ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/AnyEvent/Changes
Revision: 1.652
Committed: Mon Sep 20 21:46:30 2021 UTC (3 years, 1 month ago) by root
Branch: MAIN
CVS Tags: HEAD
Changes since 1.651: +1 -1 lines
Log Message:
*** empty log message ***

File Contents

# Content
1 Revision history for Perl extension AnyEvent.
2
3 TODO: document TCP_*** constants
4 TODO: add debug/trace logging to some modules
5 TODO: use the eval-with-cb-call trick also to speed up JSON?
6 TODO: FAQ: common pitfalls? invoke-before-return
7 TODO: FAQ: scope-ids? hosts file order?
8 TODO: AnyEvent::Handle with AnyEvent::IO?
9 TODO: invalid. and localhost. specialcasing inside AEDNS and not AESocket (rfc6761)
10 TODO: maybe implement env variable to give hosts precedence
11 TODO: more formats for parse_ipv6 (single ipv6 address without port, ...p80, ...#80)
12 TODO: inet_aton maybe pack C4? no, add a comment why not
13 TODO: socket prepare callbacks, can they easily signal failure as well?
14 TODO: Олег Г <verdrehung@gmail.com>, empty A record response causes
15 AnyEvent::resolve_sockaddr to eventually call pack_sockaddr with undef -
16 broken dns server, but maybe improve the response.
17 TODO: AnyEvent::Loop should not use POSIIX when ti isn't used.
18
19 TODO: AnyEvent::Log contexts should have a way to inject log messages received form elsewhere, using timestamp etc.
20 TODO: add some cheta-sheet to log synopsis regarding AE_LOG=AnyEvent::MP=+
21 TODO: tcp_connect option for AnyEvent::Handle?
22 TODO: hosts files on windows probably contain \r, which is not stripped
23 - AnyEvent::Loop no longer loads the POSIX module if Time::HiRes
24 has monotonic clock support (suggested by Felipe Gasper).
25 - since Net::SSLeay (up to 1.88-1) is _still not ported to openssl 1.1
26 and many distributions therefore will have a completely broken
27 Net::SSLeay module for the forseeable future, AnyEvent::TLS
28 will now log a warning (that is not seen by default unless you
29 increase log verbosity).
30 - expose error method of AnyEvent::Handle for general use.
31 - add tlsv1_3 disable switch to AnyEvent::TLS.
32 - add experimental "tls" method support for AnyEvent::TLS.
33 - tls methods were lowercase, but documented to be uppercase.
34 fix documentation.
35 - EV backend now uses EV::break when waiting for condvars,
36 making waiting for condvars more efficient. This might
37 "break" already-broken programs that recursively enter
38 the event loop even more.
39 - add a dependency on Coro 6.57 if the installed version is older,
40 to avoid an incompatibility.
41 - upgrade to UTS-46:13.0.0 ("Unicode 13").
42 - reduce lib/AnyEvent/Util/uts46data.pl data from ~68k to ~47k
43 by applying some simple data compression.
44
45 7.17 Wed Sep 18 03:04:49 CEST 2019
46 - work around antique openssl version in RHEL 7 by formatting
47 dh parameters differently (reported by several people).
48 - add t/13_weaken.t.
49
50 7.16 Fri Jul 19 18:00:21 CEST 2019
51 - add ffdhe group dh parameters from RFC 7919, and use ffdhe3072
52 as new default, instead of schmorp1539.
53 - AnyEvent::Log did not reassess logging status of
54 AnyEvent::Log::logger's when contexts were changed with
55 ->attach/detach/slaves, causing them to not log even though a
56 recent attach should have caused them to log.
57 - added some more logging configuration examples.
58 - mention RFC 8482 in AnyEvent::DNS.
59
60 7.15 Tue Feb 26 03:07:42 CET 2019
61 - INCOMPATIBLE CHANGE: AnyEvent::Handle's tls_detect documentation gave
62 separate major and minor versions, while code passed only a single
63 value. This version follows the documentation and now passes separate
64 major and minor values.
65 - work around Net::SSLeay not having been ported to openssl 1.1, but many
66 distributions compiling it against openssl 1.1, which unfortunately
67 succeeds and results in a very broken module.
68 - AnyEvent::DNS::dns_unpack now stores the original DNS packet
69 in the __ member, to allow decoding of undecodable resouce records
70 containing compressed domian names.
71 - AnyEvent::Socket::parse_ipv6 would NOT, as advertised, accept ipv4
72 addresses. It now does and converts them to ipv4 mapped addresses.
73 - support CAA records, based on patch by Steve Atkins.
74 - add freenom and cloudflare nameservers as dns fallback.
75 - AnyEvent::Strict would not properly ward against io watchers
76 on files when the handle passed was a file descriptor.
77 - document "internal" variables used by the dns en-/decoder to allow
78 enterprising users to extend them in a semi-official way.
79
80 7.14 Sat Jun 24 01:44:19 CEST 2017
81 - fix a crash bug in AnyEvent::Handle with openssl 1.1.0 (patched
82 by Paul Howarth and apparently tracked down by Mike McCauley).
83 - AnyEvent::Handle->keepalive was documented (and defined) twice
84 (found by Matt Merhar).
85 - AnyEvent::Socket::tcp_bind/tcp_server would immediately unlink a unix
86 listening socket unless a guard is used. Change this so that no
87 cleanup will be performed unless a guard is used and document this more
88 clearly.
89 - make tcp_bind/tcp_server error messages more regular.
90 - work around latest perl upstream breakage - this time, perl5porters were
91 apparently bullied by a bogus CVE (CVE-2016-1238), and since distros
92 stupidly apply anything that has "security fix" stamped on it, it's likely
93 going to be a problem in practise (and working around it is trivial on
94 windows and unix).
95 - add TCP_FASTOPEN/MSG_FASTOPEN and MSG_NOSIGNAL/MSG_MORE/MSG_DONTWAIT
96 to constants.
97 - update warnings to common::sense 3.74 standards.
98
99 7.13 Sat Sep 17 04:31:49 CEST 2016
100 - Only call tlsext_host_name for non-empty common names (reported
101 by Maxime Soulé).
102 - log a (single) notice message if SNI is not supported.
103 - upgrade to UTS-46:9.0.0 draft and switch to non-transitional
104 behaviour, beating thunderbird, ie, edge, chrome and safari to it :)
105 (see also https://bugzilla.mozilla.org/show_bug.cgi?id=1218179)
106 - turns out the UTS-46 IDNA testcase failures were indeed bugs in the
107 testcases and the specification and not in the code - the post-9.0.0
108 unicode files have all known problems fixed, so finally the AnyEvent
109 IDNA implementation can pass the full IDNA testsuite - without needing
110 a single fix :)
111 - guarantee (and document) that condvar callbacks will be removed
112 on invocation - important to avoid circular references.
113
114 7.12 Wed Jan 27 19:12:26 CET 2016
115 - use common name as hostname for tls connects, if Net::SSLeay
116 supports SNI.
117 - fix documentation of tls_autostart read type in AnyEvent::Handle,
118 analyzed by Felix Ostmann.
119
120 7.11 Thu Jul 16 14:36:00 CEST 2015
121 - AnyEvent::Socket::parse_ipv6 could accept malformed ipv6
122 addresses (extra "::" at end and similar cases).
123 - add a more explicit warning to AnyEvent::Handle that it doesn't
124 work on files, people keep getting confused.
125 - new function AnyEvent::Socket::tcp_bind.
126 - new functions AnyEvent::fh_block and AnyEvent::fh_unblock.
127 - aligned ipv6 address formatting with RFC 5952 (by not shortening
128 a single :0: to ::).
129 - added stability canary support.
130
131 7.09 Sat May 2 16:38:53 CEST 2015
132 - AnyEvent::Debug called an internal function (AnyEvent::Log::ft)
133 that was renamed to AnyEvent::Log:format_time. under its old name
134 (analyzed by Michael Stovenour).
135 - update AnyEvent::DNS fallback resolver addresses:
136 seems google effectively killed most other free dns resolvers,
137 so remove them, but add cable and wireless (ecrc) since it was
138 stable for 20 years or so, official or not, and there should be
139 an alternative to google.
140 - perl5porters broke windows error codes in 5.20, and mapped
141 WSAEWOULDBLOCK on the (different) EWOULDBLOCK error code, and
142 WSAEINPROGRESS into the incompatible ERINPROGRESS code, probably
143 because they were so cool. They probably broke other error codes
144 for no reason, too, but I didn't care for checking, it's too
145 depressing. This version only works around the WSAEWOULDBLOCK
146 issue, because I don't have a nice way to work around the
147 WSAEINPROGRESS bug.
148
149 7.08 Wed Dec 10 05:27:17 CET 2014
150 - work around a newly introduced bug in Socket 2.011 (an
151 errornous sun_length check) (analyzed by Maxime Soulé).
152 - AnyEvent::TLS didn't load (but refer to) AnyEvent::Socket
153 (analyzed by Ben Magistro).
