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

Comparing AnyEvent/Changes (file contents):
Revision 1.301 by root, Mon Jul 27 22:49:22 2009 UTC vs.
Revision 1.460 by root, Fri Aug 12 18:41:25 2011 UTC

1Revision history for Perl extension AnyEvent. 1Revision history for Perl extension AnyEvent.
2 2
3TODO: document TCP_*** constants
4TODO: catch AE::xxx for anyevent strict and anyevent debug
5TODO: anyevent::trace? or debug tracing all watchers?
6TODO: anyevent strict checking for $_ changes?
7
8TODO: redesign Impl
9TODO: make ae::run mroe robust against eintr, Andrew Suffield
10TODO: IO::Async sometimes cancels the wrong timer when you call cancel_timer -wait for IO::Async release.
11 - INCOMPATIBLE CHANGE: AnyEvent::DNS resource records now include
12 the TTL as fourth element - this affects ->resolve directly,
13 and ->request and ->dns_unpack indirectly. Convenience functions
14 such as AnyEvent::DNS::{a,txt,mx,srv} are not affected.
15 - INCOMPATIBLE CHANGE: backend authors now should not implement
16 one_event or loop, but instead the AnyEvent::CondVar::_wait and _poll
17 methods.
18 - DEPRECIATION: please change
19 "use AnyEvent::Impl::Perl" to "use AnyEvent::Loop" and
20 "AnyEvent::Impl::Perl::loop" to "AnyEvent::Loop::run"
21 in your programs, for future compatibility.
22 - *blush* idle watcher emulation was borked.
23 - querying too long domain names would barf deep inside AnyEvent::DNS - since
24 this is hard to avoid for applications, AnyEvent:.DNS::request now checks for
25 for too long domain names explicitly and passes an undef to the callback
26 instead (reported by James Bromberger).
27 - new function: AnyEvent::postpone.
28 - IO::Async sometimes cancels the wrong timer when you call cancel_timer.
29 Invest extra overhead to work around this peculiar behaviour.
30 - split the pure perl backend into a loop module and an impl
31 module (AnyEvent::Loop and AnyEvent::Impl::Perl).
32 - some read types (regex, netstring, json, storable) could
33 errornously keep an AE::Handle object alive on their own
34 (reported by Mohammad Toossi).
35 - add more workarounds around child watcher bugs in IO::Async
36 and (newly introduced) in POE.
37 - use glib child watchers and try to work around its limitations,
38 if possible (based on a patch by Kevin Ryde).
39 - add support for FLTK (fltk2), via AnyEvent::Impl::FLTK2.
40 - many read types in AnyEvent::Handle would malfunction during
41 transitions to TLS because of a "delete $self->{rbuf}".
42 (reported by Gerald Galster).
43 - resolve AnyEvent::Util::guard at first-use-time, not at load time,
44 for further speed and memory savings when not used.
45 - improve condvar blocking wait performance (probably only noticable
46 with faster backends), also make it easier for event loops that
47 do not support blocking waits.
48 - PERL_ANYEVENT_MODEL now also allows full module names, although
49 the usefulness of this feature is not yet clear.
50 - add tests for all supported event loops, conditional on PERL_ANYEVENT_LOOP_TESTS.
51 - add t/80_ssltext.t.
52
535.34 Thu May 12 09:59:41 CEST 2011
54 - AE::TLS didn't support both cert_file and key at the same time,
55 due to some copy&paste bug (reported and analyzed by vti).
56 - improve AE::Handle timeout documentation slightly.
57
585.33 Mon Feb 7 21:51:41 CET 2011
59 - never appeared on CPAN, for unknown reasons.
60 - AnyEvent::Handle on_connect_error - do not try to destroy a handle
61 object when it's already gone (analyzed by Ryan Bullock).
62 - update idna algorithm(s) to UTS#46 version 6.0.0.
63 - preserve trailing dot in idn_to_ascii - testsuite says to preserve,
64 uts46 says it's an error, and testsuite makes more sense.
65 - remove leading dots in idn_to_ascii - testsuite says to remove,
66 uts46 says it's an error, and testsuite makes more sense.
67 - fixed a bug in idn ignored character processing.
68 - passes the uts46 6.0.0 testsuite except for three cases which
69 apparently are bugs in the testsuite itself.
70 - passes the uts46 6.0.1 testsuite except for eleven cases which
71 apparently are bugs in the testsuite itself.
