Pratyush Yadav <me@xxxxxxxxxxxxxxxxx> writes:
> I'm afraid I don't follow what exactly this would do, and how it would
> help differentiate between the "what the amend does" and "what the
> amended commit does".
The resulting history would be
O---A
\
B
where
O = HEAD^ = HEAD@{1}^
A = HEAD@{1} - HEAD before the amend
B = HEAD - result of the amend
I wonder if
git diff -c B O A
(with possibly different permutations of three revisions) is a
reasonable way to show what the final state is and where it differs
from the previous one (i.e. HEAD@{1}) and the original one
(i.e. HEAD^) in the combined diff format.
>> 2. would it make sense to show the differences between
>> HEAD^..HEAD@{1} and between HEAD^..HEAD using the range-diff
>> machinery.
>
> I considered using range-diff, but didn't go with it because of my
> personal dislike for range-diff.
For a single-commit amend, the normal diff between HEAD@{1} and HEAD
would be far easier to read than such a range-diff, I would think.
