 - work around a bug in PerlIO (setting $SIG{__WARN__} to a PVCV).
20 - update ecb.h.
21
226.06 Mon Aug 8 23:59:48 CEST 2011
23 - cygwin unfortunately patches the stack at runtime, so we use the pthreads
24 backend, which is an order of magnitude slower. unfortunately, cygwins
25 pthread implementation isn't very complete either, so allocate the stack
26 twice just to be sure.
27 (note: cygwin also enables mymalloc, which is NOT THREADSAFE ON WINDOWS,
28 in its ithreaded perl - best recompile cygwin and use the 'w'indows
29 backend for much better performance. also disable ithreads for
30 even better performance...).
31
326.05 Thu Aug 4 21:36:36 CEST 2011
33 - blush, condvar values would not be propagated from send to recv anymore
34 (reported by Chip Salzenberg).
35 - use exponential increase for the readline buffer length in
36 Coro::Handle. also reduce initial allocation to 1020 from 4096 bytes.
37
386.04 Wed Aug 3 17:15:45 CEST 2011
39 - use even more efficient and more compatible condvars for
40 compatibility to AnyEvent 6.x :)
41 - more inconsequential ecb.h updates.
42
436.03 Wed Aug 3 11:41:30 CEST 2011
44 - change how Coro patches AnyEvent condvars for compatibility to
45 AnyEvent 6.x.
46 - update ecb.h, to no longer include <pthread.h> in case WinNT.h
47 hasn't been included.
48
496.02 Wed Jul 13 04:35:19 CEST 2011
50 - "improve portability to Gentoo" - gentoo manages to put perl variables
51 in memory areas that are farther than 2gb apart, which the jit couldn't
52 handle and barfed. now it's just a bit slower on gentoo and similar
53 systems.
54
556.01 Sun Jul 3 12:31:14 CEST 2011
56 - workarounds are good, but the test for whether pthreads are used
57 was not good. thisone should be better.
58 - check differently whether gcc generates cfi instructions itself.
59
606.0 Wed Jun 29 19:43:35 CEST 2011
61 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be
62 destroyed and cleaned up automatically (e.g. async { schedule }).
63 - implement a JIT compiler for part of the thread switch code,
64 which gives a 50% speed improvement on threaded perls, and
65 about 4% on non-thraeded perls (so threaded perls now finally
66 reach about half the speed of non-threaded perls).
67 - slightly modernise Coro::Intro, add section about rouse functions.
68 - avoid DEFSV and ERRSV, giving another 10% improvement
69 in thread switching.
70 - Coro::State->is_destroyed is now called is_zombie.
71 - implement a Coro->safe_cancel method that might fail, but
72 cancels in a "safer" way if it succeeds.
73 - add preliminary support for DEBUGGING perls.
74 - get rid of two hash-accesses when initialising a new Coro - this
75 speeds up coro creation by almost a factor of two.
76 - croak when a coro that is being cancelled tries to block
77 (e.g. while executing a guard block), instead of crashing or
78 deadlocking.
79 - use a more robust and also faster method to identify Coro::State
80 objects - speeds up everything a bit.
81 - implement Coro->cancel in XS for a 20% speed improvement, and to
82 be able to implement mutual cancellation.
83 - speed up context switches by a percent or two by more efficiently
84 allocating context stack entries.
85 - implement Coro->join and Coro->on_destroy in XS for a speedup and
86 a reduction in memory use.
87 - cancelling a coro while it itself is cancelling another coro is
88 now supported and working, instead of triggering an assertion.
89 - be a bit more crash-resistant when calling (buggy) on_destroy
90 callbacks (best effort).
91 - move on_destroy into the slf_frame, to allow extension slf
92 functions to have destructors.
93 - get rid if coro refcounting - simply crash in other interpreter
94 threads by nulling the pointers on clone.
95 - simplify warn/die hook handling when loading Coro - the convoluted
96 logic seems to be no longer neccessary.
97 - use libecb instead of our own home-grown gcc hacks.