72
735.31 Sun Jan 23 11:49:19 CET 2011
74 - work around a POE bug where POE would first poll for new events and
75 invoke handlers BEFORE handling the message queue.
76 - fix parsing of "domain" statement in resolv.conf (Johannes Plunien).
77 - stop_read now does just that in TLS mode. see its
78 documentation for caveats.
79 - update to common::sense 3.4 warning set (some warnings in perl impose
80 an arbitrary 32kb limit on regex matches, which affects
81 AnyEvent::Handle).
82 - AE::Handle had a broken rbuf_max method and missing wbuf_max method.
83 - new experimental AE::Handle::resettls method.
84
855.3 Fri Dec 31 05:48:13 CET 2010
86 - major IOAsync patch by Paul Evans - 0.33 is required, as the major
87 issues with IO::Async have been solved in that version, including
88 the availabilty of a default loop!
89 - implement new wbuf_max parameter for AnyEvent::Handle.
90 - added Cocoa::EventLoop backend by Daisuke Murase.
91 - the default testsuite now honors PERL_ANYEVENT_MODEL (Paul Evans).
92 - clarify handle on_prepare documentation.
93 - set CLOEXEC flag on parent-side fd's in AnyEvent::Util::run_cmd,
94 to avoid leaking them to any child processes.
95 - try to force files into the arch-dependend lib directory, for the
96 half-"turly obsessive compulsive person"(s) that depend on it.
97
985.29 Sun Dec 5 10:49:21 CET 2010
99 - convert EV backend to EV 4.00 API (so better upgrade EV too).
100 - AE::Handle: implement dynamic read_size adjustment, add
101 max_read_size parameter, reduce default read_size to 2048.
102 - add some FAQ entries for servers and/or on_eof vs. on_error.
103 - work around OS bugs (cygwin again) when lingering in AE::Handle.
104 - reduce memory usage slightly when connect was used in AE::Handle.
105 - wrok around more segfaults in Tk.
106 - document tls_ctx => undef as valid.
107 - detect WNOHANG value at compiletime.
108 - include some probably linux-only support for building AnyEvent
109 as part of the perl core.
110 - improve compatibility of t/08_idna.t to perl 5.8.
111 - make AnyEvent::Strict truly optional (in case somebody builds a
112 minimal perl...).
113
1145.28 Wed Oct 13 04:14:23 CEST 2010
115 - due to a glitch, AnyEvent's internal getprotobyname wasn't
116 actually used in tcp_server.
117 - implement AnyEvent::CondVar->new method, which was documented but
118 missing.
119 - support autoloading of read types in unshift_read.
120 - AnyEvent::Strict: do the test for invalid fh arguments differently,
121 resulting in more informative error messages.
122 - new AnyEvent::FAQ manpage.
123 - clarify that you can actually call push_read etc. in on_read.
124 - put keys and certs in /tmp, not into the current directory,
125 and document this in the SECURITY CONSIDERATIONS sections
126 of AnyEvent::TLS.
127 - support ";" as resolv.conf comment character in AnyEvent::DNS, as
128 requested by Ron Isaacson.
129 - document $AnyEvent::DNS::RESOLVER variable.
130 - incorporated major typo patches by Abhijit Menon-Sen.
131 - AnyEvent::Handle now croaks when negative timeouts are passed.
132 - add a shitload of TCP_xxx sockopt name constants to AnyEvent::Util,
133 undocumented, but who knows when they might come in handy.
134
1355.271 Tue Jun 8 12:05:46 CEST 2010
136 - backport to perl 5.8.x.
137
1385.27 Sun Jun 6 12:12:05 CEST 2010
139 - postpone differently in AnyEvent::Socket now, as
140 when not, canceling the connection attempt might fail
141 (found by Felix Antonius Wilhelm Ostmann).
142 - explicitly check for non-stream sockets in AE::Handle, too many
143 clueless people fell into the trap of this somehow working.
144 - simplified and reworked the "OTHER MODULES" section.
145 - better/more condvar examples.
146
1475.261 Wed Apr 28 16:13:36 CEST 2010
148 - AF_INET6 was not properly used from Socket6 during configuration
149 time (found by Andy Grundman).
150
1515.26 Mon Apr 12 04:49:35 CEST 2010
152 - don't generate spurious readyness notifications when select
153 returns EINTR in the pure perl backend (analysed and testcase
154 by Bas Denissen).
155 - give same examples in SYNOPSIS sections of AnyEvent and AE.
