Re: What's cooking in git.git (Oct 2023, #01; Mon, 2)

From: Junio C Hamano <gitster@pobox.com>
Date: Oct 5, 2023, 21:47 UTC
Message-ID: <xmqqh6n4eo7n.fsf@gitster.g>
In-Reply-To: <87lecgeqfu.fsf@osv.gnss.ru>

Sergey Organov <sorganov@gmail.com> writes:

> Overall, as an example, I'd understand if you had deflected the patch
> with "let's rather use -d for '--decorate=short', or '--date=relative'",
> or something like that, but you don't, leaving me uncertain about your
> actual worries and intentions.

Oh, I would be very much more sympathetic if somebody wanted to make a short-and-sweet single-letter option to stand for "--first-parent -p", if they come with the "first-parent chain is special---it is the trunk history of the development" world view. And the resulting behaviour would be "give me the diffs" in their world view, so I would understand if they wanted to use "-d" for such an operation.

However, to folks who do not subscribe to "the first parent chain is the trunk history" world view, "give me the diffs" is not an explanation of the resulting behaviour, because in "-d" there is no trace of hint that it is also about first-parent traversal.

So "-d" may not be a perfect fit for it, either. But at least it is based on a more consistent world view, I would think, than "--diff-merges=1 -p", whose behaviour becomes unexplainable when it hits "reverse" merges in a world where the first parent chain is not necessarily the trunk.

Anyway, I've tentatively queued the "--dd" round. Naming is hard, I cannot tell what "dd" stards for, and I suspect no user can X-<.

Thanks.