Tue, 27 Apr 1999 13:50:59 -0400 (EDT)
Yikes, the mailing list is getting heated....LET"S NOT START A FLAMEWAR.
These things start by miscommunication, let's all understand that, and I'm
not for the bad karma generated.
On the other hand, I can understand Brian's point of view, it brings the
frustration of this "project" to the forefront. Nothing useful has been
done, I've been on this list since high school and I'm now in University!
We've got to face a bit of reality and realize that with current rate, we
won't get anywhere
We have all got to agree to a common goal
Our goals right now are quite frankly ridiculously disjoint. We've got to
look at the big picture. I don't know if what Brian has put out is
correct/useful as I have not fully read the paper, but I am very amazed
least he and I are on the same track (and I didn't really realize it
until the paper) and what I am sure about is that the goals he proposes
are the ones we should follow, at least first, for our project. As for tcn
and _QZ (I am speaking
objectively here, please do not take this personally); you're efforts
really are going to waste..the world doesn't need a half baked ukerenel.
You will never impliment all the features to make it a viable system the
way you are doing it (aka the way it has always been done); and I assure
you you will not spawn another linux movement to suppourt your systems. We
don't need a low level API, there are millions out there, and just about
all of them have a greater amount of coding hours put into them than
Tom and _QZ, if you are doing your ukernels for yourself, then say
so, because having another microkernel just won't do anything..
There are so many things you haven't done (I am referring to the ver in
the zip found on tunes)
Look at all the issues, how's your mem protection model going to work?
Selectors? As for IPC are you going to have distributed computing
concepts at the messaging primitve level? DO you know the ideal message
size? What effects will parallel processing have on your messaging
system primitves? Is it still an assembler, you're making yourself a lot
of work if you want to port it to all the platforms you want to..there's
large amount of device suppourt..where are you going to get all the
hardware for this...what about plug'nplay...all this for a bootstrapped
scheme interpeter? And you'll have time to write apps?
I have walked down this path before, and no ukernel will help solve the
numerous problems in software that have been cited in this group numerous
times; to mention them briefly, programmers spend a good deal of their
time doing what a computer could be, the inertia of large software
We need to introduce a sort of metric over ideas so that we can work with
them. Some say Scheme is good enough for this, I don't, if that's what we
really think than we should call ourself a scheme group. English also
isn't a good idea, simply for the fact that it makes the task of
representing and working with ideas harder than it should be (much of the
compiler science involves figuring out what you wrote, and doesn't have to
be that way). Is Brian's system the answer or full of meta babble? I
don't know, but just looking it over I feel that there is something
important there. And I believe that's where our focus should be,
Also we come from varying degrees of background in CS...I believe it is
possible to coordinate such an effort, but the goals must be written down
in one clear cut manner such that there is no ambiguity (which we are
soaked in right now).
Ok so what's the point? Before we do anything we have to explore concepts
of how to represent concepts...once we agree how to do it then we can go
on. For now I suggest that everyone read Brian's paper, read up on
catagorey theory and other background took...then we can decide.
Hopefully it won't take us long. I also want to us to get a group
agreement to stay away from designing a traditional OS...it's been done
many many times; so why doesn't everyone who agrees with the idea of
focusing on the metaprogramming aspect and leaving the OS design alone for
awhile please say so.
I don't mean to come across liek the big boss, but something has to be
done about the confusion and lack of direction and it might take a bit of
forced opinion to get there.