| 1 | Revision history for libev, a high-performance and full-featured event loop. |
1 | Revision history for libev, a high-performance and full-featured event loop. |
| 2 | |
2 | |
| 3 | TODO: ev_walk |
3 | TODO: ev_walk |
| 4 | TODO: remain |
4 | TODO: signal handling per loop |
| 5 | TODO: EV_MINIMAL |
|
|
| 6 | |
5 | |
| 7 | - new functionality: ev_set_userdata, ev_userdata, |
6 | TODO: nsig |
| 8 | ev_set_invoke_pending_cb, ev_set_loop_release_cb, |
7 | - incompatible change: do not necessarily reset signal handler |
| 9 | ev_invoke_pending, together with a long example about thread |
8 | to SIG_DFL when a sighandler is stopped. |
| 10 | locking. |
9 | - ev_default_destroy did not properly free or zero some members, |
|
|
10 | potentially causing crashes and memory corruption on repated |
|
|
11 | ev_default_destroy/ev_default_loop calls. |
|
|
12 | - take advantage of signalfd on GNU/Linux systems. |
|
|
13 | - document that the signal mask might be in an unspecified |
|
|
14 | state when using libev's signal handling. |
|
|
15 | - take advantage of some GNU/Linux calls to set cloexec/nonblock |
|
|
16 | on fd creation, to avoid race conditions. |
|
|
17 | |
|
|
18 | 3.7 Fri Jul 17 16:36:32 CEST 2009 |
| 11 | - ev_unloop and ev_loop wrongly used a global variable to exit loops, |
19 | - ev_unloop and ev_loop wrongly used a global variable to exit loops, |
| 12 | instead of using a per-loop variable (bug caught by accident...). |
20 | instead of using a per-loop variable (bug caught by accident...). |
|
|
21 | - the ev_set_io_collect_interval interpretation has changed. |
|
|
22 | - add new functionality: ev_set_userdata, ev_userdata, |
|
|
23 | ev_set_invoke_pending_cb, ev_set_loop_release_cb, |
|
|
24 | ev_invoke_pending, ev_pending_count, together with a long example |
|
|
25 | about thread locking. |
|
|
26 | - add ev_timer_remaining (as requested by Denis F. Latypoff). |
|
|
27 | - add ev_loop_depth. |
| 13 | - calling ev_unloop in fork/prepare watchers will no longer poll |
28 | - calling ev_unloop in fork/prepare watchers will no longer poll |
| 14 | for new events. |
29 | for new events. |
| 15 | - Denis F. Latypoff corrected many typos in example code snippets. |
30 | - Denis F. Latypoff corrected many typos in example code snippets. |
| 16 | - the ev_set_io_collect_interval interpretation has changed. |
|
|
| 17 | - honor autoconf detection of EV_USE_CLOCK_SYSCALL, also double- |
31 | - honor autoconf detection of EV_USE_CLOCK_SYSCALL, also double- |
| 18 | check that the syscall number is available before trying to |
32 | check that the syscall number is available before trying to |
| 19 | use it (reported by ry@tinyclouds). |
33 | use it (reported by ry@tinyclouds). |
| 20 | - use GetSystemTimeAsFileTime instead of _timeb on windows, for |
34 | - use GetSystemTimeAsFileTime instead of _timeb on windows, for |
| 21 | slightly higher accuracy. |
35 | slightly higher accuracy. |
| 22 | - properly declare ev_loop_verify and ev_now_update even when |
36 | - properly declare ev_loop_verify and ev_now_update even when |
| 23 | !EV_MULTIPLICITY. |
37 | !EV_MULTIPLICITY. |
| 24 | - implement ev_loop_depth. |
|
|
| 25 | - do not compile in any priority code when EV_MAXPRI == EV_MINPRI. |
38 | - do not compile in any priority code when EV_MAXPRI == EV_MINPRI. |
| 26 | - support EV_MINIMAL==2 for a reduced API. |
39 | - support EV_MINIMAL==2 for a reduced API. |
| 27 | - actually 0-initialise struct sigaction when installing signals. |
40 | - actually 0-initialise struct sigaction when installing signals. |
| 28 | - add section on hibernate and stopped processes to ev_timer docs. |
41 | - add section on hibernate and stopped processes to ev_timer docs. |
| 29 | |
42 | |
