Always parse diffs as binary.
Review Request #6926 — Created Feb. 9, 2015 and submitted — Latest diff uploaded
In our diff parsing code, we would convert to unicode first, and then parse in
unicode mode. This wasn't ideal, because things likesplitlines()behave
differently for unicode, and once we add support for diffX, we'll need to parse
in binary anyway (since each file will have different encodings).
- Ran unit tests.
- Uploaded a diff with some unicode characters.