154 - AnyEvent::Strict will now confess, not croak. This is in line with
155 it being a development/debugging tool.
156 - work around a number of libglib bugs (debug builds of libglib enforce
157 certain undocumented behaviour patterns such as not being able to
158 remove a child watch source after it has fired, which we will try
159 to emulate to avoid "criticals". what where they thinking...).
160 - mention json security issues in AnyEvent::Handle, now that Douglas
161 Crockford has foolishly and incompatibly changed JSON.
162 - changed default dns resolver "max_outstanding" value from 1 to 10,
163 the latter beinfg the intended value all along
164 (reported by Ilya Chesnokov).
165 - added new "AnyEvent::Impl::UV" interface module to the UV event lib
166 (written by Mike Lowell).
167
168 7.07 Tue Dec 17 17:45:02 CET 2013
169 - the documentation for custom tls verify schemes was wrong. make it agree
170 with the code (reported by Maxime Soulé).
171 - added cbor read and write types to AnyEvent::Handle (using CBOR::XS).
172 - work around an API change in openssl that could cause wrong tls connection
173 aborts, likely on windows only (analyzed by sten).
174 - calling AnyEvent->now_update with AnyEvent::Impl::Perl caused an
175 endless loop (reported by Dietrich Rebmann).
176 - add tlsv1_1 and tlsv1_2 protocols to AnyEvent::TLS
177 (patch by Maxime Soulé).
178 - document AnyEvent::Impl::IOAsync::set_loop and
179 $AnyEvent::Impl::IOAsync::LOOP. Though only documented now, this
180 functionality has _always_ been available.
181 - force a toplevel domain name in t/81_hosts.t (analyzed by
182 David Jack Wange Olrik).
183 - document that AnyEvent::Log uses AnyEvent::IO.
184 - warn about AnyEvent::Filesys::Notify performance.
185 - praise the joys of AnyEvent::Fork::*.
186 - time for an =encoding directive.
187 - do no longer use JSON to create a default json coder, use
188 JSON::XS or JSON::PP directly.
189
190 7.05 Wed Aug 21 10:38:08 CEST 2013
191 - uts46data.pl couldn't be found due to wrong naming of the file
192 (reported by Fulko Hew).
193 - handle lone \015's properly in AE::Handle's default line read
194 (reported by various people).
195 - untaint ip addresses found in /etc/hosts (patch by José Micó).
196 - the memleak fix in 7.03 caused resolving via /etc/hosts to always
197 fail on first use (reported and testcase by Andrew Whatson).
198 - expose AnyEvent::Log::format_time, and allow users to redefine it.
199 - expose AnyEvent::Log::default_format, and allow redefinition.
200 - expose AnyEvent::Log::fatal_exit, to allow redefinition.
201 - AnyEvent::Debug shell can now run coro shell commands, if available.
202 - t/63* tests were wrongly in MANIFEST.
203 - kernel.org's finger server went MIA, switch to freebsd.org and
204 icculus.org.
205 - clarify that IO::AIO and AnyEvent::AIO are needed for AnyEvent::IO
206 to function asynchronously (reported by Daniel Carosone).
207 - hard-disable $^W in most tests, it generates too much garbage output.
208 - use a (hopefully) more future-proof method to emulate common::sense.
209 - upgrade to UTS-46:6.2.0.
210 - switch to INSTLIB from INSTLIBDIR, as INSTLIBDIR was wrongly
211 documented. should not affect anything.
212
213 7.04 Thu Nov 15 02:23:58 CET 2012
214 - debugging code left in.
215
216 7.03 Thu Nov 15 02:16:11 CET 2012
217 - AnyEvent::Socket::inet_aton did not work when dns resolution was
218 used to find the addresses (analyzed and patch by Maxime Soulé).
219 - fix a memory leak in the /etc/hosts lookup code when hosts don't
220 resolve and are not in hosts (analyzed by sten).
221
222 7.02 Tue Aug 14 04:27:58 CEST 2012
223 - AnyEvent::Util::run_cmd could block indefinitely (analyzed and test
224 program by Yorhel).
225 - verified that AnyEvent::Socket follows RFC5952.
226 - try to parse "ADDR#PORT" in addition to "ADDR PORT".
227
228 7.01 Sun May 13 01:03:17 CEST 2012
229 - fail with EPROTO in AnyEvent::Handle wqhen TLS is requested but
230 not available, instead of throwing an exception.
231 - use File::Spec to get the tmpdir in t/*, to avoid needless
232 failures on (most, not mine :) windows boxes.
233 - new handle read types: tls_detect and tls_autostart.
234
235 7.0 Fri Apr 13 06:33:30 CEST 2012
236 - child watchers are broken in POE 1.352 (also many earlier
237 versions) and there seems to be no way to work around it, as POE
238 itself is inherently racy. Document this regression and add a delay
239 in t/68_poe_03_child.t for the time being.
240 - new module AnyEvent::IO, that is a frontend to either a pure-perl
241 synchronous I/O implementation (AnyEvent::IO::Perl), or to the
242 asynchronous I/O extension IO::AIO (AnyEvent::IO::IOAIO).
243 - load /etc/hosts only when DNS has no answer.
244 - stat /etc/hosts on every access and reload it if it changed.
245 - load /etc/hosts via AnyEvent::IO - potentially asynchronous.
246 - fix a buggy croak in the dh parameter reading in AnyEvent::TLS.
247 - AnyEvent::Log log_to_file and log_to_path now use AnyEvent::IO. As
248 a side effect, they now use true appending as opposed to libc
249 appending, and the file might not have been opened when the function
250 returns.
251 - the default logging level was not properly documented in a variety of
252 places, this has hopefully been rectified.
253 - updated uts46data.pl for unicode 6.1.0.
254 - made log messages generated by AnyEvent submodules not
255 include the package name anymore, as it will be logged
256 by default already.
257 - upgrade to the trick used by common-sense 3.5 to work
258 around extra warning torture/breakage under perl 5.15.x.
259 - log messages by anyevent are now ucfirst, are usually full
260 sentences and do no longer include the package name.
261 - the storable read type would sometimes throw an exception instead
262 of causing EBADMSG (analyzed by Maxime Soulé).
263 - work around a bug in openssl 1.0.1 which enforces a minimum rsa
264 keysize (reported by Paul Howarth).
265 - documented Rocco fabricating statements about AnyEvent and me.
266
267 6.14 Tue Jan 31 20:00:24 CET 2012
268 - AnyEvent::Impl::Tk was broken due to a mysterious "die" inside,
269 probably an editing mistake (reported by Darin McBride).
270
271 6.13 Thu Jan 12 07:27:01 CET 2012
272 - AnyEvent::Util::fork_call checked for POSIX availability in the wrong way,
273 causing it to fail if POSIX wasn't loaded already (analysed by Rock Power).
274 - AnyEvent::Handle::push_read (line => did pass $1 directly, so
275 regex-matching inside the callback would change the parameter, despite
276 $1 being dynamically scoped per-block. this perl bug is now being
277 worked around (testcase by Cindy Wang).
278
279 6.12 Mon Dec 12 13:21:10 CET 2011
280 - $! was clobbered by subsequent calls in tcp_connect, due to
281 the postpone not saving/restoring it (analyzed by Richard Garnish).
282
283 6.11 Tue Nov 22 10:36:05 CET 2011
284 - Tk cannot create windows when tainted unless you set a title, so
285 set a dummy title for AnyEvent's dummy mainwindow. How dummy.
286 - escape any nonprintable/nonascii characters when stringifying
287 backtraces.
288 - log the reason loading an interface module fails to load at level debug.
289
290 6.1 Tue Oct 4 19:44:30 CEST 2011
291 - INCOMPATIBLE CHANGE: the default log level is now "4" (error
292 and above), and some messages inside AnyEvent have been elevated
293 to higher log levels to print by default.
294 - AnyEvent::log, unlike AnyEvent::Log::log, did not always exit on fatal.
295 - work around yet another signal race bug, newly introduced by POE. My
296 how that module sucks.
297 - add google public ipv6 dns as fallback servers, so AE::DNS can finally
298 support ipv6-only operation even with defaults (also use ::1 instead
299 of 127.0.0.1).
300 - give better guidelines on how to chose a logging level.
301 - AnyEvent::Log can now cap (limit) log message priority by context.
302 - AE::log will now save and restore $! and $@, so callers don't have to.
303 - AE::Strict now checks that registered fds are still valid at regular
304 intervals.
305 - skip frequently called 'require' calls if the module appears loaded.
306 - increase signal latency test timer considerably for poor little osx,
307 which can't be bothered with basic posix functionality such as signal
308 delivery.
309 - mention some debugging aids in the tutorial.
310 - new env variable PERL_ANYEVENT_MAX_SIGNAL_LATENCY.
311 - private (%namespace) log contexts in PERL_ANYEVENT_LOG now have the name
312 of the context as title.
313 - AnyEvent::Strict would not restore read-write access to $_ when a callback
314 died - this is a user bug, but since AE::Strict exists to help dbeugging...
