| 1 | Revision history for Perl extension AnyEvent. |
1 | Revision history for Perl extension AnyEvent. |
| 2 | |
2 | |
| 3 | TODO: implement getprotobyname |
3 | TODO: document TCP_*** constants |
|
|
4 | TODO: catch AE::xxx for anyevent strict and anyevent debug |
|
|
5 | TODO: anyevent::trace? or debug tracing all watchers? |
|
|
6 | TODO: anyevent strict checking for $_ changes? |
|
|
7 | |
|
|
8 | TODO: wbuf_max for anyevent |
|
|
9 | - major IOAsync patch by Paul Evans - 0.33 is required, as the major |
|
|
10 | issues with IO::Async have been solved in that version, including |
|
|
11 | the availabilty of a default loop! |
|
|
12 | - the default testsuite now honors PERL_ANYEVENT_MODEL (Paul Evans). |
|
|
13 | - set CLOEXEC flag on parent-side fd's in AnyEvent::Util::run_cmd, |
|
|
14 | to avoid leaking them to any child processes. |
|
|
15 | - try to force files into the arch-dependend lib directory, for the |
|
|
16 | half-"turly obsessive compulsive person"(s) that depend on it. |
|
|
17 | |
|
|
18 | 5.29 Sun Dec 5 10:49:21 CET 2010 |
|
|
19 | - convert EV backend to EV 4.00 API (so better upgrade EV too). |
|
|
20 | - AE::Handle: implement dynamic read_size adjustment, add |
|
|
21 | max_read_size parameter, reduce default read_size to 2048. |
|
|
22 | - add some FAQ entries for servers and/or on_eof vs. on_error. |
|
|
23 | - work around OS bugs (cygwin again) when lingering in AE::Handle. |
|
|
24 | - reduce memory usage slightly when connect was used in AE::Handle. |
|
|
25 | - wrok around more segfaults in Tk. |
|
|
26 | - document tls_ctx => undef as valid. |
|
|
27 | - detect WNOHANG value at compiletime. |
|
|
28 | - include some probably linux-only support for building AnyEvent |
|
|
29 | as part of the perl core. |
|
|
30 | - improve compatibility of t/08_idna.t to perl 5.8. |
|
|
31 | - make AnyEvent::Strict truly optional (in case somebody builds a |
|
|
32 | minimal perl...). |
|
|
33 | |
|
|
34 | 5.28 Wed Oct 13 04:14:23 CEST 2010 |
|
|
35 | - due to a glitch, AnyEvent's internal getprotobyname wasn't |
|
|
36 | actually used in tcp_server. |
|
|
37 | - implement AnyEvent::CondVar->new method, which was documented but |
|
|
38 | missing. |
|
|
39 | - support autoloading of read types in unshift_read. |
|
|
40 | - AnyEvent::Strict: do the test for invalid fh arguments differently, |
|
|
41 | resulting in more informative error messages. |
|
|
42 | - new AnyEvent::FAQ manpage. |
|
|
43 | - clarify that you can actually call push_read etc. in on_read. |
|
|
44 | - put keys and certs in /tmp, not into the current directory, |
|
|
45 | and document this in the SECURITY CONSIDERATIONS sections |
|
|
46 | of AnyEvent::TLS. |
|
|
47 | - support ";" as resolv.conf comment character in AnyEvent::DNS, as |
|
|
48 | requested by Ron Isaacson. |
|
|
49 | - document $AnyEvent::DNS::RESOLVER variable. |
|
|
50 | - incorporated major typo patches by Abhijit Menon-Sen. |
|
|
51 | - AnyEvent::Handle now croaks when negative timeouts are passed. |
|
|
52 | - add a shitload of TCP_xxx sockopt name constants to AnyEvent::Util, |
|
|
53 | undocumented, but who knows when they might come in handy. |
|
|
54 | |
|
|
55 | 5.271 Tue Jun 8 12:05:46 CEST 2010 |
|
|
56 | - backport to perl 5.8.x. |
|
|
57 | |
|
|
58 | 5.27 Sun Jun 6 12:12:05 CEST 2010 |
|
|
59 | - postpone differently in AnyEvent::Socket now, as |
