ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/Coro/Changes
Revision: 1.206
Committed: Tue Feb 13 19:21:29 2007 UTC (17 years, 3 months ago) by root
Branch: MAIN
Changes since 1.205: +2 -0 lines
Log Message:
*** empty log message ***

File Contents

# Content
1 Revision history for Perl extension Coro.
2
3 TODO: better (earlier) error messages on broken transfer's
4 TODO: destroy gets only called on cancel: fix(?)
5
6 3.5
7 - do AnyEvent model detection earlier, avoiding problems
8 caused by first using AnyEvent and later Coro::Event.
9 - implement and document Coro::Event event objects.
10 - fix a potential problem in Coro::Event causing crashes.
11 - initialise PL_comppad when creating a new coroutine,
12 avoids crashes on early coro destruction.
13
14 3.41 Mon Jan 22 19:19:49 CET 2007
15 - readline on Coro::Handle did not support undefined $/,
16 nor did it deliver partial lines on EOF or error.
17 - implement malloc fallback for stack allocation because
18 stupid broken idiotic OSX has a stupid broken
19 idiotic fits-the-whole-os mmap "implementation" and
20 my dick feels longer if Coro is portable even to
21 obsolete platforms.
22
23 3.4 Fri Jan 19 21:52:54 CET 2007
24 - remove t/09_timer.t, as it isn't really testing much
25 but was rather flaky in practise.
26 - async_pool coro would keep arguments and callback alive until
27 it was reused.
28 - cancellation of a coroutine could cause spurious idle calls
29 in cede_notself.
30
31 3.3 Sat Jan 6 03:45:00 CET 2007
32 - implement $coro->on_destroy.
33 - Coro::Event blocking semantics have been changed,
34 documented and - hopefully - improved.
35 - fix nice adding, not subtracting, from priority.
36 - fix ->prio and api_is_ready (patch by Mark Hinds).
37 - fixed an assert ("... == prev__cctx->idle_te")
38 that could errronously trigger.
39 - fix various large and small memleaks.
40 - use a (hopefully) more stable cancel implementation
41 that immediately frees the coroutine data.
42 - cede/cede_notself return a status now.
43 - added Coro::guard function.
44 - added a global coroutine pool for jobs (on my machine,
45 I can create and execute 48k simple coros/s with async,
46 and 128k coros with async_pool).
47 - Coro::AIO now uses the coroutine priority as io priority.
48
49 3.2 Fri Dec 22 05:07:09 CET 2006
50 - improve portability to slightly older perls.
51 - use cleaner coroutine destruction.
52 - simplify configuration for users.
53 - optionally (unrecommended) prefer perl functions over
54 their coro replacements.
55
56 3.11 Tue Dec 5 13:11:24 CET 2006
57 - fixed some bogus assert's, but as perl.h disables assert even
58 without NDEBUG (thank you very much), not too many people should
59 notice (that did include myself). Andreas König noticed, though :)
60 - do not save/restore PL_sortcxix on >= 5.9.x, it doesn't seem to have
61 it. Also noticed by Andreas König :)
62 - save/restore tainted status.
63 - verified to pass the testsuite on my 5.9.5.
64
65 3.1 Mon Dec 4 23:03:40 CET 2006
66 - INCOMPATIBLE CHANGE: $/ is now per-coroutine (but slow).
67 - incompatible change: transfer flags are now per-state.
68 - give a better error message on deadlock.
69 - document Coro::nready.
70 - enhanced testsuite.
71
72 3.01 Sun Dec 3 23:47:42 CET 2006
73 - forgot to include Coro::Timer.
74
75 3.0 Sun Dec 3 22:57:25 CET 2006
76 - the "FINALLY COMPLETELY STABLE" release (coming soon:
77 the "FAMOUS LAST WORDS" release).
78 - implement a new stack sharing algorithm, which uses a stack
79 pool (size currently hardcoded).
80 - make stack sharing mandatory (it no longer uses a heuristic).
81 - eval/die no longer cause weird problems under heavy use.
82 - Coro::Event could cause livelocks if it was used but
83 no Coro::Event watchers were used.
84 - Coro::Event now uses asynccheck as prepare does not
85 check for changed watchers.
