Re: [Gcl-devel] Current state of the ansi test suite

[Paul F. Dietz]

Camm Maguire wrote:
> Greetings!
>
> OK, I'll defer to your judgement here.  What about supplying functions
> that aren't in the spec -- i.e. implementation-dependent functions?
> Is this generically OK as long as I don't tread on the canonical
> namespace? 

It's fine to add extra functions (all implementations do), but they can't
be exported from the COMMON-LISP package.

> Which leads me to another question -- there is no truly
> 'static' function declaration in lisp, is there?  I.e. if one does not
> export a function from a package, one can still always access it with
> package::function, right?  Is there any way to make an internal
> function  visible only to a certain defined group of other functions,
> no matter how it is called, i.e. like a static declaration in a C
> 'file'?

You can wrap the defuns for those functions in a FLET:

(flet ((sneaky-function (...) ...))
  (defun visible-function (...)
     ... (sneaky-function 'foo 'bar) ...)
  ...
)

I think you could also do something like use LOAD-TIME-VALUE to get
the symbol-function slot of a symbol in its callers, then unbind the
that slot.  That would make the function inaccessible elsewhere:

(defun foo (...) ...)

(defun bar (...) ... (funcall (load-time-value #'foo) ...) ...)

(eval-when (load) (fmakunbound 'foo))

        Paul