I'm using wxPython version 2.2.0 on M$ Windows (NT4, if that matters).
When I try to invoke the SetEnds method on a wxLineShape object,
the result is an error box saying "OleMainThreadWndName: python.exe -
Application Error", complaining about being unable to read memory at
address 0x00000010.
A brief visit to the debugger indicates that it's happening in this
function:
00FC583B push ebp
00FC583C mov ebp,esp
00FC583E mov eax,dword ptr [ecx+17Ch]
00FC5844 fld qword ptr [ebp+8]
00FC5847 mov edx,dword ptr [eax+10h] <--- HERE.
00FC584A mov eax,dword ptr [eax+14h]
00FC584D mov edx,dword ptr [edx+8]
00FC5850 mov eax,dword ptr [eax+8]
00FC5853 fstp qword ptr [edx]
00FC5855 fld qword ptr [ebp+10h]
00FC5858 fstp qword ptr [edx+8]
00FC585B fld qword ptr [ebp+18h]
00FC585E fstp qword ptr [eax]
00FC5860 fld qword ptr [ebp+20h]
00FC5863 fstp qword ptr [eax+8]
00FC5866 fld qword ptr [ebp+8]
00FC5869 fadd qword ptr [ebp+18h]
00FC586C fmul qword ptr ds:[0FDAA40h]
00FC5872 fstp qword ptr [ecx+20h]
00FC5875 fld qword ptr [ebp+10h]
00FC5878 fadd qword ptr [ebp+20h]
00FC587B fmul qword ptr ds:[0FDAA40h]
00FC5881 fstp qword ptr [ecx+28h]
00FC5884 pop ebp
00FC5885 ret 20h
so presumably this is an attempt to access something via a
null pointer. I don't know enough x86 assembler to be very
sure what the function's doing; coordinate transformation,
maybe?
The call stack contains two entries in OGLC.DLL followed
by a number in Python itself.
I don't get behaviour like this from (e.g.) a line whose
endpoints are anchored to other shapes instead of being
specified explicitly.
Is this a known bug? Or is my setup broken? If it's a bug,
is there a fix?
I'm not subscribed to this list. I'd be grateful if any replies
could be cc'ed to me at gmccaughan@synaptics-uk.com .
Thanks muchly.
···
--
g