[Stackless] Consistent Crashes in Stackless

Andrew Francis andrewfr_ice at yahoo.com
Fri Nov 21 19:43:29 CET 2008


Hello Timothy:


>I upgraded to python 2.6 (stackless 3.1b) and it no longer crashes,
>however the program still ends after a few iterations. From my testing
>it almost seems like stackless.run() returns before all the original
>messages have been sent.

I am using Stackless 3.1b under python 2.5.2 on an ASUS EEE (linux) and
I get a segmentation fault (I am wondering if a TaskletExit exception is being thrown? ) Funny, when I added more statements after stackless.run(),
then segmentation fault disappeared.

> I just now realized though that with random sending, it could be 
>possible for every single task to be waiting to
>send to another task. So basically everyone is talking and no one is
listening.

>From what I see, some of your tasklets' channels are waiting for senders. Other tasklets are waiting for receivers. However when they are all blocked, stackless.run() will return

> The lack of async communication in Stackless is really
>starting to bug me. I guess it's time for some more hacking....

I find your programme logic hard to follow. Concerning queues. Look at the  uthread library (uthread.py) for implementation of queues. I had a problem similar to yours. Queues seem to solve the problem. However queues really hide a problem in the programme logic.

Cheers,
Andrew 




      




More information about the Stackless mailing list