RRB is a pretty heavy usage test and I don't think you need to prioritize this.
The RTL branch ran RRB just fine without JIT, at basically the same speed as release Ruby 2.6. But it crashed rapidly when run with --jit. Here's the stack traces I got when it did:
Exception in worker thread: undefined method `setsockopt' for nil:NilClass
/home/ubuntu/mount-mjit/lib/ruby/2.6.0/net/http.rb:953:in `connect'
/home/ubuntu/mount-mjit/lib/ruby/2.6.0/net/http.rb:930:in `do_start'
/home/ubuntu/mount-mjit/lib/ruby/2.6.0/net/http.rb:919:in `start'
/home/ubuntu/.rvm/gems/ext-rtl-mjit/gems/rest-client-1.8.0/lib/restclient/request.rb:413:in `transmit'
/home/ubuntu/.rvm/gems/ext-rtl-mjit/gems/rest-client-1.8.0/lib/restclient/request.rb:176:in `execute'
/home/ubuntu/.rvm/gems/ext-rtl-mjit/gems/rest-client-1.8.0/lib/restclient/request.rb:41:in `execute'
/home/ubuntu/rails_ruby_bench/user_simulator.rb:53:in `request'
/home/ubuntu/rails_ruby_bench/user_simulator.rb:72:in `action_from_args'
/home/ubuntu/rails_ruby_bench/user_simulator.rb:150:in `block in time_actions'
/home/ubuntu/rails_ruby_bench/user_simulator.rb:149:in `each'
/home/ubuntu/rails_ruby_bench/user_simulator.rb:149:in `time_actions'
/home/ubuntu/rails_ruby_bench/user_simulator.rb:179:in `block (2 levels) in multithreaded_actions'
#<Thread:0x0000562a0645be20@/home/ubuntu/rails_ruby_bench/user_simulator.rb:170 run> terminated with exception (report_on_exception is true):
Traceback (most recent call last):
11: from /home/ubuntu/rails_ruby_bench/user_simulator.rb:179:in `block (2 levels) in multithreaded_actions'
10: from /home/ubuntu/rails_ruby_bench/user_simulator.rb:149:in `time_actions'
9: from /home/ubuntu/rails_ruby_bench/user_simulator.rb:149:in `each'
8: from /home/ubuntu/rails_ruby_bench/user_simulator.rb:150:in `block in time_actions'
7: from /home/ubuntu/rails_ruby_bench/user_simulator.rb:72:in `action_from_args'
6: from /home/ubuntu/rails_ruby_bench/user_simulator.rb:53:in `request'
5: from /home/ubuntu/.rvm/gems/ext-rtl-mjit/gems/rest-client-1.8.0/lib/restclient/request.rb:41:in `execute'
4: from /home/ubuntu/.rvm/gems/ext-rtl-mjit/gems/rest-client-1.8.0/lib/restclient/request.rb:176:in `execute'
3: from /home/ubuntu/.rvm/gems/ext-rtl-mjit/gems/rest-client-1.8.0/lib/restclient/request.rb:413:in `transmit'
2: from /home/ubuntu/mount-mjit/lib/ruby/2.6.0/net/http.rb:919:in `start'
1: from /home/ubuntu/mount-mjit/lib/ruby/2.6.0/net/http.rb:930:in `do_start'
/home/ubuntu/mount-mjit/lib/ruby/2.6.0/net/http.rb:953:in `connect': undefined method `setsockopt' for nil:NilClass (NoMethodError)
Release Ruby 2.6 didn't get these errors on the same EC2 instance.
RRB is a pretty heavy usage test and I don't think you need to prioritize this.
The RTL branch ran RRB just fine without JIT, at basically the same speed as release Ruby 2.6. But it crashed rapidly when run with --jit. Here's the stack traces I got when it did:
Release Ruby 2.6 didn't get these errors on the same EC2 instance.