git-diff bug?

From: Eli Barzilay <eli@barzilay.org>
Date: Nov 2, 2020, 06:53 UTC
Message-ID: <CALO-guviA4xKjUi0HfA+RLkTPPaQw7KArj__A9fKz0oP3m5MGw@mail.gmail.com>

Is the following a bug?

    $ printf "aaa\nbbb\nccc\n\n" > 1
    $ printf "aaa\nbbb\n\nccc\n" > 2
    $ git diff --ignore-blank-lines 1 2

This shows a weird output, as if `ccc` was removed and then re-added. Flipping the 1 & 2 names makes it show no difference at all. I tried a bunch of variants, including --minimal, and the four algorithms, and all show the same results. (Similar brokenness happens with an empty line at the beginning on one side and after the first line on the other.)

I'm really not sure that the following is a bug, because I see the same behavior from `diff` (which is what made me try git-diff, hoping that it would be more consistent). (But I can't think of any rational that would make it not a bug.)

-- 
                   ((x=>x(x))(x=>x(x)))                  Eli Barzilay:
                   http://barzilay.org/                  Maze is Life!