Re: Down with Upvalues

[ramsdell@... (John D. Ramsdell)]

Edgar, 

I sent a note this morning continuing the discussion on lexical
scoping before I read your note suggesting we take this discussion off
line.  Let me simply say that all the definitions of lexical scoping
either explicitly say, or by example show, that nested functions in a
lexically scoped language can use the value of variables defined in
their enclosing functions.  This is a fact.

John

Edgar Toernig <froese@gmx.de> writes:

> "John D. Ramsdell" wrote:
> > 
> > > You mix scope with accessibility.
> > 
> > The purpose of scoping rules is to define accessibility.
> 
> No.  Possible accessibility or better visibility.  Access rules
> are a completely different thing.  I.e. take a pseudo language
> that has attributes for input and output parameters (input are
> read only, output are write only):
> 
>   define foo(in: a,b; out: x)
>     x:=a+b  -- ok
>     a:=1    -- error: trying to change read-only var a
>     z:=x    -- error: trying to read write-only var x
>   end
> 
> Scope rules define which a, b, and x are taken (the formal parameters
> of foo).  Access rules define what you may do with them.  Or take an
> attribute 'private' that prohibits access from nested functions i.e.
> to better document usage or avoid closure generation:
> 
>   define foo()
>     private x
>     define bar()
>       print x  -- error: unable to access private x
>   ...
> 
> The same here.  The scope rules define what declaration is chosen and
> access rules what you are allowed to do with them.
> 
> Access rules may even be dynamic (especially in dynamically typed lang-
> uages like Lua).  I.e. you may set the setglobal/getglobal tag methods
> to restrict access to global variables.  Or you may create references
> to variables and suddenly they become accessible from anywhere, ignoring
> any scoping rules.
> 
> > Look at the FOLDOC definition of lexical scoping.
> > 
> >      In a lexically scoped language, the scope of an identifier is
> >      fixed at compile-time to be the smallest block (begin/end or
> >      function/procedure body) containing the identifier's
> >      declaration. This means that an identifier declared in some block
> >      is only accessible within that block and from procedures declared
> >      within it.
> > 
> > It defines lexical scoping by describing when and were a variable is
> > accessible.  Accessibility is what scoping is all about.
> 
> Again.  The FOLDOC entry is not a formal definition.  And I read the
> last sentence as: An identifier declared in some block is never visible
> outside of that block - only within it.  [Note: access may even be possible
> from outside of that block via references.]  Btw, this definition is so
> short-sighted that even C++ would not conform.
> 
> 
> To your other posting:
> >
> > Let me make it clear that any definition of lexical scoping taken from
> > a Scheme source will further my case.  In the eighties, general
> > agreement was found.  The Scheme reference on
> >
> >   http://lua-users.org/wiki/LuaScopingDiscussion
> >
> > is
> >
> >   http://www.htdp.org/2001-01-18/Book/node104.htm
> >
> > a classic example of a link that furthers my cause.
> 
> This is a "How is lexical scoping done in some kind of Scheme".  Try to
> get away from that specific rule set.  Take a more general view of scoping
> and you may have much less problems with different styles ;-)
> 
> 
> We should take this discussion off the mailing list.  Nobody else seems
> to be interested in it any more.  If you want to continue this discussion
> (which I find very interesting) then I think it would be better to reply
> by private mail.
> 
> Ciao, ET.