|
|
60 | when not, canceling the connection attempt might fail |
|
|
61 | (found by Felix Antonius Wilhelm Ostmann). |
|
|
62 | - explicitly check for non-stream sockets in AE::Handle, too many |
|
|
63 | clueless people fell into the trap of this somehow working. |
|
|
64 | - simplified and reworked the "OTHER MODULES" section. |
|
|
65 | - better/more condvar examples. |
|
|
66 | |
|
|
67 | 5.261 Wed Apr 28 16:13:36 CEST 2010 |
|
|
68 | - AF_INET6 was not properly used from Socket6 during configuration |
|
|
69 | time (found by Andy Grundman). |
|
|
70 | |
|
|
71 | 5.26 Mon Apr 12 04:49:35 CEST 2010 |
|
|
72 | - don't generate spurious readyness notifications when select |
|
|
73 | returns EINTR in the pure perl backend (analysed and testcase |
|
|
74 | by Bas Denissen). |
|
|
75 | - give same examples in SYNOPSIS sections of AnyEvent and AE. |
|
|
76 | - provider faster implementations for AE::time/now/now_update for |
|
|
77 | EV and Perl backends. |
|
|
78 | - provide faster AE::cv and AE::time implementations for all |
|
|
79 | backends. |
|
|
80 | - no longer support lower/mixed-case signal names (this was |
|
|
81 | never documented nor universally supported). |
|
|
82 | - some more memory ยต-optimisations, and somewhat less messy |
|
|
83 | code to cope with both AnyEvent and AE APIs. |
|
|
84 | |
|
|
85 | 5.251 Sat Mar 13 00:58:21 CET 2010 |
|
|
86 | - make test failed if it was run before make install on perl <= |
|
|
87 | 5.8.8, and older versions of perl make our live very difficult, |
|
|
88 | so write constants.pl during Makefile.PL time. |
|
|
89 | |
|
|
90 | 5.25 Sat Mar 13 00:23:14 CET 2010 |
|
|
91 | - fix a race condition in AnyEvent::Handle that would cause |
|
|
92 | a "bio_read: unsupported method" error (in _another_ |
|
|
93 | ssl connection) after user code threw an exception. |
|
|
94 | - added AnyEvent::Handle->destroyed method. |
|
|
95 | - speed up AnyEvent::Socket::format_address and ::format_ipv6. |
|
|
96 | - the AnyEvent::Util::fh_nonblocking function worked only |
|
|
97 | by lucky accident on win32. |
|
|
98 | - smaller and faster AnyEvent::Util::fh_nonblocking. |
|
|
99 | - when the (required!) Time::HiRes module is missing, AnyEvent |
|
|
100 | did not fall back to built-in time properly. |
|
|
101 | - do not load Fcntl at runtime, saving memory and loading time. |
|
|
102 | - precompile a number of constants and use them instead of runtime |
|
|
103 | detection and eval. |
|
|
104 | - free detection code after detection and similar memory optimisations. |
|
|
105 | - Perl backend timer interval best effort drift has been improved |
|
|
106 | (same algorithm as EV). |
|
|
107 | - update unicode idna mapping table. |
|
|
108 | |
|
|
109 | 5.24 Tue Jan 5 11:39:43 CET 2010 |
|
|
110 | - cygwin never reports errors from failed connects. 1.5 just gives |
|
|
111 | you continous readyness and EAGAIN, 1.7 is even more broken and |
|
|
112 | just hangs. work around both issues in a cygwin-specific hack. |
|
|
113 | - improve idle watcher documentation slightly. |
|
|
114 | |
|
|
115 | 5.23 Sun Dec 20 23:48:00 CET 2009 |
|
|
116 | - support IDNs in resolve_sockaddr, and therefore in tcp_connect. |
|
|
117 | - implement punycode_encode/decode, idn_nameprep, |
|
|
118 | idn_to_ascii and idn_to_unicode operations in AnyEvent::Util. |
|
|
