Re: 'git diff-index' doesn't honor the 'diff.algorithm' variable

From: Dmitry Gutov <dgutov@yandex.ru>
Date: May 15, 2016, 20:36 UTC
Message-ID: <c81e4b71-29f2-6a11-7bf4-769013d34a6d@yandex.ru>
In-Reply-To: <xmqqwpmv5fi3.fsf@gitster.mtv.corp.google.com>

On 05/15/2016 09:43 PM, Junio C Hamano wrote:

> I think the paragraph is shared among the "diff" family of
> commands both plumbing and Porcelain, so I'd say "patches welcome"
> at this point ;-).

I think I've done my part here. It's not like this is a feature request.

> The script was an illustration of the logic--I am sure elisp is much
> core capable scripting environment than POSIX shell.  Perhaps (setq
> vc-diff-git-diff-use-histogram t) in ~/.emacs is not too bad ;-)

Yes, doing it via a user option is already possible in Emacs. I was concerned that the user has to configure it twice (once for console, two for Emacs), but if you think it's fine, let's leave it at that.

Show 9 quoted lines

> The Porcelain "git diff" command is not bound by any promise of
> stable output and reserves the right to change the default to better
> support human users.  I think the upcoming version of Git turns the
> diff.renames setting on by default, for example.  We might even add
> a side-by-side diff and make it the default someday.  You do not
> want to be reading these "fancy" output, and you cannot keep
> updating the invocation of "git diff" by vc-diff with unbounded
> number options, e.g. --no-side-by-side, that will be added to defeat
> configuration variables that will be invented in the future.

Fair enough.

Thanks, Dmitry.