Oki, it seems I've found. To directly use a bytearray buffer from ctypes, you must first create a compatible ctypes type (I.E, a char array of same size), and only then instanciate this new type with newtype.from_buffer (bytearray_object).
The little danger is : you must NOT change the size of bytearray as long as the ctypes array targets its inner buffer, else of cousre memory reallocations occurs, and your ctypes array points to invalid memory (awaiting a pretty segfault). But modifying chars or characters ranges from the buffer, as well via the bytearray object than via the ctypes array, is fine. I don't think problems can occur if both sides are accessed simultaneously, even though ctypes operations may release the GIL while they execute... the only race condition is the access of a memory segment, isn't it ? No crash shall occur with this... IDLE 2.6.4 >>> import ctypes >>> a = bytearray(3) >>> mytype = ctypes.c_char * 3 >>> h = mytype.from_buffer(a) >>> h <__main__.c_char_Array_3 object at 0x02B06350> >>> h[:] = "abc" >>> a bytearray(b'abc') >>> h[0] = "k" >>> a bytearray(b'kbc') -- http://mail.python.org/mailman/listinfo/python-list
