Re: bug report...

[Branko Cibej <branko dot cibej at hermes dot si>]

Joe Buck wrote:

> Ideally, the -ansi flag should warn about use of extensions (that
> doesn't mean it currently does in every case) and -ansi -pedantic
> should make the compiler really nitpicky about deviations from the
> standard.  Again, because the language is so big I'm sure egcs has
> missed a lot of these things.

Speaking of deviations from the standard; I _think_ the C++ FDIS allows an
expression on a return statement in a function returning void, and that this was
allowed specifically to simplify writing templates, e.g.:

------void.cc begins------
template <typename R> class functor {
  public:
    functor (R (*fn) ())
        : func(fn)
        {}

    R operator() ()
        {
            return func ();
        }

  private:
    R (*func) ();
};

int foo () { return 1; }
void bar () {}

int main ()
{
    functor<int> f(foo);
    functor<void> b(bar);

    f();
    b();

    return 0;
}
------void.cc ends------

Egcs snapshot 1998-05-31 has this to say about the code above:

$ /home/brane/eval/egcs/bin/gcc -v
Reading specs from /home/brane/eval/egcs/lib/gcc-lib/sparc-sun-solaris2.6/egcs-2.91.34/specs
gcc version egcs-2.91.34 19980530 (gcc2 ss-980502 experimental)

$ home/brane/eval/egcs/bin/c++ -ansi -pedantic -o void void.cc
void.cc: In method `void functor<void>::operator ()<void>()':
void.cc:25:   instantiated from here
void.cc:9: warning: `return' with a value, in function returning void

Should this really be a pedantic warning?

    Brane

--
Branko Cibej   <branko.cibej@hermes.si>
HERMES SoftLab, Litijska 51, 1000 Ljubljana, Slovenia
phone: (++386 61) 186 53 49  fax: (++386 61) 186 52 70