98 - document alternatives to Coro::LWP. Please use them :)
99 - work around another mindless idiotic NEEDLESS bug in openbsd/mirbsds
100 sigaltstack. Really. wine suffers from it, erlang suffers from it,
101 and it's known since at least 2006.
102
1035.372 Wed Feb 23 06:14:30 CET 2011
104 - apparently mingw doesn't provide a working gettimeofday, try to
105 work around that by relying on Time::HiRes (indirectly brought to
106 my attention by Max Maischein).
107 - fix some portability issues when Time::HiRes was used.
108
1095.371 Mon Feb 21 14:36:08 CET 2011
110 - backport to windows process emulation code again.
111
1125.37 Sat Feb 19 07:49:44 CET 2011
113 - add a big "Coro thread life cycle" section to "man Coro".
114 - try a tentative workaround against the breakage that 5.13 has
115 introduced without depreciation period. sigh.
116 - no longer use Time::HiRes if gettimeofday is available, which
117 saves quite a lot of memory.
118
1195.36 Sun Feb 13 05:33:41 CET 2011
120 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
121 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
122 method call.
123 - undocument Coro::Timer::sleep and obsolete whole module.
124 - optimise Coro::Timer::timeout memory and cpu usage.
125 - slightly updated Coro::Intro for recent changes.
126 - do not initialise PL_dirty anymore.
127
1285.25 Thu Nov 11 01:08:39 CET 2010
129 - try a different approach on netbsd - netbsd 5 finally has marginally
130 working pthreads, but still broken ucontext/sigaltstack.
131 - openbsd 4.8 finally got their act together, Coro works out of the box
132 with asm, setjmp and pthreads (no change, just informational).
133
1345.24 Sat Oct 23 11:27:12 CEST 2010
135 - port to the EV 4.0 API.
136 - work around bugs in mingw32, making strawberry perl work
137 out of the box.
138 - correctly modify Coro::AIO function prototypes
139 so that they reflect the "no optional parameters" rule.
140 - "ported" libcoro to C++.
141
1425.23 Mon May 17 18:50:42 CEST 2010
143 - be more resistant to ordering changes when initialising
144 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
145 Waldorf).
146 - document that perl 5.12 deliberately removed support for cloning.
147
1485.22 Wed Apr 14 03:55:35 CEST 2010
149 - correctly return udnef on errors in Coro::Handle::read/write
150 (testcase by Marc Mims).
151 - convert Coro::Util into a "perl compatibility wrapper" - the functions
152 are less useful now, but are drop-in replacements for existing
153 functions, listing better alternatives in the documentation. This also
154 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
155 with Coro::Util::inet_aton.
156 - do not override $Coro::idle unconditionally in Coro.pm, as other
157 modules could have provided their own idle coro already
158 (for exmaple, Coro::AnyEvent).
159 - fix Coro::Util::gethost* functions.
160 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
161
1625.21 Wed Dec 16 07:19:51 CET 2009
163 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
164 together.
165 - add some examples on how to combine other event loops with Coro in
166 Coro::AnyEvent, and how to run it (and not to block). Seems to be
167 the most common source of confusion.
168 - try to catch people naively blocking in an event callback.
169 - work around the perl filehandle bug issue in conjunction with
170 older common::sense (as indirectly pointed out by ZSystem).
171 - clarify the "not from signal handlers" section.
172
1735.2 Sun Oct 4 14:54:24 CEST 2009
174 - Coro::Storable destroyed the prototypes of the functions it wrapped.
175 - export rouse_cb and rouse_wait by default now.
176 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
177 - new method $state->swap_sv.
178 - added section on "windows process emulation" to the manpage,
179 after a not-so-fruitful (nor-friendly) "discussion" with chip
180 salzenberg (discussion implies arguments, but his only arguments
181 were ad-hominems, one wonders why he started it in the first
182 place). I hope this explains it well enough for him to understand,
183 and maybe well enough for others to understand.
184 - use common::sense everywhere now.
185 - idle callbacks are no longer supported, use idle coros instead.
186 - print a thread listing when a deadlock is detected.
187
1885.17 Sat Aug 22 23:09:31 CEST 2009
189 - work around a bug in the perl debugger causing crashes
190 when running under the debugger by marking _pool_handler as nodebug.
