Re: [PATCH v1 2/7] mv: add documentation for check_dir_in_index()

From: Derrick Stolee <derrickstolee@github.com>
Date: Jul 19, 2022, 17:43 UTC
Message-ID: <228ad533-477c-f16e-220d-61e52d9aee26@github.com>
In-Reply-To: <20220719132809.409247-3-shaoxuan.yuan02@gmail.com>

On 7/19/2022 9:28 AM, Shaoxuan Yuan wrote:

> Using check_dir_in_index without checking if the directory is on-disk
> could get a false positive for partially sparsified directory.

It can be helpful to point out that this is a relatively new method, introduced in b91a2b6594 (mv: add check_dir_in_index() and solve general dir check issue, 2022-06-30).

Show 5 quoted lines

> + *
> + * Note: *always* check the directory is not on-disk before this function
> + * (i.e. using lstat());
> + * otherwise it may return a false positive for a partially sparsified
> + * directory.

I'm not sure what you mean by a "false positive" in this case. The directory exists in the index, which is what the method is defined as checking. This does not say anything about the worktree.

Perhaps that's the real problem? Someone might interpret this as meaning the directory does not exist in the worktree? That would mean that this doc update needs to be changed significantly to say "Note that this does not imply anything about the state of the worktree" or something.

But I think I'd rather just see this patch be dropped, unless I am missing something important.

Thanks, -Stolee