Advogato blog for DragonFaX

4 Sep 2000

Mon, 4 Sep 2000 08:40:04 GMT

harrisj read my shpeil about a simulated observatory room.

Linux reported X using 40 Megs, I was freaking until I looked
it up to find it was counting the 32M of ram on the video card.
Weird!

I won't actually decide whether or not to use Galeon instead of
netscape until I've compiled it myself and possibly added
a few features I need to have in a web browser. Then I'll have
a good idea of how to quantitaively measure its success for me.

* problems with getting the vpn working

Can't seem to connect to the office VPN. We use an Intraport
from compatible systems. Since bought by cisco and I think made
into the Cisco 5000 Concentrator. The ipsec implimentation
of win2k is messed up and and I just get connection errors
from freeswan (linux). Cisco provides no more support than was
originally supplied by compatible system. Which was a
kernel module that just crashs the system even when your running
the exact system the module was made for. I could probably
get the win98 client to work but I use 2000 and linux and
would have to triple boot just for office connectively which
anyone that mutliboots for one purpose knows is almost unacceptible.

I miss my bike. A freind back in missouri no longer has a car so I
shipped my dual suspension mongoose back there so he'd have a means of
getting to work. I still have my rollerblades though.

=Bits and peices

Halted has mice with an entire numpad on them for $10
what a useless peice of junk that I must have.

I like the DeCSS demostration at the Oscars idea even if just to
see what the other demonstrators yell when we're dragged off by police
"Let go of me you Damn Dirty Apes!"
"Stella!!"
"Attica! Attica!"

* Association for the Revival of Danger mouse.

Its the Plan 9 from Bell Labs "Brain Tumor Bunny"

"Politics and Professional Wrestling go together like and Cookies and Ass."
-the daily show, comedy central

"<Tiberious> get a taste of religion: bite a priest!"

nymia, objective c is yummy.

* TIMMAH!

27 Aug 2000

Sun, 27 Aug 2000 13:09:15 GMT

One of those PDA's should have enough horse power to emulate
a gameboy right? And someone could make a small adapter ( say
for the Visor as an example ) that would fit gameboy cartridges
right? Multitude of handheld games available for your pda
and legally too because your buying the physical cartridges.
Heck using cheap and slow memory they could even put rather
large games on these cartridges. Too large to fit in the pda's
memory itself. Its own form of copy protection. Or has this
been done already. Personally I think the new gameboy is a joke.

Also the VMS's (VMU's) that were born in the wake of the tamagotchi's
to carry small bits of games from the consoles with you seem
to have died a withering death. But slightly larger more
robust versions of these 'mini-games' to build your character
or even just get your RPG buzz while on the train seem like
they could have a place on your PDA. With a small link
cable from your PSX2 to your PDA you could download the
code and keep your characters strengths up. Or just down
load your saves to trade with friends.

I've come to the horrifying realization that colocation at a
decent price doesn't exist.

27 Aug 2000

Sun, 27 Aug 2000 12:41:38 GMT

What keeps me up at night is the realization that soon you'll be able to build custom virii
(biological) in your own home as easily as you can do with computer virii right now.
And that the script kiddies out there currently are putting out new major
virii weekly. We cant just reboot our bodies after a crash like we do with Exchange.
We are so dead.

I agree with Dacta
about galleon and mozilla. They should have started with the renderer and then just
worked up around that, making real releases of the features one at a time as
they became complete.

I have to read 'The Making of the Atomic Bomb' as suggested by
rachel

Searching for other places such as forums using peer review or trust metrics to control the content/noise ratio I've actually found several places that have done so far beyond what I would have expected. I see these sorts of practices absolutely necessary in the future as the internet becomes more and more bloated.

Tidbits

"P.S. They say they sell buffalo meat there, but I think it's actually
just a lot of bull."

- Rich Morin on sfpug@sf.pm.org

"Some people like buffalo, but I wouldn't want to buysome. (Weak, I know...)"

- Quinn Weaver, followup

a nice savage game of hunt the grumpus.

