Re: [RFA v2 10/17] C++ify mi_parse

[Pedro Alves <palves at redhat dot com>]

On 04/12/2017 05:15 PM, Tom Tromey wrote:
>>>>>> "Pedro" == Pedro Alves <palves@redhat.com> writes:
> 
> Pedro> The patch made mi_parse be a non-POD (non-trivial dtor), so I
> Pedro> think we should get rid of that memset at the same time.
> 
> This change was a little bigger than the others.  I've added this
> constructor:
> 
>     mi_parse::mi_parse ()
>       : op (MI_COMMAND),
>         command (NULL),
>         token (NULL),
>         cmd (NULL),
>         cmd_start (NULL),
>         args (NULL),
>         argv (NULL),
>         argc (0),
>         all (0),
>         thread_group (-1),
>         thread (-1),
>         frame (-1),
>         language (language_unknown)
>     {
>     }
> 

Thanks.  Totally fine with me.  If you prefer, using in-class
initialization is also fine.  (We've been using it more in recent patches.)

> ... and changed the construction of the object like:
> 
> @@ -241,12 +241,5 @@
>  
> -  memset (parse, 0, sizeof (*parse));
> -  parse->all = 0;
> -  parse->thread_group = -1;
> -  parse->thread = -1;
> -  parse->frame = -1;
> -  parse->language = language_unknown;
> -
> -  cleanup = make_cleanup (mi_parse_cleanup, parse);
> +  std::unique_ptr<struct mi_parse> parse (new struct mi_parse);
>  

Perfect.

Thanks,
Pedro Alves