[m-rev.] for post-commit review: simplify success/failure testing

[Zoltan Somogyi]

On 2023-12-11 13:15 +11:00 AEDT, "Peter Wang" <novalazy at gmail.com> wrote:
> Hmm, are you using the 'patience' or 'histogram' git diff algorithms?

I remember that I used patience for some of the diffs I sent over the weekend,
but not which diffs.

> It turns out that the (default) 'myers' or 'minimal' algorithms happen
> to produce a much more reasonable diff for this part of the commit.

I usually use the default diff algorithm, and switch to patience if I don't like
the result, usually because it resynchronizes on something inappopriate,
and screws up the rest of the diff. I then usually check that the switch to patience
has improved the part of the diff in question. For big diffs, I don't always check
what it has done to the rest of the diff, because I have not found patience
to generate output that is *significantly* worse than the default algorithm.
If patience makes the output slightly worse in one part while significantly
improving the output in another part, I will use it.

Unfortunately, I know of no way to specify different diff algorithms for
diffing different files, much less for different parts of the same file.

Thanks for the reviews.

Zoltan.