blob: f2f964d761bfd4380c0dcec45cc2a6068d81c77e (
plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
|
======
0.1.2 - Revision 15 - 2007-08-18
+ Setup system to use new coro management as described in COROUTINE_MANAGEMENT
The callbacks are called from the event_loop 'thread' rather than that which they are
created in. This will prevent the self-resume problem as well as dead-thread problems.
- Recognized issues to fix in next release:
* Socket/event closing needs to be cleaned
* luaevent.lua needs refactoring
* luaevent.[ch] need to be cleaned up
======
0.1.1 - Revision 14 - 2007-06-13
+ Fixed event-handling code to cancel events on nothing being returned
+ Added socket/object cleanup.
+ Filed bug to libevent about the strange valgrind-released errors
- Recognized following issues:
Timeouts needed
Need to handle events setup from inside a coroutine... need to get a global Lua state from a thread...
======
0.1.0 - Revision 6 - 2007-06-10 22:00 EST
Completed mostly working version
* Moved to a mode where addevent calls a callback rather than it being instantiated within.
If the callback returns -1, then no event is ever setup,
Otherwise the integer value is used to setup the event.
This allows for using coroutine.wrap rather than a cooked-up wrapper
* Tests work, although there are a few remaining issues:
* Need to figure a good way of preserving the event object,
not sure if current method is good enough, since the socket
is the only anchor, and it is only held inside the coro..
circular reference, something that Lua 'handles' well.
* Doing more than the maximum sockets the process is allows
causes strangeness to occur in libevent.. somehow
it is getting around to epoll_add which is causing valgrind
to barf.
* Added cheap protection code for failures in callback handlers
======
0.0.0 - Revision 2 - 2007-06-10 12:00 EST
Initial public version, was broken due to self-resume coroutines
|
