Logged In: YES 
user_id=89016

OK, I've checked in the following:

Lib/codecs.py 1.44
Lib/test/test_codecs.py 1.23
Lib/codecs.py 1.35.2.7
Lib/test/test_codecs.py 1.15.2.5

with the following changes as suggested by glchapman:
If a chunk read has a trailing "\r", read one more character
even if the user has passed a size parameter. Remove the
atcr logic. This should fix most of the problems. There are
three open issues:

1) How do we handle the problem of a truncated line, if the
data comes from the charbuffer instead of being read from
the stream?

2) How do we handle error reporting? The buffering code
might read more data than the current line. If this data has
a decoding error the caller might report a wrong line
number. (See bug #1178484)

3) Fixing the tokenizer.