315 - detect the extremely evil IO::Async::Loop::AnyEvent module and refuse
316 to work with it - the author is well aware of the issues but refuses
317 to fix them. This extreme step has been taken because the module actively
318 attacks the goals of AnyEvent and makes it harder for module authors to use
319 AnyEvent.
320
321 6.02 Fri Aug 26 20:08:31 CEST 2011
322 - log_to_syslog now accepts facility strings.
323 - log_to_syslog would not allow facilities - "$facility|$level"
324 *literally* means that string. sheesh.
325 - the CPAN parser got confused and thought we provide Sys::Syslog, try
326 to work around.
327 - renamed AnyEvent::Impl::FLTK2 to FLTK.
328
329 6.01 Fri Aug 26 07:04:11 CEST 2011
330 - INCOMPATIBLE CHANGE: AnyEvent::DNS resource records now include
331 the TTL as fourth element - this affects ->resolve, ->request
332 and ->dns_unpack indirectly. Convenience functions such as
333 AnyEvent::DNS::{a,txt,mx,srv} are not affected, neither is
334 AnyEvent::Socket.
335 - INCOMPATIBLE CHANGE: backend authors now should not implement
336 one_event or loop, but instead the AnyEvent::CondVar::_wait and _poll
337 methods.
338 - INCOMPATIBLE CHANGE: verbosity levels have been redefined, see
339 AnyEvent (PERL_ANYEVENT_VERBOSE) and AnyEvent::Log for details.
340 - DEPRECIATION: please change
341 "use AnyEvent::Impl::Perl" to "use AnyEvent::Loop" and
342 "AnyEvent::Impl::Perl::loop" to "AnyEvent::Loop::run"
343 in your programs, for future compatibility.
344 - *blush* idle watcher emulation was borked.
345 - AnyEvent::Strict now checks the AE::xxx functions as well.
346 - querying too long domain names would barf deep inside AnyEvent::DNS - since
347 this is hard to avoid for applications, AnyEvent:.DNS::request now checks for
348 for too long domain names explicitly and passes an undef to the callback
349 instead (reported by James Bromberger).
350 - make AnyEvent::Util more robust against EINTR (analyzed by Andrew Suffield).
351 - new functions: AnyEvent::postpone, AnyEvent::log.
352 - new module: AnyEvent::Log, for simple logging needs.
353 - new env variable, PERL_ANYEVENT_LOG, for fine-grained logging config.
354 - new env variable, PERL_ANYEVENT_DEBUG_SHELL, to automatically start
355 a debug shell.
356 - new env variable, PERL_ANYEVENT_DEBUG_WRAP, to automatically enable
357 watcher instrumentation/wrapping (see AnyEvent::Debug::wrap).
358 - new env variable, PERL_ANYEVENT_HOSTS, to override the place where
359 /etc/hosts can be found.
360 - an empty PERL_ANYEVENT_RESOLV_CONF now selects the default
361 configuration.
362 - env variables can now be specified using AE_xyz instead of
363 PERL_ANYEVENT_xyz, subject to some rules.
364 - lots of new AnyEvent::Debug functionality.
365 - AnyEvent::Strict now checks for modifications of $_ by the callback
366 (by setting it to readonly while executing the callback, if possible).
367 - IO::Async sometimes cancels the wrong timer when you call cancel_timer.
368 Invest extra overhead to work around this peculiar behaviour.
369 - split the pure perl backend into a loop module and an impl
370 module (AnyEvent::Loop and AnyEvent::Impl::Perl).
371 - some read types (regex, netstring, json, storable) could
372 errornously keep an AE::Handle object alive on their own
373 (reported by Mohammad Toossi).
374 - AnyEvent::Socket::resolve_sockaddr and all functions using it now
375 supports /etc/hosts overrides.
376 - add more workarounds around child watcher bugs in IO::Async
377 and (newly introduced) in POE.
378 - use glib child watchers and try to work around its limitations,
379 if possible (based on a patch by Kevin Ryde).
380 - add support for FLTK (fltk2), via AnyEvent::Impl::FLTK2.
381 - many read types in AnyEvent::Handle would malfunction during
382 transitions to TLS because of a "delete $self->{rbuf}".
383 (reported by Gerald Galster).
384 - improve condvar blocking wait performance (probably only noticable
385 with faster backends), also make it easier for event loops that
386 do not support blocking waits.
387 - PERL_ANYEVENT_MODEL now also allows full module names, although
388 the usefulness of this feature is not yet clear.
389 - add tests for all supported event loops, conditional on PERL_ANYEVENT_LOOP_TESTS.
390 - add t/80_ssltext.t.
391 - no longer use AUTOLOAD for the initial loading (for no very good reason).
392 - tcp_server now tries to unlink unix domain sockets when it gets destroyed.
393 - do not run event loop detection multiple times when programs call their cached
394 copies of AnyEvent::detect.
395 - parse_hostport now accepts absolute paths als unix domain sockets.
396 - suppress the idiotic warning inside POE by patching the POE::Kernel
397 run flags instead of asking the user to jump through contortions.
398 - add Task::Weaken as dependency (as recommends only at this time).
399 - replace PF_UNSPEC by 0, for good old minix.
400
401 6.0 (not officially released, but was on CPAN for about 15 minutes :/)
402
403 5.34 Thu May 12 09:59:41 CEST 2011
404 - AE::TLS didn't support both cert_file and key at the same time,
405 due to some copy&paste bug (reported and analyzed by vti).
406 - improve AE::Handle timeout documentation slightly.
407
408 5.33 Mon Feb 7 21:51:41 CET 2011
409 - never appeared on CPAN, for unknown reasons.
410 - AnyEvent::Handle on_connect_error - do not try to destroy a handle
411 object when it's already gone (analyzed by Ryan Bullock).
412 - update idna algorithm(s) to UTS#46 version 6.0.0.
413 - preserve trailing dot in idn_to_ascii - testsuite says to preserve,
414 uts46 says it's an error, and testsuite makes more sense.
415 - remove leading dots in idn_to_ascii - testsuite says to remove,
416 uts46 says it's an error, and testsuite makes more sense.
417 - fixed a bug in idn ignored character processing.
418 - passes the uts46 6.0.0 testsuite except for three cases which
419 apparently are bugs in the testsuite itself.
420 - passes the uts46 6.0.1 testsuite except for eleven cases which
421 apparently are bugs in the testsuite itself.
422
423 5.31 Sun Jan 23 11:49:19 CET 2011
424 - work around a POE bug where POE would first poll for new events and
425 invoke handlers BEFORE handling the message queue.
426 - fix parsing of "domain" statement in resolv.conf (Johannes Plunien).
427 - stop_read now does just that in TLS mode. see its
428 documentation for caveats.
429 - update to common::sense 3.4 warning set (some warnings in perl impose
430 an arbitrary 32kb limit on regex matches, which affects
431 AnyEvent::Handle).
432 - AE::Handle had a broken rbuf_max method and missing wbuf_max method.
433 - new experimental AE::Handle::resettls method.
434
435 5.3 Fri Dec 31 05:48:13 CET 2010
436 - major IOAsync patch by Paul Evans - 0.33 is required, as the major
437 issues with IO::Async have been solved in that version, including
438 the availabilty of a default loop!
439 - implement new wbuf_max parameter for AnyEvent::Handle.
440 - added Cocoa::EventLoop backend by Daisuke Murase.
441 - the default testsuite now honors PERL_ANYEVENT_MODEL (Paul Evans).
442 - clarify handle on_prepare documentation.
443 - set CLOEXEC flag on parent-side fd's in AnyEvent::Util::run_cmd,
444 to avoid leaking them to any child processes.
445 - try to force files into the arch-dependend lib directory, for the
446 half-"turly obsessive compulsive person"(s) that depend on it.
447
448 5.29 Sun Dec 5 10:49:21 CET 2010
449 - convert EV backend to EV 4.00 API (so better upgrade EV too).
450 - AE::Handle: implement dynamic read_size adjustment, add
451 max_read_size parameter, reduce default read_size to 2048.
452 - add some FAQ entries for servers and/or on_eof vs. on_error.
453 - work around OS bugs (cygwin again) when lingering in AE::Handle.
454 - reduce memory usage slightly when connect was used in AE::Handle.
455 - wrok around more segfaults in Tk.
456 - document tls_ctx => undef as valid.
457 - detect WNOHANG value at compiletime.
458 - include some probably linux-only support for building AnyEvent
459 as part of the perl core.
460 - improve compatibility of t/08_idna.t to perl 5.8.
461 - make AnyEvent::Strict truly optional (in case somebody builds a
462 minimal perl...).
463
464 5.28 Wed Oct 13 04:14:23 CEST 2010
465 - due to a glitch, AnyEvent's internal getprotobyname wasn't
466 actually used in tcp_server.
467 - implement AnyEvent::CondVar->new method, which was documented but
468 missing.
469 - support autoloading of read types in unshift_read.
470 - AnyEvent::Strict: do the test for invalid fh arguments differently,
471 resulting in more informative error messages.
472 - new AnyEvent::FAQ manpage.
