[ python-Bugs-1370322 ] Float marshaling problems with test_colorsys

[SourceForge.net]

Bugs item #1370322, was opened at 2005-11-30 20:49
Message generated for change (Comment added) made by birkenfeld
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=105470&aid=1370322&group_id=5470

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: Python Interpreter Core
Group: Python 2.5
>Status: Closed
>Resolution: Works For Me
Priority: 7
Submitted By: Reinhold Birkenfeld (birkenfeld)
Assigned to: Michael Hudson (mwh)
Summary: Float marshaling problems with test_colorsys

Initial Comment:
Starting point: "make clean; make" in the 2.5 source tree.

"make test" hangs at test_colorsys.

Careful investigation shows that when the bytecode is
freshly generated
by "make all" (precisely in test___all__) the .pyc file
is different from what a
direct call to "regrtest.py test_colorsys" produces.

Curiously, a call to "regrtest.py test___all__" instead
of "make test" produces
the correct bytecode.

System is Linux 2.6, x86, gcc 3.4.4.

I have attached both versions of colorsys.pyc.

----------------------------------------------------------------------

>Comment By: Reinhold Birkenfeld (birkenfeld)
Date: 2006-01-06 20:23

Message:
Logged In: YES 
user_id=1188172

This has now vanished.
I don't know why, but I did a glibc update today, maybe...

----------------------------------------------------------------------

Comment By: Neal Norwitz (nnorwitz)
Date: 2005-12-11 21:05

Message:
Logged In: YES 
user_id=33168

I can't reproduce (make test) on linux 2.6, gcc 3.4.4, only
diff is amd64.  I have some outstanding changes, but none
that should affect the outcome.  Sorry, I don't know what to
tell you, can you try to debug any more?  Maybe run under
valgrind (or other memory debugger) and see if there's a
memory issue.

----------------------------------------------------------------------

Comment By: Reinhold Birkenfeld (birkenfeld)
Date: 2005-12-10 18:05

Message:
Logged In: YES 
user_id=1188172

Yep, after "make distclean" and even a fresh checkout the
problem is there. I configured with and without
"--with-pydebug".

Okay, Michael, you are rehabilitated: I restored
floatobject.c to the state before your patch, and it behaves
the same.

----------------------------------------------------------------------

Comment By: Michael Hudson (mwh)
Date: 2005-12-10 14:42

Message:
Logged In: YES 
user_id=6656

Umpf.  I suppose the one remaining thing is you say you use "make clean".  This 
makes me very faintly suspicious that "make clean" doesn't really clean 
everything... could you maybe try a clean checkout?  Clutching at straws here, 
obviously.

----------------------------------------------------------------------

Comment By: Reinhold Birkenfeld (birkenfeld)
Date: 2005-12-10 14:35

Message:
Logged In: YES 
user_id=1188172

Yes, it is reproducable. If it's local to me, I don't know
what could be causing it.
I have a SVN tree without local changes on my local box and
my laptop, and both of them show the same bug.

I'll boot into some CD Linux later and see if the other
environment makes it vanish.

----------------------------------------------------------------------

Comment By: Michael Hudson (mwh)
Date: 2005-12-10 12:47

Message:
Logged In: YES 
user_id=6656

Also, is the problem reproducable on *your* system?  AFAIK, noone else has 
reproduced it, but I don't know how many people are actually running make test 
on development Python these days...

----------------------------------------------------------------------

Comment By: Michael Hudson (mwh)
Date: 2005-12-10 12:43

Message:
Logged In: YES 
user_id=6656

I don't think so, no.  For a start, I can't reproduce the problem (mind you, I only 
use a Mac regularly these days).

----------------------------------------------------------------------

Comment By: Reinhold Birkenfeld (birkenfeld)
Date: 2005-12-10 12:07

Message:
Logged In: YES 
user_id=1188172

Michael, you lately corrected a problem with floatobject.
Although the problem persists here, could it be caused by a
similar bug?

----------------------------------------------------------------------

Comment By: Armin Rigo (arigo)
Date: 2005-12-01 11:32

Message:
Logged In: YES 
user_id=4771

It's again a problem with floats.  For example, the constant 0.3 in rgb_to_yiq shows up as 0.0 in bad.pyc.  The constant 1.72986 shows up as 1.0.

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=105470&aid=1370322&group_id=5470