119 | - provide $AE::VERSION. |
|
|
120 | - removed traces of "no strict 'refs'". |
|
|
121 | |
|
|
122 | 5.22 Sat Dec 5 03:51:13 CET 2009 |
|
|
123 | - downgrade-or-fail in AnyEvent::Handle::push_write, to |
|
|
124 | diagnose encoding failures earlier and more succinctly. |
|
|
125 | (this works around bugs in perl, throwing away encoding info |
|
|
126 | when passing scalar data to extensions). |
|
|
127 | - add more examples to AnyEvent::Socket manpage. |
|
|
128 | - upgrade internal warning set to the same as common::sense 2.03. |
|
|
129 | - use pack "n/a*" for pre-5.8.9 perl compatibility in AnyEvent::DNS |
|
|
130 | (John Beppu). |
|
|
131 | - AnyEvent::Socket::inet_aton now properly supports ipv6, as documented. |
|
|
132 | - add google public dns servers to fallback server set and make sure |
|
|
133 | we load-balance properly between the three sets. also add all |
|
|
134 | fallback dns servers, not just a random one, to each dns config. |
|
|
135 | |
|
|
136 | 5.21 Thu Nov 19 02:48:47 CET 2009 |
|
|
137 | - fix a problem where socket constants were called with parameters |
|
|
138 | (spotted by David Friedland). |
|
|
139 | - fork_call never use'd POSIX (reported by Daisuke Maki). |
|
|
140 | - improve perl 5.6 compatibility further (but it still won't work |
|
|
141 | unless you rip out everything but the core). |
|
|
142 | - prefer Net::DNS::Resolver over ipconfig, if installed, on win32. |
|
|
143 | uses 10MB of RAM, but doesn't open a console window. *sigh*. |
|
|
144 | |
|
|
145 | 5.202 Wed Oct 14 22:35:44 CEST 2009 |
|
|
146 | - AnyEvent::DNS would unexpectedly clobber $_ under windows |
|
|
147 | (analysed by Matthias Waldorf). |
|
|
148 | - AnyEvent::Handle::run_cmd can now pass the PID of the |
|
|
149 | newly-created process, which is much less useful than it might |
|
|
150 | sound (based on patch by Yann Kerherve). |
|
|
151 | |
|
|
152 | 5.201 Tue Sep 29 12:09:25 CEST 2009 |
|
|
153 | - AnyEvent:Handle::on_starttls/on_stoptls methods were broken |
|
|
154 | (reported by Torsten Foertsch). |
|
|
155 | - common::sense 2.0 could cause tcp_server to throw an exception |
|
|
156 | (analysed by elmex). |
|
|
157 | |
|
|
158 | 5.2 Mon Sep 14 07:04:49 CEST 2009 |
|
|
159 | - INCOMPATIBLE CHANGE: do no longer support register_read_type |
|
|
160 | and register_write_type in AnyEvent::Handle, instead support |
|
|
161 | package names (the facility was mostly abused). |
|
|
162 | - implement "packagename-as-read/write type" support in |
|
|
163 | AnyEvent::Handle. |
|
|
164 | - AnyEvent::Handle: new options "keepalive" and "oobinline". |
|
|
165 | - oobinline set by default to avoid security issues. |
|
|
166 | - the pure-perl event loop backend wrongly detected times() fallback |
|
|
167 | support (spotted by Pavel Boldin). |
|
|
168 | |
|
|
169 | 5.12 Tue Sep 1 20:26:50 CEST 2009 |
|
|
170 | - be more lenient when parsing resolv.conf files, as some people |
|
|
171 | use hashmarks as comment indicator inside directives (reported |
|
|
172 | by Michael S. Fischer). |
|
|
173 | - use same set of warnings as common::sense 2.0. |
|
|
174 | - fix a potential 32 bit overflow issue due to perl having problems |
|
|
175 | with large hex constants in the Perl backend. |
|
|
176 | - do not use the slower 5.6 version fo the enc_name code in AnyEvent::DNS |
|
|
177 | by default. |
|