473 - clarify that you can actually call push_read etc. in on_read.
474 - put keys and certs in /tmp, not into the current directory,
475 and document this in the SECURITY CONSIDERATIONS sections
476 of AnyEvent::TLS.
477 - support ";" as resolv.conf comment character in AnyEvent::DNS, as
478 requested by Ron Isaacson.
479 - document $AnyEvent::DNS::RESOLVER variable.
480 - incorporated major typo patches by Abhijit Menon-Sen.
481 - AnyEvent::Handle now croaks when negative timeouts are passed.
482 - add a shitload of TCP_xxx sockopt name constants to AnyEvent::Util,
483 undocumented, but who knows when they might come in handy.
484
485 5.271 Tue Jun 8 12:05:46 CEST 2010
486 - backport to perl 5.8.x.
487
488 5.27 Sun Jun 6 12:12:05 CEST 2010
489 - postpone differently in AnyEvent::Socket now, as
490 when not, canceling the connection attempt might fail
491 (found by Felix Antonius Wilhelm Ostmann).
492 - explicitly check for non-stream sockets in AE::Handle, too many
493 clueless people fell into the trap of this somehow working.
494 - simplified and reworked the "OTHER MODULES" section.
495 - better/more condvar examples.
496
497 5.261 Wed Apr 28 16:13:36 CEST 2010
498 - AF_INET6 was not properly used from Socket6 during configuration
499 time (found by Andy Grundman).
500
501 5.26 Mon Apr 12 04:49:35 CEST 2010
502 - don't generate spurious readyness notifications when select
503 returns EINTR in the pure perl backend (analysed and testcase
504 by Bas Denissen).
505 - give same examples in SYNOPSIS sections of AnyEvent and AE.
506 - provider faster implementations for AE::time/now/now_update for
507 EV and Perl backends.
508 - provide faster AE::cv and AE::time implementations for all
509 backends.
510 - no longer support lower/mixed-case signal names (this was
511 never documented nor universally supported).
512 - some more memory µ-optimisations, and somewhat less messy
513 code to cope with both AnyEvent and AE APIs.
514
515 5.251 Sat Mar 13 00:58:21 CET 2010
516 - make test failed if it was run before make install on perl <=
517 5.8.8, and older versions of perl make our live very difficult,
518 so write constants.pl during Makefile.PL time.
519
520 5.25 Sat Mar 13 00:23:14 CET 2010
521 - fix a race condition in AnyEvent::Handle that would cause
522 a "bio_read: unsupported method" error (in _another_
523 ssl connection) after user code threw an exception.
524 - added AnyEvent::Handle->destroyed method.
525 - speed up AnyEvent::Socket::format_address and ::format_ipv6.
526 - the AnyEvent::Util::fh_nonblocking function worked only
527 by lucky accident on win32.
528 - smaller and faster AnyEvent::Util::fh_nonblocking.
529 - when the (required!) Time::HiRes module is missing, AnyEvent
530 did not fall back to built-in time properly.
531 - do not load Fcntl at runtime, saving memory and loading time.
532 - precompile a number of constants and use them instead of runtime
533 detection and eval.
534 - free detection code after detection and similar memory optimisations.
535 - Perl backend timer interval best effort drift has been improved
536 (same algorithm as EV).
537 - update unicode idna mapping table.
538
539 5.24 Tue Jan 5 11:39:43 CET 2010
540 - cygwin never reports errors from failed connects. 1.5 just gives
541 you continous readyness and EAGAIN, 1.7 is even more broken and
542 just hangs. work around both issues in a cygwin-specific hack.
543 - improve idle watcher documentation slightly.
544
545 5.23 Sun Dec 20 23:48:00 CET 2009
546 - support IDNs in resolve_sockaddr, and therefore in tcp_connect.
547 - implement punycode_encode/decode, idn_nameprep,
548 idn_to_ascii and idn_to_unicode operations in AnyEvent::Util.
549 - provide $AE::VERSION.
550 - removed traces of "no strict 'refs'".
551
552 5.22 Sat Dec 5 03:51:13 CET 2009
553 - downgrade-or-fail in AnyEvent::Handle::push_write, to
554 diagnose encoding failures earlier and more succinctly.
555 (this works around bugs in perl, throwing away encoding info
556 when passing scalar data to extensions).
557 - add more examples to AnyEvent::Socket manpage.
558 - upgrade internal warning set to the same as common::sense 2.03.
559 - use pack "n/a*" for pre-5.8.9 perl compatibility in AnyEvent::DNS
560 (John Beppu).
561 - AnyEvent::Socket::inet_aton now properly supports ipv6, as documented.
562 - add google public dns servers to fallback server set and make sure
563 we load-balance properly between the three sets. also add all
564 fallback dns servers, not just a random one, to each dns config.
565
566 5.21 Thu Nov 19 02:48:47 CET 2009
567 - fix a problem where socket constants were called with parameters
568 (spotted by David Friedland).
569 - fork_call never use'd POSIX (reported by Daisuke Maki).
570 - improve perl 5.6 compatibility further (but it still won't work
571 unless you rip out everything but the core).
572 - prefer Net::DNS::Resolver over ipconfig, if installed, on win32.
573 uses 10MB of RAM, but doesn't open a console window. *sigh*.
574
575 5.202 Wed Oct 14 22:35:44 CEST 2009
576 - AnyEvent::DNS would unexpectedly clobber $_ under windows
577 (analysed by Matthias Waldorf).
578 - AnyEvent::Handle::run_cmd can now pass the PID of the
579 newly-created process, which is much less useful than it might
580 sound (based on patch by Yann Kerherve).
581
582 5.201 Tue Sep 29 12:09:25 CEST 2009
583 - AnyEvent:Handle::on_starttls/on_stoptls methods were broken
584 (reported by Torsten Foertsch).
585 - common::sense 2.0 could cause tcp_server to throw an exception
586 (analysed by elmex).
587
588 5.2 Mon Sep 14 07:04:49 CEST 2009
589 - INCOMPATIBLE CHANGE: do no longer support register_read_type
590 and register_write_type in AnyEvent::Handle, instead support
591 package names (the facility was mostly abused).
592 - implement "packagename-as-read/write type" support in
593 AnyEvent::Handle.
594 - AnyEvent::Handle: new options "keepalive" and "oobinline".
595 - oobinline set by default to avoid security issues.
596 - the pure-perl event loop backend wrongly detected times() fallback
597 support (spotted by Pavel Boldin).
598
599 5.12 Tue Sep 1 20:26:50 CEST 2009
600 - be more lenient when parsing resolv.conf files, as some people
601 use hashmarks as comment indicator inside directives (reported
602 by Michael S. Fischer).
603 - use same set of warnings as common::sense 2.0.
604 - fix a potential 32 bit overflow issue due to perl having problems
605 with large hex constants in the Perl backend.
606 - do not use the slower 5.6 version fo the enc_name code in AnyEvent::DNS
607 by default.
608 - fix some prototypes.
609 - (optionally) require Async::Interrupt 1.02.
610 - replace opendns fallbacks by verizon/level3 public dns servers.
611
612 5.112 Fri Aug 21 13:59:04 CEST 2009
613 - AnyEvent::Handle could cause hard crashes in Net::SSLeay when
614 using starttls with an illegal/uncreatable tls context.
615 - port to broken win32 perls.
616
617 5.111 Sun Aug 16 18:44:36 CEST 2009
618 - perl errornously requires sockaddr_un structures returned by the
619 kernel to have a specific length. work around this bug by padding
620 sockaddr_un structures (found to fail on OS X, as perl only uses
621 the correct code for GNU/Linux, but likely to fail on other
622 OSes as well).
623
624 5.11 Wed Aug 12 17:49:37 CEST 2009
625 - fixed a bug in run_cmd with input from filename.
626 - create a json coder object when none is given in write direction
627 to reduce dependency on JSON.
628
629 5.1 Tue Aug 11 03:17:32 CEST 2009
630 - tcp_connect now reports the correct error in $!, not an unrelated
631 one (was broken since 4.91).
632 - AE::Handle did not properly clear rtimeout/wtimeout watchers
633 when retrying a tcp_connect.
634 - new functions: AnyEvent::Util::run_cmd and close_all_fds_except.
635 - fixed wrong documentation in AE::Socket.
636
637 5.01 Mon Aug 10 03:16:32 CEST 2009
638 - last release broke EV child watchers :/.
639
640 5.0 Sun Aug 9 17:32:09 CEST 2009
641 - officially introduce and document the AE API.
642 - lowering the timeout at runtime did not have immediate effect
643 in AnyEvent::Handle.
644 - AE::Handle now has separate and independent read and write timeout
645 settings.
646 - probe for Guard module when AnyEvent::Util is loaded, not at
647 runtime (Event doesn't like eval "use" from callbacks, and it avoids
648 exporting the probe function).
649 - Impl::IOAsync's io watchers did not use the filehandle emulation
650 code to work around it's watcher limits.
651 - work around IO::Async's problems with overloaded objects.
652 - try to disable wearnings inside IO::Async, it's just too buggy.
