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