tested with ruby 1.9.3-p286 and ruby 2.0.0dev attempted rubinius fixes add rubyisms.h header for encapsulating helpful ruby logic add em_blocking_region as an abstraction on top of available primitives: rb_thread_call_without_gvl, rb_thread_blocking_region, TRAP_BEG/TRAP_END macros, or rb_enable_interrupt/rb_disable_interrupt/rb_thread_check_ints fix (?) an oddity where epoll_wait was marked as a blocking region, but was being called with a timeout of 0 (return immediately). this is now a blocking region behaving similary to other blocking regions in EventMachine, with a timeout based on _TimeTilNextEvent() add pry to Gemfile for debugging provide accessors on Bindable_t for BindingBag's max size and current size, expose to ruby as EM.max_bindings and EM.live_bindings add test case verifying BindingBag accessor behavior static analysis cleanup, make clang/gcc behave more consistently
…begin called on the connection when epoll is enabled. Fix this by only setting EPOLLIN in the new ConnectionDescriptor if SelectForRead is true. If pause is called in post_init, SelectForRead will have become false during EventCallback.
Code gated on EM.reactor_running? will cause a RuntimeError to be thrown if executed *after* a call to release_machine() but *before* @reactor_running is set to false. A prime example of this is code executed in a signal handler. Conflicts: lib/eventmachine.rb
ruby-pg project no longer uses 1.0.0a version but 1.0.1c instead. When follow the warning instructions users will still fail to obtain the right version. Allowing to provide an alternative version is good :-)