'goto" gone from computer languages or is it!

[Sean 'Captain Napalm' Conner]

It was thus said that the Great Dave Dunfield once stated:
> 
> >  Hmmm ... 
> >
> >	void print_arg(int *aryp,size_t n)
> >	{
> >	  printf("%u",*aryp++);
> >	  while(--n)
> >	  {
> >	    printf(",%u",*aryp++);
> >	  }
> >	  putchar('\n');
> >	}
> 
> But has two calls to printf, with different format
> strings. More than doubles the static string space.
> Plus two complete printf call frames (bigger code)...
> one being used only once.

  Well ... one could do:

	void print_arg(int *aryp,size_t n)
	{
	  static const char text[] = ",%u";

	  printf(text+1,*aryp++);
	  while(--n)
	  {
	    printf(text,*aryp++);
	  }
	  putchar('\n');
	}

  Which I stuffed through GCC 2.7.2.3 (admittedly not a recent version of
GCC) using the following command line:

	% gcc -S -O2 -fomit-frame-pointer t.c

  And got the following (cleaned up):

		.section	rodata
text		.text		",%u"

		.section	code

print_arg	push	esi
		push	ebx
		mov	esi,[esp + 12]
		mov	ebx,[esp + 16]
		push	dword ptr [esi]
		add	esi,4
		push	offset text + 1
		jmps	.L11

.L9		push	dword ptr [esi]
		add	esi,4
		push	offset text

.L11		call	printf
		add	esp,8
		dec	ebx
		jne	.L9
		push	dword $_IO_stdout
		push	dword 10	
		call	_IO_putc
		add	esp,8
		pop	ebx
		pop	esi
		ret

  Surprised me!  A single call to printf().  Not bad---then again, GCC had
been in development for quite a long time.  And the printf() stack frame is
no different from any other C stack frame (in declaration yes, but not in
how it works).

> gotos and other such structures are not evil - lack
> of understanding of when such constructs are
> appropriate (and not appropriate) is the real
> problem. "Banning" the constructs just serves to
> emphasize that this is not obvious to some people.
> 
> Hmmm... we seem to be sailing away from the topics
> again!

  Oh dear, we are, aren't we?

  -spc (And the assembly output has one unconditional GOTO and one
	conditional GOTO ... but who's counting?)