Rewrite the computer thread system

[SquidDev]

This makes a couple of significant changes to the original system, to reduce the number of threads created and allowing for multiple threads in the future. There are several notable changes from the original implementation:

• A blocking queue is used for the main task queue queue. This removes the need for the "monitor" variable and allows for multiple threads polling this queue in the future.
• The thread used to execute tasks is "cached" between tasks, significantly reducing the number of threads which need to be created. If a task needs to be stopped then the thread is then terminated and a new one constructed, though this rarely happens.

I'd really like someone with more multi-threading/concurrency knowledge to go over this and ensure it is sound. I'm not entirely sure if I need the semaphore class, or if .wait() and .notify() are sufficient. Nor am I sure if I've got my locks in entirely the correct place.

[dmarcuse]

Looked through quickly and didn't see anything concerning, might want to wait for someone else with better experience with concurrency.

[JLLeitschuh]

Reset the interrupt flag??

Thread.currentThread().interrupt();

http://www.yegor256.com/2015/10/20/interrupted-exception.html

[SquidDev]

Yeah, I saw your comments on one of the older issues about this. I meaning to do it and forgetting :).

[JLLeitschuh]

What if the interrupted exception gets thrown?
Do you want to handle that differently?
What about something like an OutOfMemoryError?

[SquidDev]

This part was just a direct copy of the original implementation :). I think it would probably be safe just to catch RuntimeException, though we may also want to catch OrphanedThread.

[JLLeitschuh]

Your whole starting/stopping infrastructure seems to be covered by guava's service helper methods.

https://github.com/google/guava/wiki/ServiceExplained

Perhaps consider using those instead?

[JLLeitschuh]

See my comment about the InterruptedException below.

[SquidDev]

Is this really needed here? The thread is going to terminate at this point anyway it doesn't seem worth it. I might be wrong though.

[JLLeitschuh]

Can submit be called by multiple threads concurrently?

[SquidDev]

No. Each TaskRunner will handle exactly one TaskExecutor.

[SquidDev]

Your whole starting/stopping infrastructure seems to be covered by guava's service helper methods.

It sort of is. However, using that does seem rather overkill for a very simple problem. I'd also rather avoid introducing a dependency on Guava, so emulators don't have to bundle it (though I accidentally did that in #159).

[SquidDev]

One thing worth noting is that if the computer cannot be aborted (such as in #174 or in string pattern matching) then the computer will spin forever. From what I can tell, LuaThread.State.lua_resume (namely line 326) does not appear to respond to interruptions. At a guess (I haven't investigated too much), it's because LuaThread.State.run also locks on the monitor.

Also worth noting that the synchronized blocks on m_machine cannot be interrupted, causing problems when a new thread is created. It might be better to switch to a ReentrantLock and use lockInterruptibly.

[JLLeitschuh]

I'd also rather avoid introducing a dependency on Guava, so emulators don't have to bundle it (though I accidentally did that in #159).

Doesn't Minecraft/forge have a dependency on Guava already?

However, using that does seem rather overkill for a very simple problem

The reason that I suggest this is because multi threading is a complicated problem and I always try to use guava's threading libraries because I know it will be better tested than my own attempts.