heapsort in frame.c, conflict with BSD/OS definition

[Shigeya Suzuki <shigeya at foretune dot co dot jp>]

I was trying to build latest CVS snapshot (98/04/18 22:06 JST) and
found following.

BSD/OS (and probably others) have heapsort definition in stdlib.h. And 
it is now conflicting with definition in gcc/frame.c.

in stdlib.h, it defined as:

int      heapsort __P((void *, size_t, size_t,
            int (*)(const void *, const void *)));

and in gcc/frame.c, 

/* This is O(n log(n)). */
static inline void
heapsort (fde_vector *erratic)
{

so conflicting.

I guess there are similar problem in other platforms. How I can modify
this? My quick and dirty solution is change heapsort function name as
follows..

Shigeya


*** ../egcs-latest/egcs/gcc/frame.c.FCS	Sat Apr 18 22:07:12 1998
--- ../egcs-latest/egcs/gcc/frame.c	Sat Apr 18 23:17:49 1998
***************
*** 271,277 ****
  
  /* This is O(n log(n)). */
  static inline void
! heapsort (fde_vector *erratic)
  {
    /* For a description of this algorithm, see:
       Samuel P. Harbison, Guy L. Steele Jr.: C, a reference manual, 2nd ed.,
--- 271,277 ----
  
  /* This is O(n log(n)). */
  static inline void
! inline_heapsort (fde_vector *erratic)
  {
    /* For a description of this algorithm, see:
       Samuel P. Harbison, Guy L. Steele Jr.: C, a reference manual, 2nd ed.,
***************
*** 366,372 ****
    fde_split (&accu->linear, &accu->erratic);
    if (accu->linear.count + accu->erratic.count != count)
      abort ();
!   heapsort (&accu->erratic);
    fde_merge (&accu->linear, &accu->erratic);
    free (accu->erratic.array);
    return accu->linear.array;
--- 366,372 ----
    fde_split (&accu->linear, &accu->erratic);
    if (accu->linear.count + accu->erratic.count != count)
      abort ();
!   inline_heapsort (&accu->erratic);
    fde_merge (&accu->linear, &accu->erratic);
    free (accu->erratic.array);
    return accu->linear.array;