TUNSE and Retro
Mon, 31 May 1999 11:00:50 -0400
Ken Evitt writes:
>TUNES is not about features. TUNES is about integrity and consistency. TUNES
>is about simplicity and elegance. Linux started out as one man's personal
>project. But because it was useful, and free, people made of it what they
>wanted. The purpose of Retro and Brian's Arrow system is not to implement
>every feature anyone would like but to make those implementations possible,
>in a unique way--and that is TUNES.
Actually, life itself is neither simple nor elegant. It's true that
the underlying foundation is simple and elegant, but the actual
implementation is an insanely complex soup of kludges, hacks, and
People's computing needs are similarly chaotic. A typical day behind a
computer involves missing information, looming deadlines, and continutally
changing needs. I doubt that a computer system which is an elegant,
unified whole will be able to match these "organic" demands. Linux's
messy pool of features reflects the messy pool of its users' needs.
In that light, Linux's weakness is not its redundancy of features or
lack of structure, but rather its lack of adaptability to new tasks.
Hierarchy-based object-oriented language designers have an analagous
problem; they expect you to plan the whole project before you start
it, and are very unforgiving if you need to go back and restructure
your design mid-way. But, unless you're work is totally unoriginal,
restructuring is the fundamental process of programming!
Therefore a good operating system (and a good programming language,
although at some level they're the same thing) should provide powerful
mechanisms for assimilating old code and reliably reusing it for new
purposes. Again, from that viewpoint it doesn't matter what low-level
OS you start with or how its features are organized; what's important
is *what* features are already there. The first great rule of
programming is to avoid writing new programs wherever possible. :-)
This is not a criticism of Retro directly, but rather a criticism of
the "academic" design mentality at large. There are perhaps better
motivations for working on Retro.