Re: [RFA] Add support for __VA_OPT__

[Pedro Alves <palves at redhat dot com>]

On 09/23/2017 04:03 AM, Tom Tromey wrote:
>>>>>> "Pedro" == Pedro Alves <palves@redhat.com> writes:
> 
>>> But I wonder if gdb should just error() on the invalid ones.
>>> My first thought was no, why make life harder -- but at the same time,
>>> the invalid cases really aren't that useful either.
> 
> Pedro> Yeah, error might be better - e.g., for someone trying
> Pedro> to write a "macro define" interactively (without
> Pedro> going via the compiler first), and puzzling about why it
> Pedro> doesn't exactly work [due to some typo].  But we can
> Pedro> decide to do that incrementally.  Fine with me to push
> Pedro> as is if you'd like.
> 
> I've switched it; and good thing, too, because this caught a bug in the
> previous patch that could cause an infinite loop.
> 

Nice.  Patch looks good.  Please push.


> +  /* Whether the loop should keep going.  */
> +  bool keep_going = true;

'keep_going' doesn't appear used after the loop.  Could it be
declared in the for's init statement ?

  for (bool keep_going = true;
       keep_going;
       ...

> +
> +  for (;
> +       keep_going;
> +       get_next_token_for_substitution (&replacement_list,
> +					&tok,
> +					&original_rl_start,
> +					&lookahead,
> +					&lookahead_rl_start,
> +					&lookahead_valid,
> +					&keep_going))
>      {

Thanks,
Pedro Alves