Just, I'm guessing this has to do with the import
modifications you made, although I really don't know. 
I just started getting a crash which originates in
_PyImport_Fini().

When I run:
  ./python -E -tt ./Lib/test/regrtest.py test_logging

I get a core dump.  Here's the stack trace from a debug
build.  The same problem occurs in a normal (non-debug)
build.

(gdb) bt
#0  0x08080e09 in _Py_ForgetReference (op=0x407887c0)
at Objects/object.c:1949
#1  0x08080e79 in _Py_Dealloc (op=0x407887c0) at
Objects/object.c:1970
#2  0x08079e7e in dict_dealloc (mp=0x4026b214) at
Objects/dictobject.c:690
#3  0x08080e87 in _Py_Dealloc (op=0x4026b214) at
Objects/object.c:1971
#4  0x08079e7e in dict_dealloc (mp=0x401bea34) at
Objects/dictobject.c:690
#5  0x08080e87 in _Py_Dealloc (op=0x401bea34) at
Objects/object.c:1971
#6  0x080d5d66 in _PyImport_Fini () at Python/import.c:224
#7  0x080dff7a in Py_Finalize () at Python/pythonrun.c:233
#8  0x08055146 in Py_Main (argc=4, argv=0xbffff814) at
Modules/main.c:465
#9  0x08054884 in main (argc=4, argv=0xbffff814) at
Modules/python.c:23

Any ideas?

Logged In: YES 
user_id=92689

Hm, I wish I could reproduce this on my OSX box, but I can't: the loggin test passes just fine. I also doubt it has to do with my mods, as _PyImport_Fini doesn't touch anything that has anything to do with import hooks directly.  Of course it's still not impossible I caused this, but right now I don't see how it could.

Logged In: YES 
user_id=45365

Could this be a duplicate of #675341? You can check this by updating: Martin just fixed that bug, which was in a new unicode codec.

Logged In: YES 
user_id=33168

This works now, so I'm closing.  Since the problem occurred
after iconv, it was most likely related to that and Martin
fixed it.