Re: [PATCH] setup: suppress implicit "." work-tree for bare repos

From: Junio C Hamano <gitster@pobox.com>
Date: Mar 8, 2013, 07:54 UTC
Message-ID: <7vboaujphx.fsf@alter.siamese.dyndns.org>
In-Reply-To: <20130308071554.GB24429@sigill.intra.peff.net>

Jeff King <peff@peff.net> writes:

Show 12 quoted lines

> diff --git a/cache.h b/cache.h
> index e493563..070169a 100644
> --- a/cache.h
> +++ b/cache.h
> @@ -344,6 +344,7 @@ static inline enum object_type object_type(unsigned int mode)
>  #define GIT_DIR_ENVIRONMENT "GIT_DIR"
>  #define GIT_NAMESPACE_ENVIRONMENT "GIT_NAMESPACE"
>  #define GIT_WORK_TREE_ENVIRONMENT "GIT_WORK_TREE"
> +#define GIT_IMPLICIT_WORK_TREE_ENVIRONMENT "GIT_IMPLICIT_WORK_TREE"
>  #define DEFAULT_GIT_DIR_ENVIRONMENT ".git"
>  #define DB_ENVIRONMENT "GIT_OBJECT_DIRECTORY"
>  #define INDEX_ENVIRONMENT "GIT_INDEX_FILE"

Not adding any user documentation is fine (you explained why in the log message), but I would really prefer to have some in-code comment to clarify its meaning. Is it "Please do use implicit work tree" boolean? Is it "This is the path to the work tree we have already figured out" string? Is it something else? What is it used for, who sets it, what other codepath that will be invented in the future need to be careful to set it (or unset it) and how does one who writes that new codepath decides that he needs to do so (or shouldn't)?

I would know *today* that it is a bool to affect us, after having discovered that we are in bare and we have set GIT_DIR (so if the end user already had GIT_DIR, we shouldn't set it ourselves), and also our child processes, but I am not confident that I will remember this thread 6 months down the road.