03:38 <byteit101[m]> ominous: * @shyouhei doesn't think there is any needs of this function for 3rd party

03:38 <byteit101[m]> * extension libraries.

03:39 <byteit101[m]> How can I let MRI people know that I will be using rb_io_set_write_io so they don't deprecate it on us?

03:56 <byteit101[m]> Oh hmm... can I get access to the current thread via the c api?

11:03 <lopex[m]> https://github.com/simonis/mmap-pause/blob/master/README.md

13:12 <headius> byteit101: I would assume so but I don't know the API well

13:14 <headius> -XX:-PerfDisableSharedMem seriously

13:15 <headius> worth playing with though... we trigger lots of jit and gc and probably a large amount of performance counters as a result

17:32 <headius> I did some exploration of json dumping and came around to the idea of using Panama native memory APIs to back a String that can exceed 2GB

17:32 <headius> https://github.com/jruby/jruby/issues/6265#issuecomment-1492242640

17:32 <headius> a workable project as of JDK 19, might be time to start playing with it

20:32 <byteit101[m]> I found GET_THREAD() in the x86_64-linux/rb_mjit_min_header-3.0.0.h header, but nowhere else, and it seemed suspicously not api

20:44 <headius> yeah sounds pretty sketch

21:23 <byteit101[m]> reimplementing $? seems easier than setting the original $?