From: Jeff King <peff@peff.net>
Date: Thu, 15 Feb 2018 05:49:45 GMT
Subject: Re: [PATCH v3 14/23] ref_filter: add is_atom_used function
Message-ID: <20180215054945.GL18780@sigill.intra.peff.net>
In-Reply-To: <01020161890f4357-05e636bc-8b44-425b-a252-ff2341f91cdd-000000@eu-west-1.amazonses.com>

On Mon, Feb 12, 2018 at 08:08:54AM +0000, Olga Telezhnaya wrote:

> Delete all items related to split_on_whitespace from ref-filter
> and add new function for handling the logic.
> Now cat-file could invoke that function to implementing its logic.

OK, this is a good direction. I think in a more compact series we'd
avoid moving the split-on-whitespace bits over to ref-filter in the
first place, and have two commits:

 - one early in the series adding is_atom_used()

 - one late in the series switching cat-file over to is_atom_used() as
   part of the conversion to ref-filter

> diff --git a/builtin/cat-file.c b/builtin/cat-file.c
> index 6db57e3533806..3a49b55a1cc2e 100644
> --- a/builtin/cat-file.c
> +++ b/builtin/cat-file.c
> @@ -382,8 +382,7 @@ static int batch_objects(struct batch_options *opt)
>  {
>  	struct strbuf buf = STRBUF_INIT;
>  	struct expand_data data;
> -	int save_warning;
> -	int retval = 0;
> +	int save_warning, is_rest, retval = 0;

Try to avoid reformatting existing code that you're not otherwise
touching, as it makes the diff noisier. Just adding "int is_rest" would
make this easier to review.

I also think the variable name should probably still be
"split_on_whitespace". It's set based on whether we saw a "%(rest)"
atom, but ultimately we'll use it to decide whether to split.

-Peff