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