--- a/luaevent/CHANGELOG Tue Aug 28 17:50:26 2007 -0400 +++ /dev/null Thu Jan 01 00:00:00 1970 +0000 @@ -1,37 +0,0 @@ -====== -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