From: Junio C Hamano <gitster@pobox.com>
Date: Fri, 13 Aug 2010 00:49:28 GMT
Subject: Re: [PATCH v4 2/9] stash: extract stash-like check into its own function
Message-ID: <7v8w4bnxpj.fsf@alter.siamese.dyndns.org>
In-Reply-To: <1281398688-21008-3-git-send-email-jon.seymour@gmail.com>

Jon Seymour <jon.seymour@gmail.com> writes:

> A similar code block is used in three places; this change
> factors the stash-like check into a separate function so that
> it can be used in these and other places.

"git stash show" used to be able to show a single-parent commit, and we
are forbidding that "feature".  I'd actually call that a bugfix though ;-)

> +# if this function returns, then:
> +#   s is set to the stash commit
> +#   w_commit is set to the commit containing the working tree
> +#   b_commit is set to the base commit
> +#   i_commit is set to the commit containing the index tree
> +# otherwise:
> +#   the function exits via die
> +#
> +assert_stash_like() {
> +	# stash records the work tree, and is a merge between the
> +	# base commit (first parent) and the index tree (second parent).
> +	s=$(git rev-parse --revs-only --quiet --verify --default $ref_stash "$@") &&
> +	w_commit=$(git rev-parse --quiet --verify "$s:") &&
> +	b_commit=$(git rev-parse --quiet --verify "$s^1:") &&
> +	i_commit=$(git rev-parse --quiet --verify "$s^2:") ||
> +		die "$*: no valid stashed state found"

Don't these "rev-parse" calls return tree objects?  It is too confusing to
call them x_commit if that is the case.