191 - speed up Coro::async considerably.
192 - try some hacks to get netbsd to work "more often" - their broken
193 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
194 - convert Coro to AE 5.0 API.
195
1965.162 Tue Jul 28 04:04:03 CEST 2009
197 - perl 5.8.2 is now minimum requirement.
198 - skip t/19_handle.t on broken windows perls.
199
2005.161 Wed Jul 22 04:47:38 CEST 2009
201 - Coro::AnyEvent::poll could have a different prototype when EV was
202 used as backend (analyzed by Tatsuhiko Miyagawa).
203 - Coro::AnyEvent errornously initialised the event loop when loaded,
204 not on demand.
205 - try to workaround rare */t/01_unblock.t failures.
206
2075.16 Tue Jul 21 01:44:37 CEST 2009
208 - Coro::AnyEvent failed to hook into the event loop
209 when no threads had been readied between detecting
210 the event loop and actually running it.
211 - considerably speed up Coro::Select by taking avdantage
212 of AnyEvent > 4.8 and some other optimisations.
213 - implement paragraph readline mode in Coro::Handle
214 (based on patches by Zsbán Ambrus).
215 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
216 and analyzed by Yasuhiro MATSUMOTO).
217 - clarified libcoro license and copyright.
218 - someone stole my EXTRA_META!!!
219 - implement Coro::Select::patch_pp_sselect and it's brother,
220 for hardcode select overriding.
221
2225.151 Mon Jul 6 05:41:57 CEST 2009
223 - backport to windows process emulation code again (patch by
224 Yasuhiro MATSUMOTO).
225 - slightly update Coro::MakeMaker.
226
2275.15 Tue Jun 30 10:28:06 CEST 2009
228 - deprecate Coro::Socket, document how to get ipv6 support via
229 AnyEvent::Socket instead.
230 - implement signal->wait ($cb) interface, similar to semaphores.
231 - work around SvOK not supporting getmagic, so we have to getmagic
232 to test for undef :/ (reported by Matthias Waldorf).
233 - load Coro::AnyEvent in all modules using AnyEvent.
234 - work around perl corrupting our internal data structures,
235 reported by Tokuhiro Matsuno.
236 - enable per-coroutine real and cpu time gathering
237 (Coro::State::enable_times).
238
2395.14 Wed Jun 24 01:37:48 CEST 2009
240 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
241 as this seems to have been a difficult concept (poll, sleep, idle,
242 idle_upto).
243 - add Coro::AnyEvent::readable/writable functions.
244 - clarify Coro::EV/Event/AnyEvent manpages.
245 - free per-thread global scalars in the thread calling ->cancel, to
246 avoid crashes when $_, $@ etc., are magical but some of those
247 had already been freed.
248 - "unexperimentalise" the callback interface for Coro::Semaphore.
249 - speed up ready queue management/context switching by using a linked
250 list instead of an array (~5-10%).
251 - implement "watch" command in Coro::Debug shells.
252 - for fun, implement time-slicing as an example in the manpage.
253 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
254 Coro::Event, use the normal AnyEvent handling instead of dieing
255 (the same is true for Coro::Handle).
256 - properly document Coro::EV::timed_io_once.
257 - avoid unneccessary ->cancel calls in Coro::Handle.
258 - maybe make it work on mingw32 with win32 backend
259 (based on patch by Yasuhiro Matsumoto).
260
2615.132 Fri May 29 09:00:39 CEST 2009
262 - do not keep a reference to the argument itself in
263 Coro::Semaphore::guard, as it could change later.
264 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
265 should support a prepare callback.
266
2675.131 Mon Mar 16 23:20:37 CET 2009
268 - implement and document Coro->suspend, Coro->resume.
269 - fix Coro::Select implementation to not (often) close
270 the passed file descriptors (testcase provided by pippijn).
271
2725.13 Mon Dec 15 21:51:42 CET 2008
273 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
274 - don't set diehook to C<undef> but instead to NULL, to avoid
275 spurious warnings.
276 - fix a lot of bugs in Coro::SemaphoreSet.