156 - provider faster implementations for AE::time/now/now_update for
157 EV and Perl backends.
158 - provide faster AE::cv and AE::time implementations for all
159 backends.
160 - no longer support lower/mixed-case signal names (this was
161 never documented nor universally supported).
162 - some more memory ยต-optimisations, and somewhat less messy
163 code to cope with both AnyEvent and AE APIs.
164
1655.251 Sat Mar 13 00:58:21 CET 2010
166 - make test failed if it was run before make install on perl <=
167 5.8.8, and older versions of perl make our live very difficult,
168 so write constants.pl during Makefile.PL time.
169
1705.25 Sat Mar 13 00:23:14 CET 2010
171 - fix a race condition in AnyEvent::Handle that would cause
172 a "bio_read: unsupported method" error (in _another_
173 ssl connection) after user code threw an exception.
174 - added AnyEvent::Handle->destroyed method.
175 - speed up AnyEvent::Socket::format_address and ::format_ipv6.
176 - the AnyEvent::Util::fh_nonblocking function worked only
177 by lucky accident on win32.
178 - smaller and faster AnyEvent::Util::fh_nonblocking.
179 - when the (required!) Time::HiRes module is missing, AnyEvent
180 did not fall back to built-in time properly.
181 - do not load Fcntl at runtime, saving memory and loading time.
182 - precompile a number of constants and use them instead of runtime
183 detection and eval.
184 - free detection code after detection and similar memory optimisations.
185 - Perl backend timer interval best effort drift has been improved
186 (same algorithm as EV).
187 - update unicode idna mapping table.
188
1895.24 Tue Jan 5 11:39:43 CET 2010
190 - cygwin never reports errors from failed connects. 1.5 just gives
191 you continous readyness and EAGAIN, 1.7 is even more broken and
192 just hangs. work around both issues in a cygwin-specific hack.
193 - improve idle watcher documentation slightly.
194
1955.23 Sun Dec 20 23:48:00 CET 2009
196 - support IDNs in resolve_sockaddr, and therefore in tcp_connect.
197 - implement punycode_encode/decode, idn_nameprep,
198 idn_to_ascii and idn_to_unicode operations in AnyEvent::Util.
199 - provide $AE::VERSION.
200 - removed traces of "no strict 'refs'".
201
2025.22 Sat Dec 5 03:51:13 CET 2009
203 - downgrade-or-fail in AnyEvent::Handle::push_write, to
204 diagnose encoding failures earlier and more succinctly.
205 (this works around bugs in perl, throwing away encoding info
206 when passing scalar data to extensions).
207 - add more examples to AnyEvent::Socket manpage.
208 - upgrade internal warning set to the same as common::sense 2.03.
209 - use pack "n/a*" for pre-5.8.9 perl compatibility in AnyEvent::DNS
210 (John Beppu).
211 - AnyEvent::Socket::inet_aton now properly supports ipv6, as documented.
212 - add google public dns servers to fallback server set and make sure
213 we load-balance properly between the three sets. also add all
214 fallback dns servers, not just a random one, to each dns config.
215
2165.21 Thu Nov 19 02:48:47 CET 2009
217 - fix a problem where socket constants were called with parameters
218 (spotted by David Friedland).
219 - fork_call never use'd POSIX (reported by Daisuke Maki).
220 - improve perl 5.6 compatibility further (but it still won't work
221 unless you rip out everything but the core).
222 - prefer Net::DNS::Resolver over ipconfig, if installed, on win32.
223 uses 10MB of RAM, but doesn't open a console window. *sigh*.
224
2255.202 Wed Oct 14 22:35:44 CEST 2009
226 - AnyEvent::DNS would unexpectedly clobber $_ under windows
227 (analysed by Matthias Waldorf).
228 - AnyEvent::Handle::run_cmd can now pass the PID of the
229 newly-created process, which is much less useful than it might
230 sound (based on patch by Yann Kerherve).
231
2325.201 Tue Sep 29 12:09:25 CEST 2009
233 - AnyEvent:Handle::on_starttls/on_stoptls methods were broken
234 (reported by Torsten Foertsch).
235 - common::sense 2.0 could cause tcp_server to throw an exception
236 (analysed by elmex).
237
2385.2 Mon Sep 14 07:04:49 CEST 2009
239 - INCOMPATIBLE CHANGE: do no longer support register_read_type
240 and register_write_type in AnyEvent::Handle, instead support
241 package names (the facility was mostly abused).