86 - Coro::Event allows multiple waiting coros and will wake up one
87 per event.
88 - Coro::Event should be cleaner and more efficient now.
89 - new utility function Coro::unblock_sub.
90 - document the sad fact that Event is no longer reentrant.
91 - putting a coroutine into the ready queue twice could under
92 some circumstances lead to stack corruption.
93 - minor incompatibility: subclassing Coro::State is supported
94 directly now without going through a _coro_state member.
95 - state/coro switching is much faster now.
96 - very minor optimisations and code/documentation cleanup.
97 - avoid problems due to compiler inlining.
98 - removed timers from Coro::Timer -> use AnyEvent instead.
99 - replaced Coro::idle coroutine by (cleaner) idle handler.
100 - updated to newest libcoro.
101 - implement enhanced support for valgrind.
102 - implement is_ready and return value for ready.
103 - removed Coro::Cont, it was a misnomer (it's generators),
104 and it was somewhat annoying to get it right. Will come back
105 if somebody asks for it :)
106 - many ->wait methods and Coro::Event could return
107 spuriously without the event having happened.
108
109 2.5 Tue Nov 7 12:22:33 CET 2006
110 - made Coro::Util, Coro::Select, Coro::Handle and Coro::Socket
111 use AnyEvent, moved them to Coro/.
112 - added Coro::LWP which contains all the uglyness required to
113 make LWP non-blocking.
114 - should work with perl 5.9.x now (Andreas König made me do it).
115 - fixed another bug in Coro::Select when the timeout was undef.
116 - reuse PL_start_env for all coros, saves some memory per coroutine.
117 - manage PL_top_env differently, hopefully to avoid panic: top_env.
118 - timeout argument was not properly used in Coro::Socket.
119 - allow limited forms of subclassing in Coro::Handle/Coro::Socket.
120 - emulate undocumented(!) functionality of IO::Socket required
121 by LWP(!!).
122 - updated eg/lwp to work with newer lwp's.
123 - remove "FATAL: uncaught exception" prefix. Coroutines that die
124 kill the whole process, just as exceptions in the main "coroutine"
125 did already.
126
127 2.1 Wed Nov 1 23:01:13 CET 2006
128 - fix a long-standing bug in Coro::Select where select with
129 zero timeout would instead change the current default filehandle.
130 - use a simpler and hopefully more robust way to clone padlists
131 (uses less memory and a perl function instead of our own).
132 - coro can now create a stack guard area on many architectures.
133 - Coro::AIO properly reexports additional functions from IO::AIO.
134 - updated libcoro with a workaround for OS X,
135 pach and testing by Michael Schwern.
136
137 2.0 Tue Oct 24 05:47:17 CEST 2006
138 - support additional aio requests in Coro::AIO.
139
140 1.9 (never properly released due to a glitch)
141
142 1.8 Thu Feb 2 00:59:06 CET 2006
143 - applied suggested patch by SAMV to avoid problems during stupid
144 mark & sweep gc run.
145 - applied patch by Scott Walters for 5.9.3 compatibility.
146
147 1.7 Tue Dec 27 01:41:58 CET 2005
148 - added Coro::AIO, a thin wrapper around IO::AIO.
149 - improved Makefile.PL explanations.
150
151 1.6 Mon Dec 12 21:30:05 CET 2005
152 - additionally save PM_curpm might fix as-of-yet
153 unseen problems with regex matches being attributed to the
154 wrong package.
155 - add t/10_bugs.t, which currently checks against imho broken
156 perls that use 0.26MB of stack space per Perl_magic_get
157 invocation (newer linuxes) as opposed to the 0.0002MB perl
158 normally uses.
159 - make stacksize configurable for the ultimate debian experience.
160
161 1.51 Mon Dec 12 18:48:36 CET 2005
162 - remove debugging warn() accidentally left in Coro::Select.
163
164 1.5 Tue Nov 29 13:32:44 CET 2005
165 - use Coro::Event inside Coro::Select to avoid spurious deadlocks.
166 - fix Coro::Select 'select'.
167 - strict'ify some modules.
168 - libcoro errornously restored the SIGUSR2 handler to SIGUSR1.
169 - use XSLoader in selected modules.