|
178 | - fix some prototypes. |
|
|
179 | - (optionally) require Async::Interrupt 1.02. |
|
|
180 | - replace opendns fallbacks by verizon/level3 public dns servers. |
|
|
181 | |
|
|
182 | 5.112 Fri Aug 21 13:59:04 CEST 2009 |
|
|
183 | - AnyEvent::Handle could cause hard crashes in Net::SSLeay when |
|
|
184 | using starttls with an illegal/uncreatable tls context. |
|
|
185 | - port to broken win32 perls. |
|
|
186 | |
|
|
187 | 5.111 Sun Aug 16 18:44:36 CEST 2009 |
|
|
188 | - perl errornously requires sockaddr_un structures returned by the |
|
|
189 | kernel to have a specific length. work around this bug by padding |
|
|
190 | sockaddr_un structures (found to fail on OS X, as perl only uses |
|
|
191 | the correct code for GNU/Linux, but likely to fail on other |
|
|
192 | OSes as well). |
|
|
193 | |
|
|
194 | 5.11 Wed Aug 12 17:49:37 CEST 2009 |
|
|
195 | - fixed a bug in run_cmd with input from filename. |
|
|
196 | - create a json coder object when none is given in write direction |
|
|
197 | to reduce dependency on JSON. |
|
|
198 | |
|
|
199 | 5.1 Tue Aug 11 03:17:32 CEST 2009 |
|
|
200 | - tcp_connect now reports the correct error in $!, not an unrelated |
|
|
201 | one (was broken since 4.91). |
|
|
202 | - AE::Handle did not properly clear rtimeout/wtimeout watchers |
|
|
203 | when retrying a tcp_connect. |
|
|
204 | - new functions: AnyEvent::Util::run_cmd and close_all_fds_except. |
|
|
205 | - fixed wrong documentation in AE::Socket. |
|
|
206 | |
|
|
207 | 5.01 Mon Aug 10 03:16:32 CEST 2009 |
|
|
208 | - last release broke EV child watchers :/. |
|
|
209 | |
|
|
210 | 5.0 Sun Aug 9 17:32:09 CEST 2009 |
|
|
211 | - officially introduce and document the AE API. |
|
|
212 | - lowering the timeout at runtime did not have immediate effect |
|
|
213 | in AnyEvent::Handle. |
|
|
214 | - AE::Handle now has separate and independent read and write timeout |
|
|
215 | settings. |
|
|
216 | - probe for Guard module when AnyEvent::Util is loaded, not at |
|
|
217 | runtime (Event doesn't like eval "use" from callbacks, and it avoids |
|
|
218 | exporting the probe function). |
|
|
219 | - Impl::IOAsync's io watchers did not use the filehandle emulation |
|
|
220 | code to work around it's watcher limits. |
|
|
221 | - work around IO::Async's problems with overloaded objects. |
|
|
222 | - try to disable wearnings inside IO::Async, it's just too buggy. |
|
|
223 | - do no longer use Event by default, only EV or Perl, Event is too buggy. |
|
|
224 | - hackishly provide fast AE:: interface when EV or Perl is the backend. |
|
|
225 | - inherit VERSION from AnyEvent::VERSION in most modules. |
|
|
226 | |
|
|
227 | 4.91 Thu Aug 6 15:42:45 CEST 2009 |
|
|
228 | - AE::Handle::starttls could get out of sync when the read buffer |
|
|
229 | already contains some TLS handshake. |
|
|
230 | - AE::Handle did not properly free the TLS session, causing bigger |
|
|
231 | memory leaks in Net::SSLeay than expected (reported by toaster). |
|
|
232 | - AE::Socket::tcp_connect will now artificially delay invoking |
|
|
233 | the callback to avoid returning after invoking the callback. |
|
|
234 | - convert many internal watcher uses to AE API. |
|
|
235 | |
|
|
236 | 4.9 Sat Aug 1 11:07:01 CEST 2009 |
|
|
237 | - Glib silently fails when registering a timeout with |
|
|
238 | a negative value, so avoid that. |
