Re: [PATCH v2] merge-recursive: honor diff.algorithm

From: Antonin Delpeuch <antonin@delpeuch.eu>
Date: Jul 13, 2024, 16:50 UTC
Message-ID: <9ff7c0e6-af29-48a2-bec7-be4554681671@delpeuch.eu>
In-Reply-To: <xmqqmsmpw2mp.fsf@gitster.g>

On 10/07/2024 19:58, Junio C Hamano wrote:

Show 7 quoted lines

> Since be733e12 (Merge branch 'en/merge-tree', 2022-07-14),
> merge-tree is no longer an interrogator but works as an manipulator.
> As it is meant to be used as a building block that gives a reliable
> and repeatable output, I am tempted to say it should be categorized
> as a plumbing, but second opinions do count.  Elijah Cc'ed as it was
> his "fault" to add "--write-tree" mode to the command and forgetting
> to update command-list.txt ;-)

I'm happy to change it to use the "basic" config if you prefer. I have to admit I don't have a good overview of what's porcelain or plumbing.

> Isn't log-tree shared with things like "git diff-tree" porcelain?

I'm happy to also change this, but looking at the call hierarchy we seem to have a complicated entanglement of porcelain and plumbing commands here, so to separate them is probably more involved.

> Can't we save the scarce resource that is test number
> and make these not about "I test cherry-pick" and "I test merge"?

Yes I agree, let me submit a new patch which does that already. I wasn't sure if it was worth including tests for different commands anyway.

Antonin