Re: enable fdiagnostics-show-option by default

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

On Thu, Apr 15, 2010 at 6:22 PM, Ian Lance Taylor <iant@google.com> wrote:
> Gabriel Dos Reis <dosreis@gmail.com> writes:
>
>> This would add to the substantial amount of output, especially if the
>> warnings are repeated.
>
> This is a valid concern...
>
>
>> I would prefer this behaviour to be configurable at configure-time, so that
>> users or packagers can decide which behaviour is
>
> ...however, I don't think this would be a good approach. ?I don't
> think this is the kind of thing which should be controlled on a
> user-by-user basis.

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.

The suggestion of configuration-time basis is that it is done for all
at configuration time -- either by the packagers, or by the user
who built the compiler.  That is better than the current behaviour, and less
disruptive than what the current patch proposes.

> ?If we think that the majority of users--in
> particular, the majority of naïve users--would be helped by this
> change, then I think we should make it, and let the minority run gcc
> via a shell script.

How do we measure that?
We are talking about changing a default that has been there for ages.


>
> Many compilers have some sort of code attached to their warning
> messages. ?This would be gcc's equivalent to that.

Note that I am not saying that there should not be any annotation to
diagnostic messages. We are discussing what the defaults should be
and how they are enabled.

>
> Ian
>