Re: [MLUG] guess what

[Jonathan King <EMAIL:PROTECTED>]

On Fri, 2 Apr 2004 EMAIL:PROTECTED wrote:

> > The *point*, though, is that this is a demonstration of what
> > true lexical scoping would really look like and what it allows.  
> > There is a lot more out there than Python.  I don't have any
> > more time to talk about lexical closures and their uses.  Just
> > google it, Michael.
> 
> Lexical closures are quite useful. Perl has had them since 1995.
> Java got them with "Inner Classes" in 1.4, I believe.

And Scheme has had them since the Very Beginning.  Seriously, people 
could do worse in developing a new language than to re-syntax some 
of the nice lisp-like languages out there that have all of the 
"modern" conveniences.  Actually, R, a language that has come up in 
this thread, *is* now Scheme-based, I believe.
 
> Damien Conway in his "OO Perl" book (Manning), gives a convincing
> argument that lexical closures are the only way to get "truly"
> private object attributes in languages that expose "namespaces"
> and class definitions through "reflection".

I have the book, but I will admit to not having completely read it, 
yet.  If Perl6 really does ever happen, Damien will make sure that 
this aspect of the language gets even better.
 
> His book is is really more about the interesting nooks of OO,
> multiple inheritance, data inheritance vs method inheritance,
> closures, public/private object interfaces, methods as objects and
> data as objects, and the like, than about Perl, but really shows
> how OO Perl *can be*.

One reason I haven't read the whole thing is that my feelings on OO 
are that people way over-use inheritance when (say) composition is 
what they really want.  Ousterhout (of Tcl/Tk fame) had something 
about this point, too, in his somewhat controversial scripting 
paper.

jking


_______________________________________________
members mailing list
EMAIL:PROTECTED
http://mlug.missouri.edu/mailman/listinfo/members