From: Junio C Hamano <gitster@pobox.com>
Date: Mon, 17 Mar 2014 23:07:00 GMT
Subject: Re: [PATCH 1/1] general style: replaces memcmp() with starts_with()
Message-ID: <xmqqd2hka0h7.fsf@gitster.dls.corp.google.com>
In-Reply-To: <20140317230139.GB19578@sigill.intra.peff.net>

Jeff King <peff@peff.net> writes:

> On Mon, Mar 17, 2014 at 03:52:51PM -0700, Junio C Hamano wrote:
>
>> > diff --git a/builtin/for-each-ref.c b/builtin/for-each-ref.c
>> > index 3e1d5c3..4135980 100644
>> > --- a/builtin/for-each-ref.c
>> > +++ b/builtin/for-each-ref.c
>> > @@ -193,7 +193,7 @@ static int verify_format(const char *format)
>> >  		at = parse_atom(sp + 2, ep);
>> >  		cp = ep + 1;
>> >  
>> > -		if (!memcmp(used_atom[at], "color:", 6))
>> > +		if (starts_with(used_atom[at], "color:"))
>> >  			need_color_reset_at_eol = !!strcmp(used_atom[at], color_reset);
>> >  	}
>> >  	return 0;
>> 
>> Good.
>
> Actually, I found this one confusing. We are looking for "color:", but
> if we find it, we _don't_ skip past and look at what comes after.
> Instead, we compare the whole string. Which works because color_reset
> actually contains "color:reset", and we end up just re-comparing the
> first bit of the string. So the memcmp here is redundant, and this can
> simply become:
>
>   need_color_reset_at_eol = !!strcmp(used_atom[at], color_reset);
>
> Or am I missing something?

What if used_atom[at] is not related to color at all?  We do not
want to touch the variable.