RE: [RFC PATCH 01/35] merge: improve fatal fast-forward message

From: Felipe Contreras <felipe.contreras@gmail.com>
Date: Jul 6, 2021, 22:14 UTC
Message-ID: <60e4d5b03dbe8_1c4281208b1@natae.notmuch>
In-Reply-To: <04df01d772ad$c9ac7430$5d055c90$@nexbridge.com>

Randall S. Becker wrote:

> On July 6, 2021 5:12 PM, Felipe Contreras wrote:

Show 7 quoted lines

> >But this is not a `git merge`, this is a `git merge --ff-only`; they
> >are different operations. There *never* is a need for `--abort` with
> >`git merge --ff-only`.
> 
> Well, you know that and I know that, but having to explain this to
> every new git user who will operationally use git merge --ff-only
> within hours or days of their first clone is a different matter.

New users don't do `git merge --ff-only`, they very likely don't do `git merge` either, nor do they do `git pull --ff-only`. This is obvious from the countless threads where users do `git pull` without knowing that they will create a true merge by mistake.

Even less likely is that they will know about `git merge --abort`, and even less that they will know both `git merge --ff-only` and `git merge --abort`.

Even even less likely that they will mistakenly expect `git merge --ff-only` to be left inside a temporary stated to be resolved by `git merge --abort`.

But as Junio said, *if* such a hypothetical user exists, we don't want to keep misleading them into thinking that a true merge was somehow aborted, because it wasn't.

It's much better to simply let go the current error message.

Cheers.

-- 
Felipe Contreras