|
|
239 | - call condvar callback immediately when it is set |
|
|
240 | after the condition is already signalled. |
|
|
241 | - check rbuf_max condition only after trying to consume |
|
|
242 | data. |
|
|
243 | - normalise signal numbers to names when using the ae's |
|
|
244 | signal handling, but do not document this (yet). |
|
|
245 | - pure perl signal emulation did not properly set nonblocking |
|
|
246 | mode on the signal pipe on win32 (this is unlikely to have caused |
|
|
247 | any issues). |
|
|
248 | - new module: AnyEvent::Debug. |
|
|
249 | - AnyEvent::Strict now checks that a signal was specified |
|
|
250 | by name (not name or number). |
|
|
251 | - reduce memory footprint in typical cases by ~50kb by |
|
|
252 | only compiling signal code when necessary. |
|
|
253 | - add AnyEvent::Handle->rbuf_max. |
|
|
254 | - grab the AE namespace for future new API, implement stubs for |
|
|
255 | said future/alternative API. |
|
|
256 | - new function: AnyEvent::Socket::format_hostport. |
|
|
257 | |
|
|
258 | 4.881 Tue Jul 28 12:51:53 CEST 2009 |
|
|
259 | - work around a bug in local in pre-5.10 perls, causing |
|
|
260 | AnyEvent::Handle to recurse when it should not |
|
|
261 | (analysed by elmex). |
|
|
262 | |
|
|
263 | 4.88 Tue Jul 28 04:04:37 CEST 2009 |
|
|
264 | - re-bless the handle into a dummy package after calling |
|
|
265 | AnyEvent::Handle::destroy, so the user does not need to check |
|
|
266 | for errors after every push_write etc. |
|
|
267 | - do not attempt to run t/02_signals.t on obviously broken |
|
|
268 | platforms. |
|
|
269 | |
|
|
270 | 4.87 Sun Jul 26 02:06:16 CEST 2009 |
|
|
271 | - do not attempt to linger when there is no longer a valid fh. |
|
|
272 | - enforce tls mode to be either accept or connect, do not simply |
|
|
273 | segfault in Net::SSLeay. |
|
|
274 | - AnyEvent::Handle can now call tcp_connect itself (new parameters |
|
|
275 | connect, on_prepare, on_connect and on_connect_error). Updated |
|
|
276 | tutorial accordingly. |
|
|
277 | - add AnyEvent::Impl::Irssi backend. |
|
|
278 | |
|
|
279 | 4.86 Mon Jul 20 23:52:29 CEST 2009 |
|
|
280 | - since the verbose warning is not enough, explicitly document |
|
|
281 | that versions before 1.33 of Net::SSLeay are not secure. |
|
|
282 | - work around signal handling races in Event and (...) Event::Lib. |
|
|
283 | - try to align signal-race timer to full-second boundaries. |
|
|
284 | - work around Tk not liking negative timeouts. |
|
|
285 | - don't complain of different grades of environmental unfriendlyness |
|
|
286 | in IO::Async. |
|
|
287 | |
|
|
288 | 4.85 Sat Jul 18 06:16:14 CEST 2009 |
|
|
289 | - nail the signal race problem in perl once and for all |
|
|
290 | (see $AnyEvent::MAX_SIGNAL_LATENCY). |
|
|
291 | - take advantage of Async::Interrupt if it is available. |
|
|
292 | - load Time::HiRes and Guard modules on demand only. |
|
|
293 | - add optional/recommended modules section to AnyEvent |
|
|
294 | documentation. |
|
|
295 | - reduce memory usage considerably (and reduce startup penalty) |
|
|
296 | by not using "strict", "warnings" and "overload" modules. |
|
|
297 | - work around buggy windows/openbsd perls and provide EBADMSG |
|
|
298 | and EPROTO ourselves when missing. |
|
|
299 | - improve perl 5.6 compatibility of the core event loop. |
|
|
