Is this a bug in Python2.0 or yet another feature of Windows?

[Michael Hudson]

"Fredrik Lundh" <effbot at telia.com> writes:

> "June Kim" wrote:
> > With Python 2.0 Final Release,
> >
> > >>> from array import array
> > >>> z=array('c','hello world')
> > >>> `z`  #Note that these are back-quotes
> >
> > It returns a nice "python20.dll" error, whereas run on python 1.6,
> > it returns what we expect, the string form of the array.
> 
> explicit calls to repr and str crashes for me with a locally built
> 2.0 interpreter (for some reason, it can still *print* this array).

Oh, that's easy: array_repr and array_print are different functions.

The culprit *seems* to be this line:

                v = array_tostring(a, (PyObject *)NULL);

in array_repr, and this checkin:

--------------
revision 2.51
date: 2000/07/31 21:57:30;  author: nowonder;  state: Exp;  lines: +39 -34

replaced PyArgs_Parse by PyArgs_ParseTuple

changed error messages for extend method from "append" to "extend"
--------------

Because PyArgs_Parse can cope with a NULL args, and PyArgs_ParseTuple
can't.  Peter changed array_print, but not array_repr...

Also, we have this at present:

>>> sys.getrefcount(())
63
>>> array.array("f",[1,2])
array('f', [1.0, 2.0])
>>> sys.getrefcount(())
64
>>> array.array("f",[1,2])
array('f', [1.0, 2.0])
>>> sys.getrefcount(())
65

i.e. the current code leaks references to the empty tuple.  Not
serious, but not pretty either.

> have you reported the bug?  here's where to do it:
> 
>     http://sourceforge.net/bugs/?group_id=5470

Patch at

  https://sourceforge.net/patch/index.php?func=detailpatch&patch_id=102068&group_id=5470

hopes to fix all of the above.

brown-paper-bag-ly y'rs,
M.

-- 
  > Why are we talking about bricks and concrete in a lisp newsgroup?
  After long experiment it was found preferable to talking about why
  Lisp is slower than C++...
                        -- Duane Rettig & Tim Bradshaw, comp.lang.lisp