[PATCH resend 1/2] PR c/65403 - Ignore -Wno-error=<some-future-warning>

[Martin Sebor]

On 8/26/19 6:24 AM, Martin Liška wrote:
> On 8/23/19 6:52 PM, Alex Henrie wrote:
>> On Fri, Aug 23, 2019 at 3:13 AM Martin Liška <mliska@suse.cz> wrote:
>>> On 8/23/19 6:17 AM, Alex Henrie wrote:
>>>> +vec<const char *> ignored_wnoerror_options;
>>>
>>> Here you'll need to add a comment.
>>
>> The declaration of ignored_options in opts-global.c doesn't have a
>> comment either. What would you like the comment to say?
> 
> I would just write a comment for
> ignored_wnoerror_options

FWIW, since the vector stores pointers it doesn't own, it might
be helpful to consider (and mention in the comment) who owns them.
I assume the strings they point to stick around for the lifetime
of the vector but that they do isn't obvious from its use, and
neither is how they are allocated (does the vector need to be
annotated with GTY?)

Martin

> 
>>
>>>> +      const char * opt = ignored_wnoerror_options.pop ();
>>>
>>> No space between '*' and 'opt' please.
>>
>> Okay.
>>
>>> You don't want to append and option that is already in the ignored_wnoerror_options:
>>>
>>> $
>>> ./xgcc -B. -Wunused-variable -Werror -Wno-error=some-future-warning -Wno-error=some-future-warning /tmp/main2.c
>>> /tmp/main2.c: In function ‘main’:
>>> /tmp/main2.c:3:7: error: unused variable ‘foo’ [-Werror=unused-variable]
>>>      3 |   int foo; /* { dg-error "unused variable 'foo'" } */
>>>        |       ^~~
>>> /tmp/main2.c: At top level:
>>> cc1: error: ‘-Wno-error=some-future-warning’: no option ‘-Wsome-future-warning’ [-Werror]
>>> cc1: error: ‘-Wno-error=some-future-warning’: no option ‘-Wsome-future-warning’ [-Werror]
>>> cc1: all warnings being treated as errors
>>>
>>> As seen the error is there twice.
>>
>> Joseph explicitly asked me to make -Wno-error=some-future-warning
>> behave the same as -Wno-some-future-warning (see
>> <https://gcc.gnu.org/ml/gcc-patches/2019-03/msg00065.html>), and
>> -Wno-some-future-warning prints multiple warnings if the option is
>> given multiple times.
>>
>>> One question about the behavior:
>>>
>>> Why do I need to have another warning to get the warning: ‘-Wno-error=some-future-warning’ printed?
>>
>> If we always give a warning about -Wno-error=some-future-warning then
>> combining that option with -Werror would cause compilation to fail,
>> which we don't want.
> 
> Ah, I see. It's not easy to implement that in a reasonable way. Now I understand that.
> 
> Thanks,
> Martin
> 
>>
>> -Alex
>>
>