170 - remove some 5.6 compatibility cruft.
171
172 1.4 Tue Sep 6 00:11:05 CEST 2005
173 - libcoro did not take into account the trampoline on amd64,
174 when 'l'inux method was used.
175
176 1.31 Tue Aug 30 23:31:33 CEST 2005
177 - some portability fixes/workarounds.
178
179 1.3 Sat Aug 20 03:08:56 CEST 2005
180 - no code changes. module seems to work fine.
181 - improve Coro::State docs, remove reference to nonexisting
182 Coro::State::flush method.
183 - no longer autodetect windows, present it as an option instead.
184
185 1.2 Mon May 16 02:00:55 CEST 2005
186 - included libcoro.c earlier in State.xs, might improve portability.
187 - use a faster and possibly more stable Coro::Cont implementation.
188 - accept x86_64 in addition to amd64 for optimized linux-amd64 support.
189 - fix bugs Coro::SemaphoreSet that could cause locks never to be freed.
190 - fix bugs in CoroAPI.h and document it in Coro::MakeMaker.
191
192 1.11 Thu Mar 3 18:00:52 CET 2005
193 - change of contact address.
194
195 1.1 Tue Feb 22 20:51:16 CET 2005
196 - support [l]inux method on amd64.
197 - allow some context switches while compiling/eval'ing.
198
199 1.0 Tue Aug 10 03:47:27 CEST 2004
200 - Event 0.89 fixes the bug again.
201 - fix the remaining(??) memleak. I didn't free the cache-AV for
202 padlists, so one AV per created anon-sub leaked.
203
204 0.97 Fri May 14 15:23:32 CEST 2004
205 - remove stupid gcc extension that was an experiment only anyway.
206
207 0.96 Thu May 13 18:09:29 CEST 2004
208 - tell people if they have yet another broken Event variant.
209 - made Coro to work _better_ with threads (it seems that, as long
210 as you use Coros in one thread only everything will work fine.
211 Looking at the perl source the problem might be missing locking
212 between Coro and some perl internal routines. That's not easily
213 fixable).
214 - (hopefully) got rid of the static xnext inside transfer.
215
216 0.95 Sun Feb 15 03:20:28 CET 2004
217 - removed Coro::State::flush, implemented a new and much
218 faster caching mechanism by attaching magic to the CV.
219 - a memleak with (real) closures remains.
220 - dramatically reduced size of stacks to 65536 (linux/x86), after
221 observing that even myhttpd never had more than a kilobyte
222 of stack in use (stack is still only physically allocated in
223 pagesize increments for systems suporting mmap).
224 - fix for a "cygwin-64" architecture added, proposed by Stefan Traby.
225 - fix for threaded perl.
226
227 0.9 Sun Nov 23 04:32:39 CET 2003
228 - fix a few very hard to track down but (of course) stupid
229 memory leaks.
230 - stringify version number to avoid locale problems :(.
231 - perl 5.9.0 does not have SvPADBUSY, reported by Scott Walters,
232 hopefully ignoring it "just works".
233
234 0.8 Wed Nov 5 19:38:40 CET 2003
235 - port to perl5.005_03, but only 5.8.x+ is supported!
236 - honor LocalAddr even without LocalPort in Coro::Socket.
237
238 0.7 Tue May 27 03:12:38 CEST 2003
239 - the version jump indicates some level of testing, not gobs
240 of new features.
241 - uh... I found the prompt function in ExtUtils::MakeMaker.
242 Highly correct stuff, that is...
243 - fixed(?) a bug with die's in coroutines causing "panic: top_env".
244 the fix is not well understood by the author, so beware :(.
245
246 0.652 Thu May 8 02:54:46 CEST 2003
247 - Applied patch by Slaven Rezic to set default to "s" on FreeBSD,
248 cause version 4 doesn't have ucontext.
249 - Benjamin Reed reported that setjmp works fine on darwin,
250 so preselect it.
251
252 0.651 Sat Mar 29 15:00:23 CET 2003
253 - fix a bug in Coro::Handle where some bytes could get lost
254 on reading, reported by jason@nichego.net.
255
256 0.65 Sun Mar 23 00:08:26 CET 2003
257 - added a README.linux-glibc.
