[Chicago] Windows API ctypes bug

[Randy Baxley]

Are the comparable cases both using the same version of windows.  There
were in '79 two versions of FORTRAN IV for IBM VS in which a program ran
fine in both except that in one it read from areas that were not memory
allocated to that program.  This could only be seen if one of the other 15
possible programs were noticed to have results in error or by doing a core
dump.


On Sat, Jul 12, 2014 at 6:38 PM, Aaron Brady <
ascription at yahoo.com.dmarc.invalid> wrote:

> Hi guys, Just thought I'd ask about a weird bug.
>
> Calling the Windows API through "ctypes".  The bug doesn't happen if:
> comment out another line, print before the call, call the ANSI version
> (instead of Unicode), run on the 32-bit build, run in 3.3, or run on a
> home-compiled build.  3.4-x64 and 3.4.1-x64 only.  But it does happen
> consistently (not even threading).  Program:
>
>
> http://home.comcast.net/~castironpi-misc/gui%2002c%20mdi%20-%20bug%20fork%20markup.html
>
> Error:
>
> Traceback (most recent call last):
>   File "D:\working\gui 02c mdi - bug fork.py", line 41, in <module>
>     mdi_child= user32.CreateWindowExW( wd.WS_EX_MDICHILD,
> mdichild_wnd_class_id, "child window", wd.WS_CHILD, 10, 50, 200, 200,
> mdi_client, 0, 0, 0 )
> OSError: exception: access violation reading 0x000000150000C268
>
> If you could help out I'd appreciate it, even brainstorming since I'm even
> running out of leads at this point, since it works fine in the source.  For
> one, can anyone else get a repro in 3.4, or could it be just my machine?
>
> _______________________________________________
> Chicago mailing list
> Chicago at python.org
> https://mail.python.org/mailman/listinfo/chicago
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/chicago/attachments/20140712/b42f024d/attachment.html>