Re: [PATCH 06/19] reset.c: remove unnecessary variable 'i'

From: Martin von Zweigbergk <martinvonz@gmail.com>
Date: Jan 10, 2013, 08:41 UTC
Message-ID: <CANiSa6jJcYRUyswSktq86pWttxBjZedZcPheRgPCf+EubC_kLg@mail.gmail.com>
In-Reply-To: <7vpq1e5ent.fsf@alter.siamese.dyndns.org>

On Wed, Jan 9, 2013 at 11:39 AM, Junio C Hamano <gitster@pobox.com> wrote:

Show 21 quoted lines

> Martin von Zweigbergk <martinvonz@gmail.com> writes:
>
>> Throughout most of parse_args(), the variable 'i' remains at 0. In the
>> remaining few cases, we can do pointer arithmentic on argv itself
>> instead.
>> ---
>> This is clearly mostly a matter of taste. The remainder of the series
>> does not depend on it in any way.
>
> I agree that it indeed is a matter of taste between
>
>  (1) look at av[i], check with (i < ac) for the end, and increment i to
>      iterate over the arguments; and
>
>  (2) look at av[0], check with (0 < ac) for the end, and increment
>      av and decrement ac at the same time to iterate over the
>      arguments.
>
> When (ac, av) appear as a pair, however, adjusting only av without
> adjusting ac is asking for future trouble.  It violates a common
> expectation that av[ac] points at the NULL at the end of the list.

Good points.

> If a code chooses to use !av[0] as the terminating condition and
> never looks at ac, then incrementing only av is fine, but in such a
> case, the function probably should lose ac altogether.

Makes sense. I've picked this style for now (i.e. dropped both 'i' and 'argc'). I was surprised by the style that referred to the variable in many places where it was know to be 0, but I'm no experienced C programmer, so if that's a common practice when it comes to argument parsing, I'm also happy to drop the patch. Let me know what you prefer.