277 - Coro::SemaphoreSet will less often create a semaphore needlessly.
278 - add Coro::SemaphoreSet::count and wait methods.
279 - take advantage of the new Guard module.
280 - deprecate Coro::guard.
281 - try to fix the dreaded 01_unblock tests once more. I hate it when
282 testsuites need more fixing than the code they are supposed to test.
283 - croak in more cases when a required callback isn't resolvable.
284 - fix some minor issues in Coro::State->call/eval.
285 - use current coroutine context instead of a temporary one
286 when temporarily switching to another coroutine.
287 - do not call C-level on_destroy handlers during global destruction,
288 to avoid needless segfaults.
289
2905.12 Sun Dec 7 13:30:38 CET 2008
291 - add default config for MirOS, which seems to be a bug-to-bug
292 compatible fork of openbsd ("world domination by releasing
293 openbsd cvs before the openbsd folks do it" or so :).
294 - free_padlist did destroy the names pad, not good, but didn't
295 seem to bother perl - this could fix issues such as eval ""
296 inside a function called from multiple coroutines.
297 - use a different method to detect destruction time.
298 - be more careful when freeing padlists just before global
299 destruction.
300 - fixed and expanded the call/cc example.
301 - renamed _terminate to _coro_run.
302 - new method Coro::Channel->shutdown.
303 - try pthreads on openbsd <4.4 (broken sigaltstack, will
304 pthreads fare better?).
305 - be less picky about destroying "a" running coroutine.
306
3075.11 Tue Nov 25 21:49:05 CET 2008
308 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
309 working code with failed assertions, introducing more bugs than
310 it fixes, requiring elaborate workarounds :(
311
3125.1 Mon Nov 24 08:54:59 CET 2008
313 - wrote a small introductory tutorial - Coro::Intro.
314 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
315 - Coro::Select did errornously dup the file descriptors
316 and didn't work with all AnyEvent backends.
317 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
318 LWP data transfers.
319 - disassociate c contexts from coro objects - this is agruably more
320 correct, but mostly allows sharing of cctxs between coro and state
321 objects, for added memory savings and speed increases.
322 - bumped $Coro::POOL_RSS up to 32kb by default.
323 - no longer set the optype to OP_CUSTOM, as B::* understandably
324 doesn't like this very much (and we *are* a type of entersub).
325 - implement state cloning, just to prove that call/cc can be done.
326 - automatically load Coro::AnyEvent in Coro::Handle.
327 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
328 do not have this method (and don't need it either).
329 - speed up generic anyevent methods in Coro::Handle by using rouse
330 callbacks.
331 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
332 others. It also makes the debugger happier, as you can trace
333 through the idle threads now.
334 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
335 - eg/event was pretty broken.
336 - better 5.8.6 compatibility.
337
3385.0 Thu Nov 20 10:35:05 CET 2008
14 - NEW ARCHITECTURE: use the latest 4.x version if you experience 339 - NEW ARCHITECTURE: use the latest 4.x version if you experience
15 stability issues. 340 stability issues.
16 - bump API version to 7 - all dependents must be recompiled. 341 - bump API version to 7 - all dependents must be recompiled.
342 - removed timed_* functions - they were not being used anyways
343 and should be replaced by a more generic mechanism -
344 and were annoying to support anyways :)
345 - removed SemaphoreSet's waiter method - use sem method instead.
17 - Coro::Semaphore->adjust didn't correctly wake up enough waiters. 346 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
18 - async_pool did free a scalar value twice 347 - async_pool did free a scalar value twice
19 ("Attempt to unreference..."). 348 ("Attempt to unreference...").
20 - fix a longstanding bug where calling terminate on a coro that 349 - fix a longstanding bug where calling terminate on a coro that
21 was waiting for a semaphore that was just becoming available 350 was waiting for a semaphore that was just becoming available
22 would cause a deadlock (semaphore would get into a state where 351 would cause a deadlock (semaphore would get into a state where
23 it was available but waiters were waiting(. 352 it was available but waiters were still blocked).