300 | - made Net::SSLeay version 1.33 a soft requirement. |
|
|
301 | |
|
|
302 | 4.83 Fri Jul 17 16:56:26 CEST 2009 |
|
|
303 | - implement AnyEvent::Socket::getprotobyname. |
|
|
304 | - AnyEvent::CondVar's will now detect recursive blocking |
|
|
305 | waits and will croak, as too many people fall into |
|
|
306 | this trap. |
| 4 | - AnyEvent::Handle will now call ->destroy on itself after |
307 | - AnyEvent::Handle will now call ->destroy on itself after |
| 5 | executing the on_error callback, instead of doing some |
308 | executing the on_error callback, instead of doing some |
| 6 | half-baked internal shutdown, for fatal errors. |
309 | half-baked internal shutdown, for fatal errors. |
|
|
310 | - clarify on_eof behaviour w.r.t. the read queue and |
|
|
311 | on_read callbacks. |
| 7 | - ignore some possible spurious wake-ups in tcp_connect. |
312 | - ignore some possible spurious wake-ups in tcp_connect. |
| 8 | |
313 | |
| 9 | 4.82 Sat Jul 11 00:34:55 CEST 2009 |
314 | 4.82 Sat Jul 11 00:34:55 CEST 2009 |
| 10 | - POE and Event backends didn't accept some callable objects as |
315 | - POE and Event backends didn't accept some callable objects as |
| 11 | callbacks. |
316 | callbacks. |
| … | |
… | |
| 183 | 4.3 Fri Oct 3 09:18:43 CEST 2008 |
488 | 4.3 Fri Oct 3 09:18:43 CEST 2008 |
| 184 | - AnyEvent will now install a no-op signal handler for SIGPIPE, |
489 | - AnyEvent will now install a no-op signal handler for SIGPIPE, |
| 185 | unless one has been installed already. |
490 | unless one has been installed already. |
| 186 | - warn about not loaidng AnyEvent::Impl::POE early enough |
491 | - warn about not loaidng AnyEvent::Impl::POE early enough |
| 187 | (patch by Adam Rosenstein). |
492 | (patch by Adam Rosenstein). |
| 188 | - fixed a great number of bugs and corner cases in AnyEvent::Handle: |
493 | - fixed a great number of bugs and corner cases in AnyEvent::Handle. |
| 189 | - fix a bug in where in SSL connect mode, the |
494 | - fix a bug in where in SSL connect mode, the |
| 190 | client would first wait for some data by the server and |
495 | client would first wait for some data by the server and |
| 191 | otherwise hang. (reported and analyzed in an absolutely |
496 | otherwise hang. (reported and analysed in an absolutely |
| 192 | exemplary manner by Adam Rosenstein). |
497 | exemplary manner by Adam Rosenstein). |
| 193 | - fix a bug in where SSL EOF would not be treated |
498 | - fix a bug in where SSL EOF would not be treated |
| 194 | as stream EOF, putting the connection into a hung state. |
499 | as stream EOF, putting the connection into a hung state. |
| 195 | (reported and analyzed in an absolutely exemplary manner by Adam |
500 | (reported and analysed in an absolutely exemplary manner by Adam |
| 196 | Rosenstein). |
501 | Rosenstein). |
| 197 | - fix a potential segfault when the TLS context |
502 | - fix a potential segfault when the TLS context |
| 198 | would go missing in a read callback (e.g. due to stoptls) - |
503 | would go missing in a read callback (e.g. due to stoptls) - |
| 199 | Net::SSLeay of course makes no type checking whatsoever. |
504 | Net::SSLeay of course makes no type checking whatsoever. |
| 200 | - AnyEvent::Handle will not stop the read watcher in TLS mode. |
505 | - AnyEvent::Handle will not stop the read watcher in TLS mode. |