242 - implement "packagename-as-read/write type" support in
243 AnyEvent::Handle.
244 - AnyEvent::Handle: new options "keepalive" and "oobinline".
245 - oobinline set by default to avoid security issues.
246 - the pure-perl event loop backend wrongly detected times() fallback
247 support (spotted by Pavel Boldin).
248
2495.12 Tue Sep 1 20:26:50 CEST 2009
250 - be more lenient when parsing resolv.conf files, as some people
251 use hashmarks as comment indicator inside directives (reported
252 by Michael S. Fischer).
253 - use same set of warnings as common::sense 2.0.
254 - fix a potential 32 bit overflow issue due to perl having problems
255 with large hex constants in the Perl backend.
256 - do not use the slower 5.6 version fo the enc_name code in AnyEvent::DNS
257 by default.
258 - fix some prototypes.
259 - (optionally) require Async::Interrupt 1.02.
260 - replace opendns fallbacks by verizon/level3 public dns servers.
261
2625.112 Fri Aug 21 13:59:04 CEST 2009
263 - AnyEvent::Handle could cause hard crashes in Net::SSLeay when
264 using starttls with an illegal/uncreatable tls context.
265 - port to broken win32 perls.
266
2675.111 Sun Aug 16 18:44:36 CEST 2009
268 - perl errornously requires sockaddr_un structures returned by the
269 kernel to have a specific length. work around this bug by padding
270 sockaddr_un structures (found to fail on OS X, as perl only uses
271 the correct code for GNU/Linux, but likely to fail on other
272 OSes as well).
273
2745.11 Wed Aug 12 17:49:37 CEST 2009
275 - fixed a bug in run_cmd with input from filename.
276 - create a json coder object when none is given in write direction
277 to reduce dependency on JSON.
278
2795.1 Tue Aug 11 03:17:32 CEST 2009
280 - tcp_connect now reports the correct error in $!, not an unrelated
281 one (was broken since 4.91).
282 - AE::Handle did not properly clear rtimeout/wtimeout watchers
283 when retrying a tcp_connect.
284 - new functions: AnyEvent::Util::run_cmd and close_all_fds_except.
285 - fixed wrong documentation in AE::Socket.
286
2875.01 Mon Aug 10 03:16:32 CEST 2009
288 - last release broke EV child watchers :/.
289
2905.0 Sun Aug 9 17:32:09 CEST 2009
291 - officially introduce and document the AE API.
292 - lowering the timeout at runtime did not have immediate effect
293 in AnyEvent::Handle.
294 - AE::Handle now has separate and independent read and write timeout
295 settings.
296 - probe for Guard module when AnyEvent::Util is loaded, not at
297 runtime (Event doesn't like eval "use" from callbacks, and it avoids
298 exporting the probe function).
299 - Impl::IOAsync's io watchers did not use the filehandle emulation
300 code to work around it's watcher limits.
301 - work around IO::Async's problems with overloaded objects.
302 - try to disable wearnings inside IO::Async, it's just too buggy.
303 - do no longer use Event by default, only EV or Perl, Event is too buggy.
304 - hackishly provide fast AE:: interface when EV or Perl is the backend.
305 - inherit VERSION from AnyEvent::VERSION in most modules.
306
3074.91 Thu Aug 6 15:42:45 CEST 2009
308 - AE::Handle::starttls could get out of sync when the read buffer
309 already contains some TLS handshake.
310 - AE::Handle did not properly free the TLS session, causing bigger
311 memory leaks in Net::SSLeay than expected (reported by toaster).
312 - AE::Socket::tcp_connect will now artificially delay invoking
313 the callback to avoid returning after invoking the callback.
314 - convert many internal watcher uses to AE API.
315
3164.9 Sat Aug 1 11:07:01 CEST 2009
317 - Glib silently fails when registering a timeout with
318 a negative value, so avoid that.
319 - call condvar callback immediately when it is set
320 after the condition is already signalled.
321 - check rbuf_max condition only after trying to consume
322 data.
323 - normalise signal numbers to names when using the ae's
324 signal handling, but do not document this (yet).
325 - pure perl signal emulation did not properly set nonblocking
326 mode on the signal pipe on win32 (this is unlikely to have caused
327 any issues).
328 - new module: AnyEvent::Debug.
329 - AnyEvent::Strict now checks that a signal was specified
330 by name (not name or number).
331 - reduce memory footprint in typical cases by ~50kb by
332 only compiling signal code when necessary.
333 - add AnyEvent::Handle->rbuf_max.
