Re: ruby

[Lennon Day-Reynolds <address@hidden>]

First off, let me apologize; it wasn't my intention to start an 
antagonistic discussion on how bad one language or another was; 
rather, I was simply trying to make a point about why Ruby failed 
to appeal to me as much as it might have. Given the somewhat mixed 
response, however, I feel that it might be best for me to clarify 
and complete my original statements, just to avoid potential 
misunderstandings.

> Gee, you really abandoned Java for such a small problem?  Is that
> honestly your major problem with Java?  Would you feel the same way
> about Smalltalk?  Anyway, it has never bothered me.  In the cases
> where there really isn't anything OO going on, which in my experience
> is usually groups of little utility functions, I just use static
> methods, with the class simply serving as a name scoping thing.

I'm still not convinced that OO is "the" design paradigm that all 
software should be based on. Maybe I'm just dense, or the 
specification and synthesis devotees I work with are finally 
rubbing off on me.

Perhaps I should have elaborated more fully on my reasons for being 
glad to have Java behind me for day-to-day programming tasks:

	* static typing
	* no interactive environment
	* no closures, continuations, or tail-call elimination
	* primitive numeric type/class instance divide
	* awkward, limited standard string processing
	* rigid, verbose syntax

I probably spent half of my time in Java trying to work around 
these and other basic (admittedly deliberate) limitations in the 
language. Certainly, most (if not all) of these problems can be 
"patched" with an appropriate add-on or extension, but then I'm no 
longer working in the standard Java that everyone knows and loves.

I do miss a some things from the Java world: some of the features 
of the J2EE environment, such as container-managed persistence, 
servlets, and message-driven objects, were actually quite nice. Of 
course, I've since found equivalents to most of the really useful 
bits in dynamic programming environments, and consider myself much 
better off.

> I suppose if you often write very little script-like programs,the text
> overhead of the "class" statement might seem like a lot of characters
> to type; is that the issue?  I guess I don't write programs small
> enough for that to be a problem.

Probably somewhere on the order of 75% of the code I write is 
either a "very little script-like" program, or a prototype that may 
never see use outside my workstation. That means I usually don't 
want to fire up Rational Rose, grab the latest 'patterns' textbook, 
and consult a wall full of inheritance charts just to get the thing 
up and running.

Obviously, we're probably working on different types of systems, 
but I think that the majority of code written *isn't* part of a 
complex system, and suffers more from the lost productivity of 
working in a statically typed, rigid language like Java than it 
does from typing errors or poor organization.

Part of the problem, I guess, is that I just don't like putting 
things into a class that have no part being there. Some things are 
better modeled as a simple package of functions, a procedural 
script, a meta-class, etc., none of which can be first-class 
citizens in standard Java.

When I need an 'architecture,' or something that I can hand off to 
a new programmer, though, I can take my lumps and "do the right 
thing" -- which in the above cases usually means writing the least 
clever, most over-engineered, and least-dynamic system I can. The 
rest of the time, I'm lucky enough to get to work with dynamic 
languages, create compact, intelligently-written code, and use a 
variety of abstraction techniques.

> I do agree that it's best to have the best of both worlds, which
> I think Lisp-with-CLOS gives you.  (I don't know Ruby at all so
> I can't comment on it.)

Ruby is certainly more dynamic than Java, but is much more 
pervasively object-oriented than, say, Python or Scheme. One 
feature of the language which I haven't been found a good 
explanation of yet is the meta-object model, though there are 
tantalizing references to it several places online.

In essence, I'm just fickle. To put my own spin on your CLOS 
comment, I think it's best to have the choice of *all* worlds, or 
at least as many as are practical. I don't need the language I'm 
working in to artificially constrain the available semantics of my 
code just to protect me from "difficult" designs. 
Object-orientation is one technique available, and is often a very 
helpful addition to my toolkit.

Again, my apologies for the digression; I'll try to move forward 
from here on out, instead of stepping back to re-explain myself.


Lennon Day-Reynolds