Integer types used in libgfortran

[FX]

> The problem is that the library routine takes a *pointer* to an integer, and
> uses that to return the status. The object it points to must be of the
> correct type.

What about a strategy like the one used inside the I/O library, with
routines called such as:
int64_t extract_int (const void *p, int len) ?

But the int64_t should really be the largest integer type available
(which is not always int64_t, as PR 19308 reminds us), shouldn't it?

FX