tail call efficiency (was: Modified allwords() function)

[Wim Couwenberg <wcou@...>]

I another post Mike Pall profiled coroutine code and found that it uses 
more time then he'd like:

> Especially because the latter needs an additional call
> frame (which could be improved by making yield a keyword/VM op).

On a related note, I wonder if it would be at all possible to optimise 
tailcalls any further?  Consider the following functions to sum an 
arithmetic sequence:


local function sumi(n, r)
  while r > 0 do
    n = n + r
    r = r - 1
  end
  return n
end

local function sumr(n, r)
  if r > 0 then
    return sumr(n + r, r - 1)
  end
  return n
end


The recursive sumr takes about 2.5 times more time than the iterative 
sumi in 5.0.2.  (In 5.1-w5 both are quicker, but the iterative version 
more so than the recursive one...)  If tailcalls get cheaper, they would 
gain in strength as "structured goto's".

Just wishful thinking probably...  ;-)

--
Wim