TUNES website homepage draft

[Brian Rice]

At 12:59 PM +0200 7/3/01, Lendvai Attila wrote:
>:: >it would be also good to warn every newcomer right at the
>:: beginning that
>:: >their head is probably full of defcato standards, like a
>:: program must be
>:: >compiled before its run and things like that. so there
>:: should be a short
>:: >"shocking" text about things like OOP without classes, the
>:: rescrictiveness
>:: >of the bitvector modell, the minimal abstraction between
>:: the way current
>:: >computers work and the computing modell of current systems.
>:: and that its a
>:: >fool thing connecting the two so tight, the only requiremet
>:: is that the
>:: >computing modell ought to be calculable (or handleable,
>:: damn my english) by
>:: >current and future computer hadrwares. and the lack of persistency.
>
>heres another one for the above list:
>
>It's fool to compile a program to binary and then directly run it on a
>hardware, which is then trying to prevent the binary, whose behaviour is
>unknown and hardly analizable, to do any harm to the system. Managing all
>this trough a central privilegized kernel, which is after all only exists to
>try closing all the backdoors opened by this way of running direct CPU code.

Okay, if I generalize what you're saying then I could make a general 
argument against various ways of programming based on their efforts 
having to work counter-productively. I'd like to put this in more 
formal terms, and this is exactly why I invoke category theory, 
because it expresses the mappings between types of things rather 
well, intuitively and formally.

>ahh, and dont forget to talk about recent FPGA chips and that if you compile
>into "cpu design" instead of the instructions of a fixed given cpu then you
>can acheeve even 1000x speed with the same number of transistors. and to do
>this you need a dynamic tunes like system, and in C or Java you definately
>can't even start thinking how to produce "cpu design" from them.

Indeed, and we should mention this on our pages somewhere, in order 
that people in those areas see that some software research is headed 
to work in that direction. However, the vague promise is worthless as 
such, mind you.

>some random toughts again...

Ah, random thoughts. You should look through the TUNES mail archives 
if you want to see some *truly* random direction. This is precisely 
the motivation for my homepage change proposals: to state up front 
and clearly exactly what TUNES is about, how we understand it to be 
possible in terms of existing ideas, and what is required to 
implement it. Any vague promises (like what we have now) will have 
every Tom who can code and thinks that the idea is "c00l" telling us 
exactly how we should do it. I'm saying that the home page right now 
makes us look like fools ourselves.

>- 101.

Thanks for listening,
~