258 - new module Coro::Select.
259 - also save/restore PL_comppad, fixes segfaults with 5.8.1.
260 (I never claimed that I know what I am doing ;).
261 - set default to setjmp/longjmp on non-x86-linux.
262
263 0.6 Thu Nov 21 11:09:06 CET 2002
264 - made lazy context switching the default.
265 - fixed the bug where SemaphoreSet::guard
266 was the same as timed_guard and timed_guard was missing.
267 - fix a memleak in Coro::Timer.
268 - reclassified context sharing as not experimental.
269
270 0.534 Sun Apr 14 03:05:12 CEST 2002
271 - fixed backspace => tab in header parsing (myhttpd).
272 - added eg/readline, for elmex.
273 - Coro::Event now calls ->start on first call instead of
274 ->again, which created wrong timeouts for the first ->next.
275 - fixed a bug where Coro::Socket returned a closed socket instead
276 of nothing on ECONNREFUSED and similar error conditions.
277
278 0.533 Mon Feb 18 18:49:40 CET 2002
279 - adapt to finally working Event-0.85.
280 - fixed buggy event initialization (trapped under windows, but how
281 could it possibly have worked anywhere?)
282
283 0.532 Wed Jan 16 02:45:32 CET 2002
284 - removed the reference to pp_entersub, might work on
285 (some) win32 perls now (testsuite works under cygwin,
286 without Event, which freezes).
287
288 0.531 Mon Dec 10 22:18:44 CET 2001
289 - Fixed a bug in SemaphoreSet::timed_down.
290
291 0.53 Tue Nov 27 21:11:13 CET 2001
292 - some tests for Event.
293 - slightly different internal architecture to get refcnt into the object
294 and not the reference. might fix a few bugs, certainly did introduce
295 new ones.
296 - Coro::Timer (independent of Event).
297 - new timed_wait functions for Coro::Signal, Semaphore, SemaphoreSet.
298
299 0.52 Tue Nov 6 21:36:18 CET 2001
300 - ported to cygwin (trivial).
301 patch by Gerrit P. Haase <gp@familiehaase.de>.
302 - small setjmp code fix by Sullivan.DanielJ@epamail.epa.gov.
303
304 0.51 Thu Nov 1 20:39:01 CET 2001
305 - terminate/cancel now work properly (otherwise termination
306 could cause a "next coroutine is not and contains not..."
307 error.
308 - added Coro::Socket::shutdown.
309 - Coro::Event::loop no is the same as Event::loop.
310 - implemented terminate with args + join.
311
312 0.5 Fri Sep 28 16:15:35 CEST 2001
313 - fixed "print" on a Coro::Handle. The print method worked.
314 - small tweaks (seem to reduce memory consumption a lot)
315 in various modules.
316 - splendid use of "no warnings" scattered throughout.
317 - added Coro::Handle::rbuf, fixed Coro::Handle::fh.
318
319 0.49 Sun Sep 16 02:42:45 CEST 2001
320 - changed some method calls to function calls for speed
321 inside Coro::Handle.
322 - make Coro::Handle use an array instead of a hash for
323 speed reasons.
324 - IRIX mystery solved: it's SGI's NT, after all: "standard, huh?".
325 sjlj and ucontext should now work.
326 - IRIX-specific port for libcoro.
327 - swapped order of accept results to match IO::Socket.
328 - changed getsock/peername to sock/peername to match IO::Socket.
329 - fixed a bug that caused segfault when returning to main task
330 under some circumstances.
331 - other bugfixes.
332
333 0.45 Sun Sep 2 02:54:01 CEST 2001
334 - new method Coro::Handle::timeout.
335 - corrected speling of set...name to get...name in Coro::Socket.
336 - Coro::Socket::accept now returns a Coro::Socket, not a Coro::Handle.
337 - Coro::Handle now supports fileno and FILENO.
338 - added eg/myhttpd, a web server.
339 - Coro::Socket now accepts numeric ports as well as "port(number)"
340 syntax.
341 - moved some scheduling primitives into xs code. more to come.
342 - new simple priority system.
343 - implemented Coro::Event using XS, almost four times faster!
344 - small memory corruption problem fixed (boy that was difficult).