653 - do no longer use Event by default, only EV or Perl, Event is too buggy.
654 - hackishly provide fast AE:: interface when EV or Perl is the backend.
655 - inherit VERSION from AnyEvent::VERSION in most modules.
656
657 4.91 Thu Aug 6 15:42:45 CEST 2009
658 - AE::Handle::starttls could get out of sync when the read buffer
659 already contains some TLS handshake.
660 - AE::Handle did not properly free the TLS session, causing bigger
661 memory leaks in Net::SSLeay than expected (reported by toaster).
662 - AE::Socket::tcp_connect will now artificially delay invoking
663 the callback to avoid returning after invoking the callback.
664 - convert many internal watcher uses to AE API.
665
666 4.9 Sat Aug 1 11:07:01 CEST 2009
667 - Glib silently fails when registering a timeout with
668 a negative value, so avoid that.
669 - call condvar callback immediately when it is set
670 after the condition is already signalled.
671 - check rbuf_max condition only after trying to consume
672 data.
673 - normalise signal numbers to names when using the ae's
674 signal handling, but do not document this (yet).
675 - pure perl signal emulation did not properly set nonblocking
676 mode on the signal pipe on win32 (this is unlikely to have caused
677 any issues).
678 - new module: AnyEvent::Debug.
679 - AnyEvent::Strict now checks that a signal was specified
680 by name (not name or number).
681 - reduce memory footprint in typical cases by ~50kb by
682 only compiling signal code when necessary.
683 - add AnyEvent::Handle->rbuf_max.
684 - grab the AE namespace for future new API, implement stubs for
685 said future/alternative API.
686 - new function: AnyEvent::Socket::format_hostport.
687
688 4.881 Tue Jul 28 12:51:53 CEST 2009
689 - work around a bug in local in pre-5.10 perls, causing
690 AnyEvent::Handle to recurse when it should not
691 (analysed by elmex).
692
693 4.88 Tue Jul 28 04:04:37 CEST 2009
694 - re-bless the handle into a dummy package after calling
695 AnyEvent::Handle::destroy, so the user does not need to check
696 for errors after every push_write etc.
697 - do not attempt to run t/02_signals.t on obviously broken
698 platforms.
699
700 4.87 Sun Jul 26 02:06:16 CEST 2009
701 - do not attempt to linger when there is no longer a valid fh.
702 - enforce tls mode to be either accept or connect, do not simply
703 segfault in Net::SSLeay.
704 - AnyEvent::Handle can now call tcp_connect itself (new parameters
705 connect, on_prepare, on_connect and on_connect_error). Updated
706 tutorial accordingly.
707 - add AnyEvent::Impl::Irssi backend.
708
709 4.86 Mon Jul 20 23:52:29 CEST 2009
710 - since the verbose warning is not enough, explicitly document
711 that versions before 1.33 of Net::SSLeay are not secure.
712 - work around signal handling races in Event and (...) Event::Lib.
713 - try to align signal-race timer to full-second boundaries.
714 - work around Tk not liking negative timeouts.
715 - don't complain of different grades of environmental unfriendlyness
716 in IO::Async.
717
718 4.85 Sat Jul 18 06:16:14 CEST 2009
719 - nail the signal race problem in perl once and for all
720 (see $AnyEvent::MAX_SIGNAL_LATENCY).
721 - take advantage of Async::Interrupt if it is available.
722 - load Time::HiRes and Guard modules on demand only.
723 - add optional/recommended modules section to AnyEvent
724 documentation.
725 - reduce memory usage considerably (and reduce startup penalty)
726 by not using "strict", "warnings" and "overload" modules.
727 - work around buggy windows/openbsd perls and provide EBADMSG
728 and EPROTO ourselves when missing.
729 - improve perl 5.6 compatibility of the core event loop.
730 - made Net::SSLeay version 1.33 a soft requirement.
731
732 4.83 Fri Jul 17 16:56:26 CEST 2009
733 - implement AnyEvent::Socket::getprotobyname.
734 - AnyEvent::CondVar's will now detect recursive blocking
735 waits and will croak, as too many people fall into
736 this trap.
737 - AnyEvent::Handle will now call ->destroy on itself after
738 executing the on_error callback, instead of doing some
739 half-baked internal shutdown, for fatal errors.
740 - clarify on_eof behaviour w.r.t. the read queue and
741 on_read callbacks.
742 - ignore some possible spurious wake-ups in tcp_connect.
743
744 4.82 Sat Jul 11 00:34:55 CEST 2009
745 - POE and Event backends didn't accept some callable objects as
746 callbacks.
747 - use Config module instead of POSIX module to detect signal names
748 in AnyEvent::Strict and AnyEvent::Impl::EventLib, as the POSIX
749 module doesn't even have all POSIX signals :/.
750 - use more workarounds around the many refcnt/corruption bugs in
751 Event::Lib.
752 - work around a race condition in perl's select, causing t/03_child.t
753 to rarely fail.
754
755 4.81 Thu Jul 9 10:30:30 CEST 2009
756 - AnyEvent::Handle didn't properly diagnose write errors
757 (it expected -1 from syswrite, how lame... :).
758 - support file descriptors in addition to file handles
759 in AnyEvent->io.
760 - new env variables: PERL_ANYEVENT_RESOLV_CONF,
761 PERL_ANYEVENT_MAX_OUTSTANDING_DNS, PERL_ANYEVENT_CA_FILE
762 and PERL_ANYEVENT_CA_PATH.
763 - provide a sensible synopsis section for AnyEvent::TLS.
764 - add a "supported backends" section to the manpage.
765 - added simple io watcher test to testsuite, using a
766 portable_socketpair.
767 - tried to improve the stability of the Event::Lib backend,
768 YMMV.
769
770 4.8 Mon Jul 6 23:45:16 CEST 2009
771 - AnyEvent::DNS did not properly follow CNAME records with
772 uppercase targets.
773 - AnyEvent::DNS would errornously return AAAA records
774 with v4 mapped addresses (a faulty record) as ipv4 addresses,
775 causing AnyEvent::Socket to throw an exception.
776 - added new module AnyEvent::TLS for easier SSL/TLS context
777 creation, with many options including hostname verification,
778 secure default configuration, lots of documentation and,
779 predefined diffie-hellman keys for perfect forward security
780 and much more. get it while it's still fresh!
781 - use AnyEvent::TLS in AnyEvent::Handle for context management.
782 - load AnyEvent::Handle only on demand in AnyEvent::DNS,
783 so AnyEvent::Socket users have smaller memory footprint
784 in the common case.
785 - add AnyEvent::Handle->push_shutdown method.
786 - add an additional error message parameter to AnyEvent::Handle's
787 on_error callback (for TLS, $! is still available).
788 - add AnyEvent::Handle on_starttls/on_stoptls callbacks.
789 - make AnyEvent::Handle more robust against early connection
790 failures (during new), and return C<undef> in that case
791 from the constructor.
792 - AnyEvent::Handle will now try to load only JSON::XS first,
793 then fall back to JSON.
794 - format_ipv4/format_ipv6 are now exported by default, for symmetry,
795 and because it was documented that way.
796
797 4.451 Fri Jul 3 00:28:58 CEST 2009
798 - do not clear rbuf when shutting down an AnyEvent::Handle
799 object - doing so breaks AnyEvent::HTTP.
800
801 4.45 Mon Jun 29 22:59:26 CEST 2009
802 - a write error could cause AnyEvent::Handle to create
803 an I/O watcher with an undefined $fh.
804 - special-case mapped ipv4 addresses in both
805 AnyEvent::Socket::format_address and parse_address,
806 to treat them just like ipv4 addresses.
807 - updated and overhauled the AnyEvent::Intro doc.
808 - implement AnyEvent::Socket::format_ipv4/ipv6.
809 - slightly speed up portable_pipe/socketpair.
810 - expand condvar begin/end documentation.
811
812 4.42 Fri Jun 26 08:32:18 CEST 2009
813 - preliminary and neccesarily incomplete support for IO::Async.
814 - reset SIGCHLD to DEFAULT when AnyEvent is loaded, in
815 case it was set to IGNORE, to ensure we can catch child statuses
816 even when the calling env acts stupidly.
817 - updated benchmarks with IO::Async, which performs very well.
818
819 4.412 Wed Jun 24 01:35:57 CEST 2009
820 - support an "untaint" attribute for AnyEvent::DNS
821 and set it on the default resolver.
822 - implement and document AnyEvent::Impl::Perl::loop.
823 - remove all anyevent-env variables from %ENV
824 when running in tainted mode.
825 - mention and extend the IO::Lambda benchmark.
826
827 4.411 Sun Jun 7 18:48:02 CEST 2009
828 - do not try to use F_SETFD on windows.
829
830 4.41 Thu May 14 06:40:11 CEST 2009
831 - work around issues in older perls (5.8.0?) when
832 a signal handler is deleted from the %SIG hash.
833 - use POSIX::_exit in child test, to avoid running
834 destructors.