24 - avoid deadlock in Coro::Channel if maxsize == 1 (Richard Hundt). 353 - calling throw on a coroutine that is waiting for a semaphore will
354 no longer make it acquire the semaphore (and thus leak a count).
25 - perl's process emulation is now not even theoretically supported 355 - perl's process emulation is now not even theoretically supported
26 anymore. 356 anymore.
357 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
358 conversion of callback-style to blocking-style.
359 - new methods $coro->schedule_to and ->cede_to, to specifically
360 schedule or cede to a specific coroutine.
361 - new function Coro::Semaphore::wait.
362 - use named constants in Coro::Channel (Richard Hundt).
27 - directly patch the entersub opcode calling SLF functions (cede, 363 - directly patch the entersub opcode calling SLF functions (cede,
28 transfer and so on). this does speed up context switching, but 364 transfer and so on). this does speed up context switching, but
29 more importanly, it frees us from the hardcoded behaviour of 365 more importanly, it frees us from the hardcoded behaviour of
30 entersub, so we might actually be able to return something from 366 entersub, so we might actually be able to return something from
31 those functions and atcually create new ones. 367 those functions and atcually create new ones.
… …
34 as "threaded" perls are running at half speed anyways). 370 as "threaded" perls are running at half speed anyways).
35 - implement execute_slf (schedule-like-function) interface that makes 371 - implement execute_slf (schedule-like-function) interface that makes
36 it possible to implement schedule-like-functions in XS. 372 it possible to implement schedule-like-functions in XS.
37 - use new SLF interface to massively speed up Coro::EV by roughly a 373 - use new SLF interface to massively speed up Coro::EV by roughly a
38 factor of two. 374 factor of two.
39 - removed timed_down and timed_guard functions - they were not being
40 used anyways and should be replaced by a more generic mechanism -
41 and were annoying to support anyways :)
42 - used new SLF interface to massively speed up Coro::Semaphore by a 375 - used new SLF interface to massively speed up Coro::Semaphore by a
43 factor of three. 376 factor of three.
44 - used new SLF interface to speed up Coro::AIO by roughly 40-60% and 377 - used new SLF interface to speed up Coro::AIO by roughly a factor of
45 reduce its memory usage considerably. 378 four and reduce its memory usage considerably.
379 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
380 for a nice speedup and vastly more correct behaviour. Also implement
381 a new method "sem" to get at the underlying semaphore object.
382 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
383 (in comparison) 20-40% speedup.
384 - used new SLF interface to reimplement Coro::Signal gaining
385 some unknown (because I was too lazy), but certain, speedup, and also
386 making signals reliable for the first time.
387 - used new SLF interface and other optimisations to speed up async_pool
388 by a factor of two. It also doesn't rely on perl's exception mechanism
389 to exit anymore. The overhead for terminating an async_pool, coro over
390 a normal async is now very small.
391 - sped up coroutine creation/destruction by 40%.
46 - forgot to include Coro/libcoro/README in the dist for all these years. 392 - forgot to include Coro/libcoro/README in the dist for all these years.
47 - work around a freebsd pthreads bug (manual testcancel is required as 393 - work around a freebsd pthreads bug (manual testcancel is required as
48 pthread_cond_wait isn't a cancellation point on freebsd). 394 pthread_cond_wait isn't a cancellation point on freebsd).
395 - use new rouse functions to speed up and simplify Coro::BDB.
396 - make "prefer perl native functions" work with threaded perls.
397 - condense Coro::Debug ps output, hint at v and w flags.
398 - (libcoro) lots of minor cleanups and portability improvements.
399
4004.914 Wed Nov 19 12:54:18 CET 2008
401 - fix a disastrous bug in the readline optimisation
402 introduced in 4.801.
49 403
504.913 Sat Nov 15 07:58:28 CET 2008 4044.913 Sat Nov 15 07:58:28 CET 2008
51 - async_pool did free a scalar value twice 405 - async_pool did free a scalar value twice
52 ("Attempt to unreference..."). 406 ("Attempt to unreference...").
53 407

Diff Legend

– Removed lines
+ Added lines
< Changed lines
> Changed lines