345
346 0.13 Wed Aug 8 16:53:07 CEST 2001
347 - new method Coro::Handle::readline.
348 - added eg/lwp and Coro::Handle::autoflush, to make LWP work.
349 (see eg/lwp on how to make LWP non-blocking).
350 - renamed Coro::Socket::new_inet to new.
351 - added Coro::Util, some utility functions.
352 - die/eval now works better (still get annoying (but true ;)
353 "Callback called exit" messages).
354 - Coro::Handle now supports timeout, and Coro::Socket Timeout.
355 - much better Coro::Socket support.
356
357 0.12 Fri Jul 27 04:19:01 CEST 2001
358 - do not use mmap if neither MAP_ANON nor MAP_ANONYMOUS is defined.
359 - very experimental stack sharing algorithm. not 100% safe but
360 should work well in practise ;)
361 - added Coro::RWLock.
362 - Coro::Specific now works.
363
364 0.11 Tue Jul 24 22:49:21 CEST 2001
365 - added specialized hack for newer and older linux versions (fast).
366 - renamed Coro::Event::IO to Coro::Handle.
367 - new module Coro::Socket.
368
369 0.1 Tue Jul 24 01:47:53 CEST 2001
370 - release candidate 3. A rather new internal structure :(
371 - the great renaming: Damian Conway gave me a suitable replacement
372 for yield.
373 - added Coro/libcoro, a portable coroutine implementation for C
374 (not even perl-dependent!!!), which can be used by Coro::State.
375 - renamed SAVE_DEFSV (on xs-level only) to avoid symbol clash in
376 perl-5.6.
377 - new function Coro::Event::idle.
378 - the idle process is now overriden
379 by default in Coro::Event.
380 - Coro::Channel now enforces the size.
381 - canceling events now works in all cases.
382 - Coro state now include $_ and $@.
383 - yet another bug workaround that I do not really understand :(
384 - new module Coro::Event::IO (very undocumented) to do
385 non-blocking i/o.
386 - performance tuning.
387
388 0.08 Thu Jul 19 06:13:25 CEST 2001
389 - release candidate 2 (new functionality).
390 - add Coro::State::flush function.
391 - transfer now supports a flags value (mostly for speed ATM).
392 - might compile and work in the presence of threads now.
393 - continuations have a saner syntax.
394 - no more memleaks.
395
396 0.07 Tue Jul 17 17:40:18 CEST 2001
397 - release candidate 1 ;)
398 - slightly nicer code.
399 - fixed a scheduling bug in Coro::Event.
400
401 0.06 Tue Jul 17 04:23:24 CEST 2001
402 - ok, I found the showstopper - the same sub must not be
403 re-entered in two different coroutines, otherwise => crash. I
404 see no easy solution to this problem, except by walking the call
405 chaing and saving/restoring the cv's, which is what I do now.
406 - memory leaks still latent, especially at thread termination.
407 - Coro::Event now works (haha).
408
409 0.05 Sun Jul 15 17:32:20 CEST 2001
410 - fixed a few issues in Coro::Event.
411 - I forgot to include Coro::Event in 0.04 :(:(:(
412
413 0.04 Sun Jul 15 05:24:59 CEST 2001
414 - @_ is now properly localized.
415 - Coro::State is now easier subclassable.
416 - Coro::Cont now coroutine-aware.
417 - Coro::Specific is a low-overhead module to create
418 coroutine-specific vars.
419 - Coro::Event provides a simple interface to Event.
420
421 0.03 Fri Jul 13 14:51:52 CEST 2001
422 - transfer() now implemented in XS (beware).
423 - new module Coro::Cont for really faked continuations.
424 - big internal architecture changes: Coro::State is now
425 really low-level and can thus be used to implement other
426 interesting things, While "Coro::" implements a process-like
427 model. Still crude and subject to change.
428 - $_ and $@ are no longer being localized.
429
430 0.02 Tue Jul 10 01:38:17 CEST 2001
431 - implemented "async" attribute.
432 - $_ and $@ are now localized.
433 - added Coro::Channel.
434 - more testcases, still no docs.
435
436 0.01 Tue Jul 3 02:18:41 CEST 2001
437 - original version; copied from Convert::Scalar.
438