Re: [RFC][PATCH V1] diff-tree: integrate with sparse index

From: Junio C Hamano <gitster@pobox.com>
Date: May 16, 2023, 23:14 UTC
Message-ID: <xmqqsfbvswci.fsf@gitster.g>
In-Reply-To: <fa24482c-7c48-9b7f-5d97-3dbf9822728c@github.com>

Victoria Dye <vdye@github.com> writes:

> Longer version:

Thanks, as usual, for a great review. A lot of the stuff you wrote should inspire and result in an improved log message that explains why this change is sufficient to teach diff-tree to take advantage of the sparse index.

Show 5 quoted lines

> However, 'DIFF_SETUP_USE_SIZE_CACHE' seems to be unused as of 6e0b8ed6d35
> (diff.c: do not use a separate "size cache"., 2007-05-07) and nothing about
> 'detect_rename' otherwise indicates index usage, so AFAICT that whole
> condition can be dropped (along with DIFF_SETUP_USE_SIZE_CACHE,
> DIFF_SETUP_REVERSE, and diff_options.setup).

True. The size cache does not exist anymore. 6b5ee137 (Diff clean-up., 2005-09-21) restructured the command line option parsing quite a bit, and we lost DIFF_SETUP_REVERSE, which is a bit that gets OR'ed in to a file-scope diff_setup_opt static of each of the command in the diff family. The bit and the diff_setup_opt variable got replaced with members of "struct diff_options", and I should have removed the macro at the same time.

> Note that, if you want to make
> that change in this series, it should be done in a separate patch _before_
> this one (since dropping the deprecated setup infrastructure isn't really
> part of the sparse index integration).

Or after this one, perhaps? I agree that the clean-up opportunity you found is very much unrelated to the work to teach diff-tree to take advantage of the sparse index.

THanks.