Re: [PATCH v17 05/14] ref-filter: introduce match_atom_name()

[Matthieu Moy <Matthieu.Moy@grenoble-inp.fr>]

Junio C Hamano <gitster@pobox.com> writes:

> Matthieu Moy <Matthieu.Moy@grenoble-inp.fr> writes:
>
>> OTOH, you are now accepting %(atom:) as a synonym to %(atom), and it's
>> not clear whether this is a deliberate decition.
>
> I would say so.  When the caller wants to reject %(atom:), the
> caller can tell it by checking val[0] == '\0' and reject that.
>
> So it is better if you did not do this:
>
>> 	if (!body[1]) {
>> 		/* "atom_name:" */
>> 		*val = NULL;
>> 		return 1;
>> 	}
>
> which robs that information from the caller.

OK. Just dropping this part lets the code fall back to

	/* "atom_name:... */
	*val = body + 1;
	return 1;

right below in my version. It also accepts it (return 1) but lets val
point to an empty string. Makes sense.

And indeed, without this, my code looks a lot like Karthik's one, just
dropping the "|| !body[1]" part in a condition.

In any case, I'd like to see "atom_name:" explicitly mentionned
somewhere in a comment, if only to make it clear that what is done with
it is deliberate (e.g. avoid having someone not following this
conversation later considering this %(atom:) thing to be a bug and try
to fix it).

-- 
Matthieu Moy
http://www-verimag.imag.fr/~moy/