835 - speed up CHLD handling by relying on SIGCHLD being
836 synchronously delivered, even when we roll our own
837 implementation.
838 - AnyEvent::DNS: add the "dname" resource record name
839 for cosmetic reasons.
840
841 4.4 Sun Apr 26 20:12:33 CEST 2009
842 - implemented idle watchers, where applicable.
843 - AnyEvent->time died when Event backend was in use.
844 - fix a memleak in the Tk backend.
845 - sped up Tk timer handling.
846 - clip negative "after" values to 0 in AnyEvent::Impl::Event
847 to avoid spamming out warning messages.
848 - fix Qt timers without interval.
849 - avoid Qt zero-timeout special case, in old libqt's.
850
851 4.352 Mon Apr 20 16:31:11 CEST 2009
852 - fix AnyEvent::Strict error messages for child watchers.
853 - fix/update Makefile.PL "recommends" versions.
854 - add AnyEvent->now_update.
855
856 4.351 Sat Apr 11 07:56:14 CEST 2009
857 - actually make the signal pipe work under win32.
858 - localise $! in signal handler to avoid changing $!, although
859 perl itself does not seem to save/restore errno either.
860 - set the cloexec flag on the signal pipe (normally set by perl too).
861
862 4.35 Fri Mar 27 11:48:20 CET 2009
863 - event models relying on AnyEvent's signal watcher emulation did
864 invoke the callback asynchronously, contradicting documentation
865 and causing signals to get lost (this includes AnyEvent's own
866 event loop). AnyEvent now uses the standard pipe trick to make
867 callback execution synchronous to the event loop.
868 - AnyEvent::Handle didn't free TLS context data on DESTROY
869 (patch by Pavel Shaydo).
870 - work around the ever-incompatibly-changing API of MakeMaker.
871 - document that changing global variables without restoring them
872 is a bad idea in Perl (noted by Adam Rosenstein).
873 - AnyEvent::Strict now barfs if ->io is passed a file.
874
875 4.34 Thu Feb 12 18:32:45 CET 2009
876 - separately buffer TLS read data, as otherwise the read queue could
877 deadlock as receiving data is not expected while draining the
878 read queue (which cna only happen with TLS).
879 - raise EBADMSG error on JSON decoding errors.
880 - fix some minor manpage bugs (reported by Maximilian Gaß).
881 - speed up select bitmask parsing quite a bit in the pure perl
882 backend.
883 - use CORE::select instead of just select in AnyEvent::Impl::Perl.
884
885 4.331 Tue Jan 6 21:07:25 CET 2009
886 - socketpair fails on many vista machines because vista has
887 completely broken accept/getpeername and getsockname functions,
888 so we provide our own socketpair emulation that kind of works
889 (AnyEvent::Util::portable_pipe).
890 - new function: AnyEvent::Util::portable_socketpair.
891 - take advantage of the Guard module if it exists.
892
893 4.33 Fri Nov 21 02:35:40 CET 2008
894 - AnyEvent::Strict did errornously flag a fileno of 0 as
895 illegal.
896 - reduce memory usage and slightly speed up the pure perl backend
897 by only storing the file descriptor, not the file handle.
898 - add missing autocork method to AnyEvent::Handle (reported
899 by Adam Rosenstein).
900 - AnyEvent::DNS->resolve errornously documented an $rcode result
901 argument, but there isn't (spotted by Henrik Krohns).
902 - the naked truth about Tk - it's basically unmaintained.
903
904 4.32 Mon Nov 3 22:46:32 CET 2008
905 - fix AnyEvent::Socket::resolve_sockaddr to properly support
906 unix sockets again and choose a proper default for
907 the socket type.
908 - fix call to resolve_sockaddr in tcp_connect.
909
910 4.31 Thu Oct 30 04:41:48 CET 2008
911 - implemented AnyEvent::Handle->destroy method.
912 - hint about unexpected effects in TLS mode.
913 - speed up AnyEvent::Impl::Perl by using more arrays
914 instead of hashes without the slightest loss of readability :)
915 - work around a perl argument refcounting bug.
916 - with some perl patches applied, AnyEvent now seems leak-free.
917
918 4.3 Fri Oct 3 09:18:43 CEST 2008
919 - AnyEvent will now install a no-op signal handler for SIGPIPE,
920 unless one has been installed already.
921 - warn about not loaidng AnyEvent::Impl::POE early enough
922 (patch by Adam Rosenstein).
923 - fixed a great number of bugs and corner cases in AnyEvent::Handle.
924 - fix a bug in where in SSL connect mode, the
925 client would first wait for some data by the server and
926 otherwise hang. (reported and analysed in an absolutely
927 exemplary manner by Adam Rosenstein).
928 - fix a bug in where SSL EOF would not be treated
929 as stream EOF, putting the connection into a hung state.
930 (reported and analysed in an absolutely exemplary manner by Adam
931 Rosenstein).
932 - fix a potential segfault when the TLS context
933 would go missing in a read callback (e.g. due to stoptls) -
934 Net::SSLeay of course makes no type checking whatsoever.
935 - AnyEvent::Handle will not stop the read watcher in TLS mode.
936 - AnyEvent::Handle->stoptls will send a TLS close notify
937 instead of simply destroying the stream now.
938 - fix a bug where large blocks of data written in TLS mode
939 would not be sent unless triggered by receives.
940 - on_drain will now take the tls write buffer into account.
941 - SSL operations have been streamlined a bit and should be faster
942 now, more to come, though.
943 - added a FAQ section, will document nontrivial issues in there.
944
945 4.234 Mon Sep 29 04:08:13 CEST 2008
946 - fix child watcher documentation: only child exits will be reported,
947 no trace events.
948 - mention SIGPIPE in AnyEvent::Handle.
949 - perl backend will now use POSIX::times as monotonic clock
950 when available and the monotonic clock option is not.
951 - run even without Time::HiRes in some cases now, but do not
952 provide sub-second accuracy (all *supported* perl versions have
953 Time::HiRes).
954
955 4.233 Fri Aug 22 01:48:05 CEST 2008
956 - fix a bug in the testsuite, causing a hang.
957 - clarified lots of AnyEvent::Handle constructor arguments.
958
959 4.232 Thu Aug 21 20:44:25 CEST 2008
960 - fix a bug in AnyEvent::Handle that could cause two fatal error
961 callbacks on EOF, which caused AnyEvent::HTTP to signal success twice
962 (testcase by Leon Brocard).
963 - clarified on_eof/on_error documentation a bit.
964
965 4.231 Tue Jul 29 13:12:15 CEST 2008
966 - remove some debugging code left in AnyEvent::Util::fork_call
967 (and no, it's impossible to implement with the broken windows
968 perls without resource leaks or worse).
969
970 4.23 Tue Jul 29 12:19:59 CEST 2008
971 - document the first parameter passed to condvar callbacks to be
972 the callback.
973 - add AnyEvent::Socket::{ntoa,aton} aliases.
974 - optimize the AE::Handle->push_read (line) for the default
975 eol marker.
976 - optimize push_read (packstring|storable) for small packets.
977 - invoke on_error callback when no on_eof callback is set.
978 - fix a bug in push_read (storable) of unknown impact.
979
980 4.22 Sun Jul 20 16:34:13 CEST 2008
981 - new function AnyEvent::Socket::parse_hostport.
982 - as the bulkheads at microsoft can't even get getprotobyname reliably
983 working on their shitty fucking broken os we need to hardcode
984 some common protocol numbers in AnyEvent::Socket. How can
985 people even bother with such a pile of shit as windows.
986
987 4.21 Thu Jul 17 14:40:05 CEST 2008
988 - INCOMPATIBLE CHANGE: AnyEvent::Util::fork_call now has a prototype
989 which is incompatible to previous invocation syntax.
990 - work around more windows perl bugs in fork_call. I cannot imagine
991 why anybody would bother implementing fork in such an obviously
992 unusable way.
993 - avoid calling $do_retry if the request has already finished in
994 AnyEvent::DNS, thus avoiding an exception (reported by Anatoly
995 K. Sharifulin).
996
997 4.2 Sat Jul 12 22:42:11 CEST 2008
998 - support an "interval" argument for repeating timers.
999 - fix a bug in the pure perl event loop that caused timers to
1000 be delayed under some circumstances.
1001 - implement optional argument checking via PERL_ANYEVENT_STRICT.
1002 - not importing anything from Event causes it to malfunction
1003 and not use Time::HiRes, work around this bug.
1004 - more bug workarounds against the endlessly broken Tk module,
1005 seems to work now on 32 bit machines, but less so on 64 bit.
1006 - minor optimisations applied to most backends.
1007
1008 4.161 Fri Jul 4 14:24:48 CEST 2008
1009 - fixed the t/handle/02_write.t, it read after write, with the
1010 assumption that all data could be written before read
1011 (reported by Jim Razmus).
1012
1013 4.160 Thu Jul 3 04:02:21 CEST 2008
1014 - re-registering signal watchers after unregistering one did
1015 not work if a backend falls back on AnyEvent's default
1016 implementation (also affected child watchers).
