Avoid trying to acquire lock twice when shutting down

[vinistock]

Motivation

Closes #3039

We found another spot where nested code was trying to acquire the same mutex lock. Essentially, our shutdown was doing this.

@global_state.synchronize do
  # Joining this thread will also try to acquire the same lock, which leads to a deadlock
  # We need to synchronize to prevent mangled output being written to the pipes
  @outgoing_queue.join
end

Implementation

I think there's no real point in joining the worker and outgoing threads. The server is already being shutdown anyway, so why would we try to return more responses back to the editor? We can just terminate the threads and prevent a deadlock from happening.

[vinistock]

• Avoid trying to acquire lock twice when shutting down #3036 👈 (View in Graphite)
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.