From: Abhijit Sarkar <lists@...>
Date: 2013-02-25T05:50:49+09:00
Subject: Re: Ruby Multithreaded producer-consumer problem

Robert Klemme wrote in post #1098804:
> Access to at least these variables needs proper synchronization:
> Worker@next_run
> OrangeTree@age
> OrangeTree@orange_count

It seems that you are looking at the Monitor version of the code. I was
trying everything under the sun to get the code working so there exist a
couple of versions. The version Sean fixed does NOT use a Monitor, it
uses a Mutex.
I've uploaded my Mutex and Monitor versions and Sean's Mutex version to
my GitHub
(https://github.com/abhijitsarkar/ruby/tree/master/learn_to_program).
The names are self-descriptive (orange_tree_*)
The Monitor version still has the deadlock issue but I have pushed it
off for later.

> No.  If the age_increaser exits before the orange_picker starts to
> wait it means that it will have set @next_run to :orange.
> orange_picker won't block then.

Again, guess we are talking about 2 different versions.

> No, the deadlock error was caused by the fact that one of the two
> threads died silently because of Math.rand().  When you fix that by
> deleting "Math." your code works.
>
You are right that Math.rand was a culprit but it wasn't alone. In Mutex
version, the deadlock happens without the following check even with
Math.rand fixed.
if @orange_tree.age < OrangeTree::AGE_TO_DIE
  @cv.wait(@mutex)
end
> Can you please share the code which is working and the change you
> believe made the difference?

https://github.com/abhijitsarkar/ruby/tree/master/learn_to_program/orange_tree_*

I apologize if there was any confusion about different versions of the
program. It is not so confusing in the forum posts but I guess may not
be in the mailing list.

-- 
Posted via http://www.ruby-forum.com/.