>
> 4. ---> And here is the problem if you execute one job after the next
> you create thread after thread until your threadpool has reached
> the maximum numbers of threads. ---> only then the old threads get
> reused ---> I think this is a problem concerning the way Ethreads
> get dequeued the BlockingQueue.
>
> This is not the way it should because maybe in the worst case
> you create a new Thread even 30 or more threads are idle.
> (Scot please don't say "Look at the code more closely." or
> something like that.)
Ah, I believe your right. The solution is to not use a BlockingQueue, but
rather a BlockingStack. That way reclaimed threads get used in preference
over new threads.
> Another thing which is not really a problem but more some kind of feeling:
> It looks not so nice if you run ps and you get the following output
> because of 40 deamons threads running (maybe it is a special feature of
> the blackdown jdk) :
> rrkapitz at Luxor:~/Freenet/thread > ps | more
> 2739 pts/0 00:00:00 java
> 2740 pts/0 00:00:00 java
> 2741 pts/0 00:00:00 java
> 2742 pts/0 00:00:00 java
> 2744 pts/0 00:00:00 java
> 2745 pts/0 00:00:00 java
> 2774 pts/0 00:00:00 java
> 2804 pts/0 00:00:00 java
> 2955 pts/0 00:00:00 java
> 2985 pts/0 00:00:00 java
> ....
Thats due to the fact that the JVM is using native threads, which are ==
to Linux processes but with a shared VM context. Annoying in PS, but
perfectly normal.
>
>
> By the way I still would like the recode the thread package :-)
>
And I still don't think its necessary. :)
Scott
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 232 bytes
Desc: not available
URL:
<https://emu.freenetproject.org/pipermail/devl/attachments/20000803/b57f0adc/attachment.pgp>
