Re: [PATCH] clone: error specifically with --local and symlinked objects

From: Taylor Blau <me@ttaylorr.com>
Date: Apr 6, 2023, 21:27 UTC
Message-ID: <ZC85ODogr79t0ZXZ@nand.local>
In-Reply-To: <xmqq7curxk22.fsf@gitster.g>

On Tue, Apr 04, 2023 at 05:13:41PM -0700, Junio C Hamano wrote:

Show 6 quoted lines

>  * if lstat() succeeds, but the path is *not* a directory; errno is
>    explicitly set to ENOTDIR.  Unfortunately, if lstat(2) failed
>    with ENOTDIR (e.g. dir_iterator_begin() gets called with a path
>    whose leading component is not a directory), the caller will also
>    see ENOTDIR, but the distinction may not matter in practice.  I
>    haven't thought things through.

Yeah. The real tragedy here is that there's no way to signal that a file *is* a symbolic link and represent that as an error code in errno.

So I think the best that we can do here is to continue to report ENOTDIR, and have the caller figure out what to do with it.

Thanks, Taylor