btw, just something for the record:

I think the example given in msg109117 above is wrong:

Depending on the read size it will produce different results, given how split() works:

Imagine a byte sequence:
>>> b"\0foo\0barbaz\0\0abcd".split(b"\0")
[b'', b'foo', b'barbaz', b'', b'abcd']


Now the same sequence, however with a different read size (here a shorter one):
>>> b"\0foo\0barbaz\0".split(b"\0")
[b'', b'foo', b'barbaz', b'']
>>> b"\0abcd".split(b"\0")
[b'', b'abcd']

=> it's the same bytes, but in the 2nd case one get's an extra b''.