1017 - new AnyEvent::Handle options: autocork and no_delay.
1018
1019 4.152 Sun Jun 22 14:15:44 CEST 2008
1020 - allow for 32-bit perls that implement shifts differently
1021 on different architectures in parse_ipv4 (reported and
1022 analysed by Keiichi DAIBA).
1023
1024 4.151 Fri Jun 6 17:34:24 CEST 2008
1025 - make sure specifying _only_ on_read and never pushing reads
1026 works.
1027
1028 4.15 Fri Jun 6 13:00:46 CEST 2008
1029 - the pure perl backend would keep some watchers alive when more than
1030 one watcher was registered for the same fd.
1031 - new "packstring" and "storable" read and write types
1032 for AnyEvent::Handle.
1033 - allow on_eof handler to be called after on_error with EPIPE returns.
1034 - do not immediately call on_read callback in handle constructor.
1035
1036 4.14 Thu Jun 5 20:29:31 CEST 2008
1037 - Fixed a bug in DNS SRV priority sorting.
1038 - AnyEvent::Util::guard now reports runtime errors while
1039 executing the guard block as warnings.
1040 - handle 0-byte-reads just before EOF correctly in AnyEvent::Handle.
1041
1042 4.13 Thu Jun 5 00:47:59 CEST 2008
1043 - AnyEvent::DNS only followed cname chains with length 2,
1044 contrary to documentation. bumped it up to 10, thanks to
1045 microsoft, the current broken-dns-config-king.
1046 - AnyEvent::DNS didn't check for socket return status
1047 but instead relied on perl not creating filehandles in that
1048 case - too bad it gives you a bogus file handle (reported
1049 and analysed by Vladimir Timofeev).
1050 - fix queue management logic in AnyEvent::Handle:
1051 when on_read was registered, the queue was empty
1052 and no progress could be made AnyEvent::Handle would
1053 enter an endless loop.
1054 - correctly start reading again when the handle became
1055 busy again after idling.
1056 - correctly treat tls shut-downs as EOF condition
1057 (for the time being, would be nice to have a callback for that).
1058 - correctly call eof callback under all conditions
1059 (hopefully).
1060 - the timeout callback did not expect that $self can go
1061 away any time.
1062
1063 4.12 Tue Jun 3 10:58:04 CEST 2008
1064 - include AnyEvent::Intro, a tutorial for anyevent,
1065 anyevent::socket and anyevent::handle.
1066 - allow more options in on_error.
1067
1068 4.11 Fri May 30 23:42:25 CEST 2008
1069 - INCOMPATIBLE CHANGE: replace ptr by real PTR lookup, provide
1070 reverse_lookup and reverse_verify to replace it, support
1071 v4mapped and v4compat addresses.
1072 - provide more documentation for the resolver class.
1073 - really replace longest run of :0: by :: in format_address,
1074 also properly convert :: and ::1 again.
1075 - support NAPTR record name and decode it.
1076 - implement random weight sampling for SRV records, as per
1077 rfc 2782.
1078 - correctly abort on srv-record targets of ".".
1079 - added AnyEvent::DNS::wait_for_slot.
1080 - in the unlikely event of a virtual circuit connection
1081 being invalidated by a delayed udp reply, AnyEvent::DNS
1082 could die. this has been fixed.
1083 - plug a probable memleak in the DNS vc code.
1084 - use configured timeout also for VC request phase.
1085 - implement timeout and max_outstanding methods
1086 for resolver class.
1087 - update version numbers in all modules.
1088
1089 4.1 Thu May 29 05:45:40 CEST 2008
1090 - INCOMPATIBLE CHANGE: renamed xxx_ip functions to xxx_address
1091 which mirrors their purpose better (old names still available
1092 for a while). Also moved AnyEvent::DNS::addr to
1093 AnyEvent::Socket::resolve_sockaddr.
1094 - implement AnyEvent->time and AnyEvent->now.
1095 - fix IPv6 support in pack_sockaddr.
1096 - officially un-experimentalise all newly introduced code.
1097 - support unix domain sockets everywhere by specifying
1098 a host of "unix/" and the pathname as service.
1099 - implement an activity timeout in AnyEvent::Handle.
1100 - added a regex read type to AnyEvent::Handle.
1101 - added a json read/write type to AnyEvent::Handle.
1102 - always croak in on_error in AnyEvent::Handle.
1103 - document how to subclass AnyEvent::Handle.
1104 - implement AnyEvent::Util::fork_call.
1105 - add support for IPv6 nameservers and nameserver statements.
1106 - work around _yet_ _another_ windows perl bug in where empty
1107 select masks cause errors in the pure perl backend (this
1108 workaround cannot be implemented for the other event loops,
1109 but EV also works around this).
1110 - supply AnyEvent::Socket::address_type and make good use of it.
1111 - clarify and fix the Handle documentation w.r.t $self vs. $handle
1112 vs. callback arguments.
1113 - add some recommends to the META.yml.
1114
1115 4.05 Mon May 26 19:44:06 CEST 2008
1116 - some platforms ignore the rfc and prepend an extra sa_len member to
1117 the sockaddr structure, cater for those. also use sockaddr_family,
1118 didn't know it was there, but comes in handy for the workaround.
1119 - undo the import WIN32 hack.
1120
1121 4.04 Mon May 26 08:03:31 CEST 2008
1122 - try to work around yet another windows bug: failed connects are reported
1123 as if it were out-of-band data. windows users: you suck. Only
1124 EV, Glib, Event and the pure perl backend can handle this condition.
1125 - optimize write algorithm in AnyEvent::Handle.
1126 - properly parse PERL_ANYEVENT_PROTOCOLS and use the info correctly.
1127 - tcp_connect now properly iterates through all targets.
1128 - check wether IPv6 sockets can actually be created, otherwise, disable ipv6
1129 support entirely. also hardcode some AF_INET6 constants for the
1130 sake of perl 5.8.
1131 - vastly improved nameserver/suffix detection algorithm for windows
1132 (still a hack...)
1133 - try to use a fallback dns server if no dns servers could be found.
1134 - splatter around some local $SIG{__DIE__}'s to work around bad
1135 die handlers.
1136
1137 4.03 Sun May 25 05:05:57 CEST 2008
1138 - work around perl bugs on windows, where perl returns undocumented
1139 error codes for sysread, syswrite etc.
1140 - fix AnyEvent::Handle::unshift_read to really unshift and not push.
1141 - AnyEvent::Handle could sometimes stumble over 0-byte writes.
1142 - fix unpack_sockaddr, now works for AF_INET6 sockaddr's.
1143 - updated documentation.
1144 - lots of workarounds for perl 5.8 quirks.
1145 - implement a more flexible type system for both reads and writes
1146 in AnyEvent::Handle.
1147 - support IPv6 in tcp_server.
1148 - pass local hostname and port to the prepare callback in tcp_server.
1149 - make types extendable in AnyEvent::Handle.
1150 - croak, not die, in AnyEvent::Handle;
1151
1152 4.0 Sat May 24 19:58:08 CEST 2008
1153 - added (experimental) AnyEvent::DNS module.
1154 - added (experimental) AnyEvent::Socket module.
1155 - added (experiemntal) support for TLS/SSL in AnyEvent::Handle.
1156 - make condvar->end call ->send as documented.
1157 - make condvar constructor support the documented cb => argument.
1158 - overload condvars so you can use them instead of code references.
1159 - support more of the POSIX forms instead of a pure dotted quad for
1160 verbatim ip addresses.
1161 - AnyEvent::Socket::inet_aton will use AnyEvent::DNS now and not fork anymore.
1162 - try very hard not to freeze on broken windows perls.
1163 - make on_eof optional in AnyEvent::Handle.
1164 - support service names in tcp_connect and tcp_server.
1165 - work around netbsd/openbsd bug causing unix sockets not to
1166 become writable until empty.
1167 - only make a dummy read for error detection when getpeername
1168 returns ENOTCONN.
1169 - defend AnyEvent::Util::inet_aton against Coro::LWP's brutal
1170 override.
1171 - new AnyEvent::Guard::cancel.
1172
1173 3.5 Sat May 17 23:17:14 CEST 2008
1174 - gracefully deal with systems not defining CLOCK_MONOTONIC
1175 (as opposed to not supporting the clock itself, which worked in 3.41).
1176 - added AnyEvent::Util::tcp_connect and AnyEvent::Util::tcp_server
1177 helpers, to replace the ill-designed AnyEvent::Socket class.
1178 - added some examples (eg/connect, eg/handle).
1179 - added AnyEvent::Util::guard.
1180 - AnyEvent::Util now exports most of its functions.
1181 - fixed a bug in *_read_chunk, where the callback was called with $self as
1182 first and second argument.
1183 - simplified and fixed regex handling in AnyEvent::Handle::*_read_line.
1184 - fix manpage condvars examples to use new syntax.
1185 - considerably improve inet_aton with EV::ADNS backend to support
1186 cname chains (common dns configuration bug, e.g. with www.google.de)
1187 and aaaa records if no a record could be found.
