From: Junio C Hamano <gitster@pobox.com>
Date: Tue, 11 Mar 2008 07:28:25 GMT
Subject: Re: [PATCH v4 2/5] git-submodule summary: show commit summary
Message-ID: <7vod9ld8iu.fsf@gitster.siamese.dyndns.org>
In-Reply-To: <1205171657-16216-2-git-send-email-pkufranky@gmail.com>

Ping Yin <pkufranky@gmail.com> writes:

> This patch does the hard work to show submodule commit summary.
> ...

I've spent considerable amount of time cleaning up your commit log
messages for the entire series from the last round when I queued them to
'pu', but you seem to have discarded all of them.  As I do not have time
to waste doing clean-ups only to have others to discard, and I do not have
infinite amount of time, I won't be replacing them with this series
tonight.

> Signed-off-by: Ping Yin <pkufranky@gmail.com>

And you did not describe anything about the improvements from the previous
round anywhere.  interdiff shows that this part is somewhat different:

> +		then
> +			case "$mod_dst" in
> +				160000)
> +				sha1_dst=$(GIT_DIR="$name/.git" git rev-parse HEAD)
> +				;;
> +				100644)
> +				sha1_dst=$(git hash-object $name)
> +				;;
> +			esac

Earlier, you were ignoring the one that is not checked out anymore, but
now you blindly assume "rev-parse HEAD" to give you a usable sha1_dst.
Why?

If a path that previously was a module is changed to a symlink, what
happens?