Pete Gonzalez | 1 Jul 03:46 2000

Re: Proposals

At 11:10 6/30/00, you wrote:
>Are you familliar with the Travelling Salesman
>problem?  
>
>[...]
>
>It has been proven mathematically that this is an
>N*P complete problem, meaning there is no way
>to solve the problem in less time than it takes
>to calculate all possible routes and pick the
>shortest one. 

Actually, it's theoretically possible that there exists a clever algorithm which can solve these
problems in polynomial time, i.e. the amount of time required to check a single solution.  This is the
"P=NP?" question, and despite years of searching we have yet to find a conclusive mathematical proof
either way.

It is most people's SUSPICION that NP complete problems can only be solved by brute force, but it would be
incorrect to assume that this is true.  History is full of surprising discoveries like simplex and
quicksort that contradicted peoples' intuitions about optimality.

-Pete

Lynn H. Maxson | 1 Jul 08:59 2000
Picon

Re: Proposals

Response to Jason Marshall.

"But we're coming to opposite conclusions based on 
the same data, which means one (or both) of us are 
not grasping the situation."

Or that one or both of us is not communicating well 
enough to show that we have the same grasp on the 
data.<g>  As I frequently find such failures on my 
part, let us assume that only one of us is to blame 
here.

"1) Exhaustive searches (searching all possible 
outcomes) lead to optimal answers, occasionally in 
suboptimal time."

True if we change the clause "searching all 
possible outcomes" to "as if testing all possible 
variations of the input".  Remember this is not my 
idea, but the one used in logic programming of 
which Prolog is an example.  This has been working 
20+ years with an approach that began with 
rule-based AI systems and has remained the same 
since.  

You are given an assertion.  You attempt to prove 
it false.  For it to be false you must "test" all 
possible enumerated variable states of the 
assertion.  Before you can do this you must insure 
that you have the logical means of performing the 
(Continue reading)

Lynn H. Maxson | 1 Jul 10:41 2000
Picon

RE: Proposals

Response to Billy.

"I'm still having a hard time figuring out what you 
mean when you say "an exhaustive true/false testing 
process."

If my this time I have not made the meaning clear 
in this set of responses, I have to refer you to 
any Prolog text or frankly anything which describes 
the logic engines of all AI systems.  I have read 
them.  I have used them.  I understand what "they" 
mean.  I'm only repeating here.  It is no invention 
on my part.

"You underestimate the complexity of modern 
programs -- a million lines is not even close to 
Windows NT's 50 million lines (ugh).  I forget what 
Win2K weighed in at."

First off I understand the difference between 
complex and complicated, the former dealing with 
the interconnections between units and the other 
with their absolute number.  I also know that these 
units have "names" and that in a manufacturing 
sense they exist as "assemblies".  Within them 
these assemblies consist of (explicit) references 
to other assemblies (who also have a name), control 
structures, and primitive operators.

I will go out on a limb and say not one of these 
(Continue reading)

Tril | 2 Jul 19:43 2000

Re: irc logs

On Fri, 30 Jun 2000, Youlian Troyanov wrote:

> 
> Please somebody fix the irc logs.
> 
> All the data since June 26th seems to be accumulated in 2000.0626
> and I am not authorized to view the page, IE5 says.
> 
> Also the new format (from June 25th for example) is hurting my eyes.
> Too many "[0m"'s.
> 
> Thanx,
> Youlian

Go to http://www.tunes.org/~lar1/logs/tunes/
I don't know how to fix the ANSI codes.  Does anyone have a problem with
replacing my log bots with lar1's?

--

-- 
David Manifold <dem <at> tunes.org> http://bespin.dhs.org/~dem/
This message is placed in the public domain.

Youlian Troyanov | 2 Jul 22:50 2000
Picon

RE: irc logs

Thanx. Now it's OK.

Youlian

> -----Original Message-----
> From: tunes-owner <at> tunes.org [mailto:tunes-owner <at> tunes.org]On Behalf Of
> Tril
> Sent: Sunday, July 02, 2000 1:44 PM
> To: Youlian Troyanov
> Cc: tunes <at> tunes.org
> Subject: Re: irc logs
> 
> 
> On Fri, 30 Jun 2000, Youlian Troyanov wrote:
> 
> > 
> > Please somebody fix the irc logs.
> > 
> > All the data since June 26th seems to be accumulated in 2000.0626
> > and I am not authorized to view the page, IE5 says.
> > 
> > Also the new format (from June 25th for example) is hurting my eyes.
> > Too many "[0m"'s.
> > 
> > Thanx,
> > Youlian
> 
> Go to http://www.tunes.org/~lar1/logs/tunes/
> I don't know how to fix the ANSI codes.  Does anyone have a problem with
> replacing my log bots with lar1's?
(Continue reading)

Alan Grimes | 5 Jul 03:48 2000
Picon

Re: OS

Baph000 <at> aol.com wrote:
> 
> I was browsing thru your website, and the OS section caught my eye.

=)  
Thanks for your feedback/response, I rarely get any and I value it when I
do. =) 

> In the news section, you mention completely rethinking your OS design 
>from the basic 386 pmode stuff to something more like that thing called 
>"tunes." What do you mean?

