Re: [PATCH 2/3] add-patch: add diff.context command line overrides

From: Junio C Hamano <gitster@pobox.com>
Date: May 7, 2025, 20:25 UTC
Message-ID: <xmqqo6w4b223.fsf@gitster.g>
In-Reply-To: <CAP9jKjGEDY78MSrZVS2FxAoikPZpGazqem2XHDVmXqMjcox6yg@mail.gmail.com>

Leon Michalak <leonmichalak6@gmail.com> writes:

Show 15 quoted lines

> I think I may be misunderstanding, so I'll elaborate on what I
> personally intended my documentation to mean.
>
> When I put in the documentation "implies --interactive/--patch" it
> reads to me as "this assumes you are also using either --interactive
> or --patch and that if you don't specify one or the other it will do
> nothing or possibly error (which is what I chose in the end, based on
> the initial discussion in a separate thread)". I didn't think it would
> read as "you must have both settings" or "if you don't specify these
> the command will assume it as if you had and effectively act as if you
> had".
>
> I'm not sure if the wording was confusing or it generally has
> different meanings to others so perhaps that might clarify at least
> what I intended :)

We use "imply" to mean quite a different thing from what you said in the above explanation, and that is where my reaction came from.

We say "option A implies option B" only when the command behaves as if the user gave option B when only option A is given (and without giving B). For example, "git commit --help" has

        `--short`::
                When doing a dry-run, give the output in the short-format. See
                linkgit:git-status[1] for details. Implies `--dry-run`.

The `--short` option does not make any sense when you are actually creating a commit; if you say "git commit --short", the command behaves as if you said "git commit --dry-run --short".

If something works only under certain condition, it seems we tend to say "this only works when/with ...", so in this case, we would have said "The -U<n> option is only effective for --interactive or --patch mode of this command" to express what you meant to convey, I think.

Thanks.