[KLUG Members] Software Theory

[Adam Tauno Williams]

>Bob Brown and I were speaking of KDE v. GNOME, and we got into the stated
>goals of toolkit designers. KDE is primarily a desktop and toolkit which is
>evolving into something more; GNOME is a toolkit, GTK+, welded to the
>CORBA/Bonobo Software Object Model, which was designed to be the Most
>Perfect, Most Complete desktop/ORB in the world. This led me to:

Actually GNOME didn't start out with a component model.  GNOME's focus
was code re-use,  and they thought they could accomplish it with
cleverly constructed shared libraries.  KDE actually started out with
more of a component model (including the use of CORBA) but abandoned it
due to performance problems and the fact that developers weren't
familiar with the technology.  Thus KDE experienced faster initial
development,  and why (according to many) GNOME application developement
is only now really starting to pick up steam.

>Descartes's Third Meditation contained the so-called Ontological Proof of
>the Existence of God. (Similar is Anselm's reductio ad absurdam Proof of
>God.) Cutting to the chase, it basically says that ideas are things, and
>since we can conceive a Perfect God, and no inferior thing can 'birth' a
>greater thing, this God must exist.

1.  Can you really conceive of a Perfect God?  Or are you just
conceiving of the conception of a Perfect God?

2.  The meaning of the attributes of "greater" and "inferior" in this
argument imply a universal scale of value or quality for which the
argument itself offers no method to construct.

>(This must be bitter irony to we who can conceive of perfect SO's, 
>but can never seem to find, date and marry them.)

Isn't it possible that the "perfect SO" could in fact be the worst
imagineable SO?  They're would be no point in discussing anything.

>In reply to this, Bob mentioned a similar phenomenon: language designers who
>regard their programming language as complete and perfect only when they can
>write the language compiler in the language itself. Thus, we find:
>1. This language is very good and complete.
>2. The compiler/interpreter of such a good language must be good, too.
>3. I can write the compiler in the language, which is even more proof of how
>   complete it is, therefore good.
>4. Repeat steps 2 and 3 until perfect.

Anyone writen a compiler using a shell script?