* dngor rolls 1d12, save vs. intelligence.

http://www.non-sequitur.net/archive/1992/02/nq920218.html

Limpidity #45: Alien Abductions

"Could this be the end of the Samurai Pizza Cats, or could it just be a good place
to put a commercial"

WikiTikiTavi!

19 Jul 2000

Wed, 19 Jul 2000 00:53:01 GMT

Your normal sense of depth perception is useless for
objects that are extremely far away such as mountains or the moon.
Instead you judge roughly how relatively far away it is by
the way it moves in the distance as you do.
Thus with litle effort you could simulate things at a great
distance. Take for example a room where you have a computer
track position of your head within it, then display across the walls of the room the stars and a well
rendered image of the moon and earth properly such that
you could simulate being in orbit to some laughably
simple degree. But if it worked even just half decently
it could be quite awe inspiring.
Would be need to try.

Read Snow Crash. Very good book.

Yeah Rachel! Kick ass article on Salon. Congrats.

18 Jul 2000

Tue, 18 Jul 2000 08:26:29 GMT

"This mind intentionally left blank"

I'm boggled by the fact that python ( a scripting language ) can
be so well optimized that they've not only written a complete OODB
in it (ZODB for Zope) but that it runs rather impressively fast
and is so full featured. Definitley something I'm going to have to
tear apart and look at the insides of while I learn python.

So people actually read these things huh?
Personal info for bma
I feel loved now. =)

Evolution of the Brain Organ

Jason J was talking about the fact that it was believed before that
reason we could not do proper AI was because of a lack of necessary
computing power (resources). But that thats no longer the case.
Now its believed that we no longer lack the resources necessary,
so that the only reason we have for not being able to emulate the
methods of the mind is a lack of understanding about how the mind
workds.

I think this is wrong. There are still many classifications of
intelligence. If what you are trying to emulate is the human brain
you have to be prepared to embrace and understand the imperfect
processes and incomplete intelligence that it provides.

The general idea is that we don't htink perfectly. This is obvious.
But there is a good reason why this is so. This allows us to
accomplish the leaps of logic necessary to do the thinking we do

They are trying to emulate what I see as 'perfect intelligence'.
Which is not going to work, especially when taking our minds as a
template. Many reason why this could not work, only one of which
is that because such a thing could be like an equation and would
solve to completion in each specific model case rather than provide
a system which could solve arbitrary problems.

Take the eyes as an example. We actualy don't see quite so clearly
as we think we do. But instead we see imperfectly, a low resolution
that changes around the eye. But to correct this our eyes jiggle
slightly in a hyperactive way now and then and thus take in more
information, several copies of the same view slightly misaligned
. Which our brain then interpolates them together to get one much
better picture.

You have to understand that while you as a person may be quite
partial to your brain as it contains, well, you. Your brain is
really just another organ of your larger body (organism). Each
organ evolved to suite a purpose. And with a long descriptiong
omitted for brevity, the brain evolved merely to be the 'problem
solving organ' One differentiates this problem solving organ from
most problems solving algorithms in computing are the variables
involved. You see to be really of any use the problem solving organ
we developed had to be able to, not decide the best way to trap
that animal, but rather the best way to trap that animal, in a
reasonable amount of time with a reasonable amount of resources,
and a reasonable ammount of work and a ton of other variables. Most
of which are purely qualitative rather than quantitative. Omitting
another long explaination your problem solving organ evolves into
the 'Ultimate Compromise Engine'. Figuring out the best solution
within a million different qualitative and always changing variables.
And in order to acheive the speed necesary in such fuzzy decisions
we had to give up calculator status. Actually we never had this
to begin with but its moot anyways. As proof we've seen people
with calculator status (autism) and their lack of ability to deal
with certain basic situations in life.

Obviously this isn't a good description of the idea but the full
one would take too long to explore here.

