Re: [PATCH] MIPS: Transform old-style macros to newer "__noreturn" standard.

["Satyam Sharma" <satyam.sharma@gmail.com>]

On 5/26/07, H. Peter Anvin <hpa@zytor.com> wrote:
> Satyam Sharma wrote:
> >
> > But __attribute__((noreturn)) is simply a _function attribute_. Of course,
> > it is legal / valid only for functions with return-type void, so it does
> > make
> > sense to combine both void and __attribute__((noreturn)) in the same
> > macro like you say. But that's not syntactically necessary. In fact,
> > grepping through the sources, a lot of people do prefer to place the
> > attribute _after_ the function declarator.
> >
> > Anyway, I'm fine either way.
> >
>
> Sorry to say, but weren't you the person who didn't recognize !! as the
> idiomatic booleanizing operator?

Yes, of course, please prove a link / connection between that and this?

> I think you need to learn that everything that the compiler accepts
> isn't necessarily idiomatic, readable code.  Consider
> __attribute__((noreturn)); it's a nonstandard feature implemented using
> a generic gcc mechanism -- thus what the compiler will accept is quite
> flexible, because it's a generic building block.  It doesn't mean it's a
> good idea.
>
> The reason it's often written at the end of the expression mostly has to
> do with bugs in some very early versions of gcc.

That might be, but I was only saying that there is no syntactical
*compulsion* to combine the attribute with the return type. As for what's
readable, it is subjective. And as for what's common / standard / idiomatic
in the kernel code as of today, nothing beats a grep. Anyway, as I said
previously, I'm fine with either way.

Satyam