Path: utzoo!news-server.csri.toronto.edu!cs.utexas.edu!swrinde!zaphod.mps.ohio-state.edu!wuarchive!uunet!visix!ip2020!adamksh
From: adam@visix.com
Newsgroups: comp.lang.forth
Subject: Re: Forth in CS
Message-ID: <1991Mar7.193258.2069@visix.com>
Date: 7 Mar 91 19:32:58 GMT
References: <9103061356.AA26910@ucbvax.Berkeley.EDU>
Sender: news@visix.com
Reply-To: adam@visix.com
Organization: Visix Software Inc., Reston, VA
Lines: 82

Okay, I guess I should explain myself more carefully.

I stated that structured programming considered GOTO's evil, primarily
because I wanted to use that example for an analogy.

I won't argue about whether GOTO's have legitimate uses.

I won't argue about whether the definition of structured programming
necessarily considers GOTO's evil.

I know that not all control structures are loops; e.g. CLU and Ada
have exception handling.  I used loops as a common example.

I only concluded recently that "control structures suck," so I am not
prepared to demonstrate equivalents for every existing control
structure, and so I wrote "used primarily", and "I suspect."

I do not think all theorists forget to ask the right questions.  I do
not think my questions are the ones theorists have to ask; I do not
think that theorists are doing the wrong thing.

> Theorists follow their interests, often without considering
> applications (nor should they).
		^^^ ^^^^^^ ^^^^ 
I do think that techniques like parsing, which may be very well
understood theoretically, are not necessarily the best way to improve
my productivity as a programmer (unless I do Natural Language).

> Ok, let's reinvent the weel (your arguments were used sometime ago... 196X).

I don't understand the purpose of this comment.  Are you claiming that
people already program in the style I suggest?  I don't find this in
the code I have seen, except for APL, some (repeat: some) Lisp
programs, and UNIX one-line command scripts.

I wouldn't claim to have invented this wheel; APL, Lisp, and UNIX are
examples.  I should also say that my train of thought was started by
the "Minimizing Control Structures" chapter of Thinking Forth.

But wheels that have been invented still need to be used.

I also can't tell what you mean about my arguments being used back in
196X.  I'm not even sure if you might have misunderstood my argument.
References would help.

> I didn't understand the last. (UNIX)

Briefly, UNIX's pipe operator allows you to construct surprisingly
sophisticated one-line command sequences, without conditionals.
Please don't ask me to explain in detail; any good UNIX book will do.

> Anyway, for the first two: Think Functional
> Paradigm! And Think OOP because there you have this too...

Functional programming is about evaluation and avoiding side-effects.
These are separate issues from how often the keyword "IF" appears in
your program.  And, Goldberg and Robson's _Smalltalk-80_ contains
neither use nor mention of the programming style I suggest.

> No IF's? How can you do that? LISP can't... APL have IFs too.
	:
	:
> Iteratio is a control structure, if you don't know. Control structures makes
>the program do things in a non-sequencial way. Calling procedures is a control
> structure, repeat many times is a control structure, and it doesn't matter if
> your LOOP is hidden or not. You HAVE to specify an iterator, don't you?

I just realized what I should have said in the first place.

	Every line should be a basic block.

A basic block has one entry point and one exit point.  In languages
like C, since a function is the unit of decomposition, it is only
required that functions be basic blocks.  Since Forth encourages a
finer decomposition, it suggests that we can make basic blocks
smaller.

Smaller basic blocks also provides enormous benefits for traditional
compilers; smaller basic blocks makes programs easier to reason about,
and therefore easier to optimize, and also easier to parallelize.

Adam