334 - grab the AE namespace for future new API, implement stubs for
335 said future/alternative API.
336 - new function: AnyEvent::Socket::format_hostport.
337
3384.881 Tue Jul 28 12:51:53 CEST 2009
339 - work around a bug in local in pre-5.10 perls, causing
340 AnyEvent::Handle to recurse when it should not
341 (analysed by elmex).
342
3434.88 Tue Jul 28 04:04:37 CEST 2009
3 - re-bless the handle into a dummy package after calling 344 - re-bless the handle into a dummy package after calling
4 AnyEvent::Handle::destroy, so the user does not need to check 345 AnyEvent::Handle::destroy, so the user does not need to check
5 for errors after every push_write etc. 346 for errors after every push_write etc.
347 - do not attempt to run t/02_signals.t on obviously broken
348 platforms.
6 349
74.87 Sun Jul 26 02:06:16 CEST 2009 3504.87 Sun Jul 26 02:06:16 CEST 2009
8 - do not attempt to linger when there is no longer a valid fh. 351 - do not attempt to linger when there is no longer a valid fh.
9 - enforce tls mode to be either accept or connect, do not simply 352 - enforce tls mode to be either accept or connect, do not simply
10 segfault in Net::SSLeay. 353 segfault in Net::SSLeay.
… …
18 that versions before 1.33 of Net::SSLeay are not secure. 361 that versions before 1.33 of Net::SSLeay are not secure.
19 - work around signal handling races in Event and (...) Event::Lib. 362 - work around signal handling races in Event and (...) Event::Lib.
20 - try to align signal-race timer to full-second boundaries. 363 - try to align signal-race timer to full-second boundaries.
21 - work around Tk not liking negative timeouts. 364 - work around Tk not liking negative timeouts.
22 - don't complain of different grades of environmental unfriendlyness 365 - don't complain of different grades of environmental unfriendlyness
23 in IO::ASync. 366 in IO::Async.
24 367
254.85 Sat Jul 18 06:16:14 CEST 2009 3684.85 Sat Jul 18 06:16:14 CEST 2009
26 - nail the signal race problem in perl once and for all 369 - nail the signal race problem in perl once and for all
27 (see $AnyEvent::MAX_SIGNAL_LATENCY). 370 (see $AnyEvent::MAX_SIGNAL_LATENCY).
28 - take advantage of Async::Interrupt if it is available. 371 - take advantage of Async::Interrupt if it is available.
… …
2254.3 Fri Oct 3 09:18:43 CEST 2008 5684.3 Fri Oct 3 09:18:43 CEST 2008
226 - AnyEvent will now install a no-op signal handler for SIGPIPE, 569 - AnyEvent will now install a no-op signal handler for SIGPIPE,
227 unless one has been installed already. 570 unless one has been installed already.
228 - warn about not loaidng AnyEvent::Impl::POE early enough 571 - warn about not loaidng AnyEvent::Impl::POE early enough
229 (patch by Adam Rosenstein). 572 (patch by Adam Rosenstein).
230 - fixed a great number of bugs and corner cases in AnyEvent::Handle: 573 - fixed a great number of bugs and corner cases in AnyEvent::Handle.
231 - fix a bug in where in SSL connect mode, the 574 - fix a bug in where in SSL connect mode, the
232 client would first wait for some data by the server and 575 client would first wait for some data by the server and
233 otherwise hang. (reported and analyzed in an absolutely 576 otherwise hang. (reported and analysed in an absolutely
234 exemplary manner by Adam Rosenstein). 577 exemplary manner by Adam Rosenstein).
235 - fix a bug in where SSL EOF would not be treated 578 - fix a bug in where SSL EOF would not be treated
236 as stream EOF, putting the connection into a hung state. 579 as stream EOF, putting the connection into a hung state.
237 (reported and analyzed in an absolutely exemplary manner by Adam 580 (reported and analysed in an absolutely exemplary manner by Adam
238 Rosenstein). 581 Rosenstein).
239 - fix a potential segfault when the TLS context 582 - fix a potential segfault when the TLS context
240 would go missing in a read callback (e.g. due to stoptls) - 583 would go missing in a read callback (e.g. due to stoptls) -
241 Net::SSLeay of course makes no type checking whatsoever. 584 Net::SSLeay of course makes no type checking whatsoever.
242 - AnyEvent::Handle will not stop the read watcher in TLS mode. 585 - AnyEvent::Handle will not stop the read watcher in TLS mode.

Diff Legend

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