[Stackless] Patch for pickle problem on 64bit platforms

Kristján Valur Jónsson kristjan at ccpgames.com
Fri Oct 19 15:26:29 CEST 2007


and again, in run_watchdog().  Here, it ultimately ends up as an int in ts->st.interval.
Again, confusion.  should we change ts->st.interval into long, or change the run_watchdog api to int?

grep for "long" in the code.  It is not used very often.  The parts in stackless are what you should watch for.
Any word on this from our wise guru?

K

-----Original Message-----
From: stackless-bounces at stackless.com [mailto:stackless-bounces at stackless.com] On Behalf Of Kristján Valur Jónsson
Sent: Friday, October 19, 2007 11:57
To: Hallgrimur H. Gunnarsson; stackless at stackless.com
Subject: Re: [Stackless] Patch for pickle problem on 64bit platforms

btw, the same problem exists in test_cframe().
See my previous email about the possible correctness of this all.

K

-----Original Message-----
From: stackless-bounces at stackless.com [mailto:stackless-bounces at stackless.com] On Behalf Of Hallgrimur H. Gunnarsson
Sent: Wednesday, October 17, 2007 01:23
To: stackless at stackless.com
Subject: [Stackless] Patch for pickle problem on 64bit platforms

Hi all,

The following patch fixes the pickle problem on 64bit platforms:

http://hhg.to/pickle-fix.diff

This problem was causing all pickle-related unit test to fail with
a message like:

"cannot execute invalid frame [..]: frame had a C
state that can't be restored."

(or see http://hhg.to/pickle-fix.txt for example output).

The problem was caused by a PyArg_ParseTuple int/long size
misalignment in cframe_setstate (core/cframeobject.c).

-- Hallgrimur

_______________________________________________
Stackless mailing list
Stackless at stackless.com
http://www.stackless.com/mailman/listinfo/stackless

_______________________________________________
Stackless mailing list
Stackless at stackless.com
http://www.stackless.com/mailman/listinfo/stackless




More information about the Stackless mailing list