[dfawcus]

depending upon how one has structured the code, a less painful way to write the same is:

    typedef char array[5];

    void do_something(array *a) {
        enum { a_Size = sizeof *a };
        memset(*a, 'x', a_Size);
    }

it rather depends upon how painful it will be to create a bunch of typedefs.

Beyond a certain point, if there are too many arrays of the same size with different purposes, my inclination is to wrap the array in a struct, and pass that around (either by pointer or value depending upon circumstances.)

The existence of the decaying form is if I recall correctly a backward compatibility thing from either B or NB; simply because in one or the other pointers were written in the (current) array syntax form.

[int_19h]

It stems from B, because it didn't have either pointers or arrays on the type level. Declaring an array allocated the storage, but the variable itself was still a word-typed pointer to said array. In fact, you could even reassign it!

   foo(a) {
      return(&a[1]);
   }
   bar() { 
      auto a[10];
      a = foo(a);
   }

The decaying system made it mostly work with minimal changes in C.