Son of Efficiency of Closures

[GODDEN%gmr.com@RELAY.CS.NET]

From:	net%"barmar@think.COM" 20-OCT-1989 05:45:35.40
Subj:	Efficiency of local fns and closures

    >  From: GODDEN%gmr.com@RELAY.CS.NET
    >  c) I have read that closures cannot be compiled.  True? False?
    >      Implementation dependent?  (For extra credit: If true, why?)

  >Closures can't be compiled separately from the function that created them.
  >If you compile a function that contains internal functions, they will all
  >be compiled.
  >						barmar
  
Let me add question d) with a different example:

 (defun return-a-closure (x)
	#'(lambda (y)
	    (some-function x y)))

 (compile 'return-a-closure)

 (compiled-function-p (return-a-closure 'bar)) --> nil

Why isn't the returned closure a compiled function?  This would indicate 
that when the closure is later called, it is evaluated by the interpreter.
If that closure contained some really large Lisp form, then it's going to be
SLOW.  Do I understand this correctly?

-Kurt Godden
 godden@gmr.com