Laurent Martelli
25 Apr 1999 00:28:32 +0200

>>>>> "Maneesh" == Maneesh Yadav <> writes:

  Maneesh> There is some stuff going on in the Tunes "project" that
  Maneesh> irking me..feel free to respond...

  Maneesh> With respect to the people who are working on Brix and
  Maneesh> RetrOS etc...  There are more important things to
  Maneesh> have invested lots of time and effort to writing an
  Maneesh> OS...not totally different from any other.  We don't need
  Maneesh> bootstrapping linkers drivers loaders and compilers right
  Maneesh> now, they're all available and better than the ones any
  Maneesh> single coder could make.  You can only develop you OSes so
  Maneesh> far...distributive computing, symmetric multiprocessing,
  Maneesh> advanced techniques in memory managment, widely implimented
  Maneesh> video suppourt, network cards simply can't be done without
  Maneesh> a sizeable development effort, and furthermore scuh an
  Maneesh> effort is simply not justified for the OS's mentioned.

Completely agreed.

  Maneesh> As far as I I understand THIS IS THE EXACT TYPE OF THING
  Maneesh> TUNES IS AGAINST.  A good deal of work that you are doing
  Maneesh> is tedious (if you want to write your own HD driver you
  Maneesh> have to look up specs and implement calls, ditto for video,
  Maneesh> network blah blah blah...).

  Maneesh> Someone slap me with a tuna if I am wrong, but Tunes is
  Maneesh> about metaprogramming, coming up with a way to specify
  Maneesh> information and methods for computers to work with that
  Maneesh> information and be able to produce from it...(i.e. here's
  Maneesh> the specifications for file system, please make
  Maneesh> it)...minimize the time from the idea to the product.  As
  Maneesh> programmers we spend way to much time doing the translating
  Maneesh> between specification and impliementation.  If we can
  Maneesh> achieve this goal of having a unified system that can work
  Maneesh> with information in such a manner (and do it dynamically)
  Maneesh> we can write anything under it we want.

The problem with specs is that they are not all written in the same
`language'. If they were, there would be no need to implement them
because the `implementation' would be obvious. In fact, specs and
implementation would be the same thing. 

I believe that what we want to achieve with Tunes amounts to defining
such a language. And I believe that Scheme is such a language, and
that the main problem is a design problem. We should not mix together
independent issues, because when you want to reuse an aspect of the
system, you pull other aspects which you don't want but which are tied
to the aspect that you want to reuse.