Re: [Chicken-hackers] Fix attempt for bug #1269 "Port or reader state corruption"

[Evan Hanson]

On 2016-03-27 21:30, Kooda wrote:
> I went for the first solution for now, I suppose it??s a good enough??
> fix for now. A better one should go to CHICKEN 5.

Agreed, though I don't have any idea what that will look like, now. I do
think this one should be applied to both, though, to avoid divergence in
the meantime.

> diff --git a/chicken.h b/chicken.h
> index be5f308..0c985cf 100644
> --- a/chicken.h
> +++ b/chicken.h
> @@ -906,7 +906,7 @@ typedef void (C_ccall *C_proc)(C_word, C_word *) C_noret;
>  # define C_vfprintf                 vfprintf
>  # define C_fflush                   fflush
>  # define C_getchar                  getchar
> -# define C_exit                     exit
> +# define C_exit(x)                  { C_fflush(NULL); _exit(x); }

I'd prefer to leave C_exit as it is and introduce C__exit instead, or
perhaps put the fflush() call into C_exit_runtime and use that directly
wherever an _exit() is needed. Otherwise, C_exit will be the one C_foo
that isn't just an alias for foo, which seems like asking for trouble.

> diff --git a/posixunix.scm b/posixunix.scm
> index f56960d..ae1b9b0 100644
> --- a/posixunix.scm
> +++ b/posixunix.scm
> @@ -1573,6 +1573,8 @@ EOF
>  (define process-fork
>    (let ((fork (foreign-lambda int "C_fork")))
>      (lambda (#!optional thunk killothers)
> +      ;; flush all stdio streams before fork
> +      ((foreign-lambda int "fflush" c-pointer) #f)

"C_fflush".

Cheers,

Evan