Win32All Crashes Python 2.3?

Tim Daneliuk tundra at tundraware.com
Tue Aug 5 23:40:07 EDT 2003


Roger Upole wrote:

> I get the same thing on Win2k, Python 2.3 anytime a unicode string is
> passed to os.lstat (without invoking any win32all modules)

This may well be my problem as well since the routine in question
sometimes gets passed a unicode string, and sometimes not ...
(I think) ...

>    ...
> Looks like the problem is in posixmodule.c
>   PyUnicodeObject *po;
>   if (PyArg_ParseTuple(args, wformat, &po)) {
> wformat at run time is "u:lstat", which should have
> a unicode pointer for the receiving argument, not a PyUnicodeObject.
> os.stat (which works for a unicode string) uses uppercase U in the format.
> 
>       Roger
> 
> 
> "Tim Daneliuk" <tundra at tundraware.com> wrote in message
> news:jchngb.pr.ln at boundary.tundraware.com...
> 
>>Some time ago I wrote a portable file system browser, twander, that,
>>when runnning on Win32, will make use of the Win32All extensions if they
>>are present on the target system.
> 
> (http://www.tundraware.com/Software/twander)
> 
>>twander primarily uses Win32All to get file attribute and ownership
>>information back in Win32-specific ways each time a new directory is
>>entered. (Ordinarily (if Win32All is not installed), it just displays an
>>'ls -al' style output and dummies up ownership and group membership.)
>>
>>Now the problem - twander has worked fine with previous versions of
>>Python/Win32All up to and including 2.2.x.  It also works fine
>>with 2.3.  HOWEVER, if I install any of the 2.3 relevant versions
>>of Win32All, when I change to a new directory, Python crashes.
>>No traceback information is given so I cannot determine exactly
>>why this is happening.  What is strange is that the program
>>starts OK, which means it is using Win32All calls to determine
>>file system information for the initial directory.  It is only
>>upon changing directories that Python gets blown out of the water.
>>
>>While it is certainly possible that this is a bug in my code, I
>>cannot imagine what it might be.  The program has worked flawlessly
>>on a wide variety of Unix/FreeBSD/Win32 systems with no reports
>>of this sort of thing.
>>
>>Anyone have any ideas what could cause this?  Is this a reported
>>Win32All bug?
>>
>>TIA,
>>--------------------------------------------------------------------------
> 
> --
> 
>>Tim Daneliuk     tundra at tundraware.com
>>PGP Key:         http://www.tundraware.com/PGP/
>>
> 
> 
> 


-- 
----------------------------------------------------------------------------
Tim Daneliuk     tundra at tundraware.com
PGP Key:         http://www.tundraware.com/PGP/





More information about the Python-list mailing list