Re: Usefulness of Formal Notions in Programming

["Dmitry A. Kazakov" <mailbox@dmitry-kazakov.de>]

[I never said that formalism is bad, on the contrary, programming is too
much informal. But that by no means should justify the kind of CS we have]

On Thu, 08 Mar 2012 00:28:37 +0100, Georg Bauhaus wrote:

> On 07.03.12 21:17, Dmitry A. Kazakov wrote:
> 
>>>> This is a great example of making a "science" out
>>>> of nothing and for nothing.
> 
[...]
> 1. Suppose someone knows what a regular language is.

I don't buy REs. For a layman wild-carded patterns are easier to use and
understand, while covering 90% all cases. For professional use (10%)
patterns is a bad idea anyway. I think we discussed that in c.l.a.

> If a person knows one simplified, abstract notion well (grammar),
> the he or she will more easily understand a different notion, provided
> there is an analogy, or correspondence.

The only problem is that formal grammars are not simple. They are hard and
difficulties grow exponentially closer you come to the real application
domain. The formalism is just too weak.

This is the problem of all poor solutions. Be it formal grammars, patterns,
domain-specific languages, scripting languages etc. You get a simple job
done simply and quickly. That makes you think - that is. But when you face
a real task you are lost.

> It seems somewhat like knowing math. If you know it, and know how to
> put math to good use, it helps understanding programs, and writing
> programs,

That is true, except that good math has a purpose and beauty. Formal
grammars (poor math) are more like learning how to divide and square Roman
numerals. That too would train you mentally and thus help by writing
program. But a sane man would convert to the decimal representation do the
job and convert back.

[... how bad Lisp is ...]

That Lisp is bad does not imply that formal grammars are any good.

-- 
Regards,
Dmitry A. Kazakov
http://www.dmitry-kazakov.de