From: Junio C Hamano <junkio@cox.net>
Date: Thu, 22 Feb 2007 10:36:13 GMT
Subject: Re: [PATCH] Teach git-diff{,-files} the new option `--no-index`
Message-ID: <7vzm76fpia.fsf@assigned-by-dhcp.cox.net>
In-Reply-To: <Pine.LNX.4.63.0702201944340.22628@wbgn013.biozentrum.uni-wuerzburg.de>

Johannes Schindelin <Johannes.Schindelin@gmx.de> writes:

> 	Is this better than diff2 or what?

Probably.

> +int run_diff_files(struct rev_info *revs, int argc, const char **argv)
>  {
>  	int entries, i;
> +	int diff_unmerged_stage;
> +	int silent_on_removed;
> +
> +	if (handle_diff_files_args(revs, argc, argv, &silent_on_removed))
> +		return -1;
> +
> +	diff_unmerged_stage = revs->max_count;
> +
> +	if (revs->max_count == -2) {
> +		if (revs->diffopt.nr_paths != 2)
> +			return error("need two files/directories with --no-index");
> +		queue_diff(&revs->diffopt, revs->diffopt.paths[0],
> +				revs->diffopt.paths[1]);
> +		diffcore_std(&revs->diffopt);
> +		diff_flush(&revs->diffopt);
> +		return 0;
> +	}

However, I do not like this part.

I do not mind the "max_count == -2 is some magic" hack too much,
but I do not think it belongs to a generic service function
whose purpose is to run "diff-files" aka "diff between index and
working tree".  run_diff_files() is used by other applications
(most notably wt-status) for example, and I expect we would need
that when we rewrite git-commit itself in C.