I'm not sure myself. I am only beginning my research on the new track,
which is so different from what I had been doing before. Lets see how I do
in explaining it. =P 

	Uh... I'm gonna assume that you are familiar with the concept of "Time
Sharing Machines". Okay you have a machine that can run several programs at
once, that is that the universal computer can magically morph itself into a
number of distinct and extremely complex machines simultainiously. The kind
of CPU in use at the time was this huge multi-chip module that cost
something on the order of $3,000. Around 1970 the video terminal became
practical so you would have like 30 of these hanging off of a machine that
had about the power of a TI-85. The problem is that when these people, who
were programming in BASIC, Fortran, and APL, and most notably assembly were
hacking on their programs on this machine, they would crash it mess up
everybody else's work. =P IBM then came up with a machine, I forget which
one, (too lazy to look it up), It was perfect. The hardware was such that a
simple program could command it to partition itself down into completely
independant "virtual machines" that could be created and deleted at will.
(Continue reading)

Lynn H. Maxson | 5 Jul 18:31 2000
Picon

Optimization, the gist of it

After a brief flurry there came a momentary pause.  
Whether that meant I answered things satisfactory 
or not, the pause occurred at an appropriate time, 
the Fourth of July weekend.  In celebration of this 
brief respite I thought I might offer a 
composition, a tympany in four movements in C 
major.

The first movement entitled "SQL Revisited" deals 
with optimization as it pertains to modern database 
usage.  Once we have laid it to rest as it were we 
enter the second movement entitled "The 
Experiential Use of Brute Force Intelligently 
Applied" or "E tu Brutus?".  Here we will lay to 
rest the issue of code optimization and whether 
available methods fall within the range of the 
possible (and solvable) or the impossible.

This leads naturally to the third movement entitled 
"Other Modern Miracles of Our Time".  Here we add 
to the roster of functional programming (Lisp 
variants, Forth, and OO) the additional entries of 
logic programming, APL2, and PL/I.  Treating these 
new entries as a Rule of Three we meld them into a 
Rule of One.

This in turn leads us to the fourth movement 
entitled "Operational Fare" or if you prefer "Fare 
Revisited".  Here we look the requirements for the 
Tunes HLL and transforming them into 
(Continue reading)

Brian Rice | 7 Jul 05:35 2000

Quick Note... Updated the Arrow-on-Squeak code

I recalled recently that we discussed improvements to the theory of Arrow 
back in October, and that I didn't get around to implementing the ideas. 
The new set of code is in the same location:

http://www.tunes.org/~water/abstract-arrows.st

This is compatible with at least every version of Squeak back to 2.0. It 
would probably work with very minor modifications in any other Smalltalk 
distribution (GNU Smalltalk comes to mind). Anyway, the specifics of the 
updates are that the CAR and CDR graphs are supported, as well as an 
application method in the graph classes.

Todo's:
1) Make an application (ApplyGraph?) graph class (this is a design pattern, 
but I don't recall presently its name).
2) Determine a basic system for quantification: Using graph application to 
follow one path of several, or to follow all paths "concurrently", or to 
return a graph of the resulting arrows (this was discussed earlier, but not 
to a definite conclusion).
3) Develop a basic system for infinitary / intensional meaning. This could 
be in the form of some boolean logic graphs or operator graphs that 
simulated the lambda calculus with some concurrency primitives.

I know this is a distraction from my work on Slate, but any progress in 
this area counts for both systems at this point. At any rate, all of these 
code additions should be child classes of Graph, using the usual OO design 
methods to keep things simple and understandable. (And yes, I acknowledge 
that this marks me as a perfectionist ;).

Thanks for the interest,
(Continue reading)

Brian Rice | 7 Jul 05:40 2000

Re: Quick Note... Updated the Arrow-on-Squeak code

At 08:35 PM 7/6/00 -0700, you wrote:
>I recalled recently that we discussed improvements to the theory of Arrow 
>back in October, and that I didn't get around to implementing the ideas. 
>The new set of code is in the same location:
>
>http://www.tunes.org/~water/abstract-arrows.st

That URL should be:
http://www.tunes.org/~water/Abstract-Arrows.st

I apologize for the inconvenience.
~

Gary Duzan | 9 Jul 02:50 2000
Picon
Picon

Schools for OS/PL/DS Research


   Hello everyone. Some of you might remember me from long ago from
the Moose project and early Tunes days. (If so, you have a better
memory than I do.) I've been on and off the list in lurk mode for
a while, to a large extent because I was busy with other things
and I didn't feel like I had the background to contribute. However,
I'm now looking to head back to school to work on a Ph.D., with a
focus on the application of programming language technologies to
building systems software and distributed systems. Of course, if
I manage to pull that off, I'd be insane not to be more active
here.  For now, though, I'm still looking for schools which would
have good support for these areas of research, and I'm looking for
suggestions. CMU, with their Fox project, is at the top of my list,
but I'd feel a lot better with a couple other places to look. If
anyone has any suggestions, I'd appreciate them.
   Thanks.

					Gary D. Duzan


Gmane