Re: enable fdiagnostics-show-option by default

[Gabriel Dos Reis <gdr at integrable-solutions dot net>]

On Sat, Apr 17, 2010 at 12:49 PM, Toon Moene <toon@moene.org> wrote:
> Ian Lance Taylor wrote:
>
>> Manuel López-Ibáñez <lopezibanez@gmail.com> writes:
>>
>>> On 16 April 2010 16:01, Ian Lance Taylor <iant@google.com> wrote:
>>>>>
>>>>> Why not? ?After all, the patch is changing the current default and
>>>>> users
>>>>> who want to have the old default would have to do it in a user-by-user
>>>>> basis.
>>>>
>>>> Because I believe that in general it serves the community better for
>>>> all naïve users to see the same presentation for warning and error
>>>> messages, modulo translation.
>>>>
>>>> Also because there are hundreds of aspects of gcc behaviour that could
>>>> in principle be controlled at configuration time. ?Why single this one
>>>> out? ?gcc is already too configurable as it is. ?Too much
>>>> configurability is not a feature. ?It adds complexity to the overall
>>>> system for minimal gain.
>>>
>>> So I guess a patch to add a configure time switch won't be accepted.
>>> It is quite a pain in the ass to write it, so I don't want to waste
>>> (more) my time. Is this a deadlock? No solution?
>>
>> So far I think Gaby is opposed to your proposed change and Diego and I
>> are in favor of it. ?Chris Lattner says this patch makes gcc more like
>> LLVM, which is, I believe, generally acknowledged to have better
>> diagnostics. ?Do any other maintainers want to express an opinion on
>> the patch? ?The original patch is
>> ? ?http://gcc.gnu.org/ml/gcc-patches/2010-04/msg00910.html
>
> I particularly found the following - brought up by Chris - important:
>
>> FWIW, clang defaults to this being enabled, we haven't heard any
>> complaints.
>
> Certainly, there are enough users of clang to guide us in this.

I believe the alternative of issue [...] note only once for a given set
of diagnostic message still give us the essential information and at
the same time reduce the noise.

It is unfortunate that what I was said was not correctly characterized
in Ian's last message.

-- Gaby