1188
1189 3.41 Sun May 11 19:53:13 CEST 2008
1190 - work around a bug in perl 5.8.8's local. Reported by Yi Ma Mao.
1191
1192 3.4 Sun May 11 00:29:25 CEST 2008
1193 - complete re-implementation of condvars with a hopefully much more useful
1194 API (backward compatible functions exist but are undocumented).
1195 - AnyEvent::Handle fully rewritten with a hopefully more powerful API
1196 (still subject to change).
1197 - take advantage of CLOCK_MONOTONIC, if available, in AnyEvent::Impl::Perl.
1198 - provide AnyEvent::post_detect and @AnyEvent::post_detect, which
1199 allows module authors to avoid forcing event loop detection.
1200 (used by Coro::AIO, Coro::BDB, Coro::AnyEvent for example).
1201 - remove coro backends: Coro now provides generic support for AnyEvent
1202 via Coro::AnyEvent.
1203
1204 3.3 Mon Apr 28 09:51:06 CEST 2008
1205 - added AnyEvent::Handle, AnyEvent::Socket and AnyEvent::Util modules.
1206 - fix a bug in the pure perl backend that kept watchers alive
1207 when multiple watchers were registered for the same fh/poll combo.
1208 - add a benchmark section showing AnyEvent overhead and comparing
1209 the different event loops with each other.
1210 - prefer pure perl over tk when autoprobing, as it's about as fast,
1211 but doesn't crash with many watchers.
1212 - declare Qt support non-experimental.
1213 - clarify the confusing section about the file descriptor being
1214 kept alive.
1215 - document the race between loading of an event module and
1216 child processes exiting.
1217 - support POE as "backend" (with some caveats, POE is not generic
1218 enough, and darn slow).
1219 - support Wx and Prima through POE.
1220 - optimise perl backend to use 20% less memory and take advantage
1221 of typical timeout behaviour. It can now compete with
1222 select/poll-based C event loops in most cases (it is usually
1223 faster than Event and Glib :).
1224 - roughly cut EV memory use in half and increase its speed by 30%,
1225 by removing undocumented functionality. Did something similar
1226 to other event loops where possible.
1227
1228 3.2 Thu Apr 24 10:10:40 CEST 2008
1229 - do not die when anyevent watchers are destroyed while running callbacks
1230 in the pureperl backend (could only happen when two watchers are registered
1231 for the same fh/poll combo).
1232 - support autoloading for child watchers, was broken in all versions.
1233 - implement PERL_ANYEVENT_MODEL env variable.
1234 - (experimental) implement interface to the (very crashy/buggy) Event::Lib module.
1235 - (experimental) implement interface to the Qt module (cannot be autoprobed).
1236 - this release is IO:AnyEvent-proof.
1237
1238 3.12 Tue Apr 22 07:11:46 CEST 2008
1239 - reinstate AUTHOR section that got lost somehow.
1240 - do not hang in the testsuite with badly broken perls (activestate,
1241 strawberry...), but instead diagnose the problem and continue.
1242 - use INT instead of CHLD in an attempt to support broken windows perls
1243 better (this decreases the test quality, unfortunately).
1244 - do not send the signal to the process group
1245 (no problem for CHLD, bad for INT :).
1246
1247 3.11 Sat Apr 19 06:57:31 CEST 2008
1248 - major documentation rework.
1249 - document the fact that child watchers only watch for zombies.
1250 - fix the child watcher example.
1251
1252 3.1 Wed Apr 16 17:09:01 CEST 2008
1253 - work around recurring bugs in Tk by dup'ing filehandles, the
1254 only method with good success chances on Tk (the bugs apparently
1255 don't get fixed anytime soon).
1256 - lift the restriction of only one watcher per fh direction
1257 (as the Tk bug workaround also lifts it and only Tk imposed
1258 such strong limits).
1259 - changed probe order to prefer coro adaptors.
1260 - explain why recursion into the event loop is not supported
1261 unless the backend supports it (only Coro::EV does without
1262 any restrictions...).
1263 - add simple manpages for all backend modules.
1264
1265 3.0 Mon Apr 7 21:30:23 CEST 2008
1266 - Coro::Signal changed semantics, roll our own, also cleaning
1267 up the Coro implementation in general.
1268 - rename Coro backend to CoroEvent.
1269 - add some decision helping paragraph to the manpage that should
1270 help people to decide whether AnyEvent is the right thing for them.
1271
1272 2.9 Mon Jan 28 13:31:54 CET 2008
1273 - update for EV 3.0 API changes.
1274
1275 2.8 Sun Nov 25 15:06:03 CET 2007
1276 - waitpid can validly return 0. accept this fact of life
1277 instead of reporting it to any watchers.
1278
1279 2.7 Fri Nov 23 11:41:14 CET 2007
1280 - force use of AnyEvent::Impl::Perl in testsuite, there is
1281 too much breakage outside AnyEvent.
1282 - deliver signals synchronously in AnyEvent::Impl::Perl
1283 or any other event loop that relies on AnyEvents child
1284 watcher emulation. *Could* help with hanging testsuite
1285 (except when Event or EV are installed).
1286
1287 2.6 Fri Nov 9 20:36:35 CET 2007
1288 - fix bug in testsuite.
1289 - move EV adaptor modules to AnyEvent.
1290 - add Coro+EV adaptor module.
1291
1292 2.55 Tue Nov 6 17:41:32 CET 2007
1293 - add EV to the list of supported event models.
1294 - do not auto-reset pid watchers, pass pid and status to them.
1295 - allow a pid of zero to watch for all children in child watchers.
1296
1297 2.54 Wed Jul 18 17:36:23 CEST 2007
1298 - work around a perl bug that results in
1299 BEGIN not safe after errors--compilation aborted without
1300 any discernible reason or error message by once
1301 again not use'ing strict.
1302
1303 2.53 Sun Jul 8 10:51:53 CEST 2007
1304 - make Glib interface work again (spotted by elmex).
1305
1306 2.52 Wed Mar 7 18:36:16 CET 2007
1307 - child watchers stopped working permanently when
1308 all current child watchers were destroyed.
1309
1310 2.51 Mon Dec 11 21:33:24 CET 2006
1311 - work around bugs in perl where eval "require Module" returns
1312 true even if the module couldn't be loaded :/.
1313
1314 2.5 Mon Dec 11 02:15:28 CET 2006
1315 - avoid $AUTOLOAD because many perls corrupt it.
1316 - AnyEvent::detect forces autodetection.
1317 - implement signal watchers (experimental)
1318 (Tk does not support async signals (<= 804.027 at least),
1319 so they might get delayed indefinitely on Tk).
1320 - implement child watchers (experimental).
1321 - moved default condvar implementation into base module, simplifying
1322 most implementation modules.
1323
1324 2.1 Fri Nov 24 15:50:48 CET 2006
1325 - better docs.
1326 - simple testsuite.
1327 - added AnyEvent->one_event method for special purposes.
1328
1329 2.0 Wed Nov 1 02:21:30 CET 2006
1330 - INCOMPATIBLE CHANGE: poll can now either be "r" or "w" but not both.
1331 - INCOMPATIBLE CHANGE: io watcher callbacks have no arguments anymore.
1332 - EXPERIMENTAL:
1333 new pure-perl model added, AnyEvent now always finds a suitable
1334 event model.
1335 - improved documentation.
1336 - reduced cpu overhead.
1337
1338 1.02 Fri Jan 13 14:15:40 CET 2006
1339 - add COPYING to clarify license.
1340
1341 1.01 Sun Jan 8 05:48:09 CET 2006
1342 - correctly cancel Event timers, avoid memory leaks.
1343
1344 1.0 Sun Jan 8 05:40:58 CET 2006
1345 - allow third-party interfaces to be added at runtime.
1346 - fix version number.
1347
1348 0.9 Fri Dec 30 02:25:41 CET 2005
1349 - fixed errors in the documentation.
1350 - going back to using $VERSION as "load check",
1351 as suggested by Jörn Reder. Let's see why I stopped using
1352 this method.
1353 - add 'hup' to the list of events to receive for glib
1354 and generate 'rw' events for it, as some versions of
1355 glib seem to require this, while this breaks other
1356 (older) versions of glib.
1357 - implemented PERL_ANYEVENT_VERBOSE env variable.
1358
1359 0.3 Sun Dec 4 10:43:33 CET 2005
1360 - no changes to the code.
1361 - fixed documentation.
1362 - clarified some corner cases.
1363 - vastly improved documentation with a long example.
1364
1365 0.2 Thu Dec 1 22:18:49 CET 2005
1366 - check whether a valid method was called, to avoid endless loops.
1367 - move condvars into their own class (cleaner, safer).
1368 - don't die when some modules are not available.
1369 - Tk is too broken to allow for words. Will work once Tk works.
1370
1371 0.1 Thu Dec 1 19:44:57 CET 2005
1372 - spiced up and cleaned up for release.
1373
1374 0.01 Tue Sep 21 18:12:09 CEST 2004
1375 - original version; copied from Convert::Scalar.
1376