General Authentication

I've seen lots of people go on about new ways to securely authenticate
users for access to systems and resources. The problems are that
each method of authentication has its own issues. So the idea is
to use more than one and to weigh them. The more you pass each
one the more access you are granted i.e. the more permissions you
are given and the more you are allowed to do. Obvious this wouldn't
be gradual at all and everyone could set their own preferences.
Also there could be whole pass/fail mechanisms but would probably
not be necessary for everything that one does with her/her files.
Giving each method of authentication rating as to how easily they
can be faked, combined with percentages of how certain each method
is that the user is who they say they are, you could effectively
improve the security of less secure methods since the person must
pass more than one method. This would be especially useful in
passive authentication schemes such as wavelength measurements gene
tests, retinal scan and others.

Half-Qwerty

One big issue blocking everyone from moving to pda type devices
for most of their work is the lack of seriously useful input methods.
The small keyboards are too hard and slow to type on. The scribbling
is to inferior and slow as well. Plus quite counterintuitive since
they can't get close to reading regular writing. The only one I've
found to work yet is the half-qwety keyboard which can have quite
a small footprint but not require anything more advanced than
regular qwerty typing. Its the first small input space solution
I've seen that actually not only works but works very well. (for
me at least) I immedialy got 16wpm and that was with no practice.
I'd like to experiment with this a bit more but unfornately one
fellow seems to own a patent on it and chargs a ridiculous ammount
for the software, practically raping the one handed individuals
that it was originally designed for. (gee love people like that)
My ideal situation would be to build a small version of a physical
half-qwerty keyboard for connecting to a pda and see if I can
actually manage to take regular notes on it. I'm not much of a
hardware engineer so this will be a feat for me.

Got my atari trackballs and several of the books I ordered. While
I was down on castro (palo alto) I stoped in at bookbuyers, a kick
ass little used book store right next to Printers Inc. This place
is great, it has a large computer section for a used book store.
I'm going to spend alot of time there.

Barbequeue =)

The Barbeque was pretty cool. We had a decent turnout. Only 20
people or so. It was fun though. Got to talk to
John McCarthy (father of lisp, invent of AI)
for a little bit about the dissemination of micro cultures today
such as whats become of native americans. Then he went to watch
the opposite of sex in the other room.

Mr_Wrong(sean) had some interesting ideas on why databases suck
and what NEEDS to be in development right now to replace them.

Puzzle fighter was going until 4am.

I think we can make a case out of programming being an addiction
because of what some of us go through to do it or while doing it.
There also might be some reason to link a discussion of a 'programmers
high'.

"Hi I'm Bob. And... I'm a programer!"

"HI BOB!"

"I first knew I had a problem when..."

Outsourcing Opensource

I've notice a recurring problem as to how the general ideas of
opensource are continually being represented to managerial people
in magazine articles and other resources. What I keep seeing are
claims that opensourcing your software will get literaly thousands
of developers to work on it for free. In some cases this is not
exactly untrue. But they tend to go so far as to push that you'd
have the programming culture of the world working on your software
for free from home rather then them having to pay developers to
get the work done. And this not only puts up a bad image but also
pushs a very sad mentality that can actually harm the promotion of
opensource ideals. I'll agree that some companies opening their
software see quite a flush of individual developers working on it
as part of a hobby.

However the idea is not free software development for your company
or its product. Rather its peer-review and development. That
companies (and individuals too )that also can benefit from the
software will be able to work on one copy, sharing advancements
and bugfixes to their mutual benefit. Rather than continually
reinventing the wheel. Pushing the idea as they are that opening
your souce will take the burden of development off your hands and
put it on the backs of so many programming hobbiest is bad for open
source and quite counter-intuitive when managers attept to seriously
look into what opensourcing their software really means. Its a bad
mentality that shouldn't be promoted. What we should be trying to
do is convince companies to invest somehow in the open source
project that can benefit them. Unfortunately since all of this is
still new is extremelly complicated for any specific company to
even attempt to invest in any opensource software let alone opening
any of theirs.

Now with companies turning toward outsourcing everything from
development to managerial staff and in some cases even HR it would
be interesting to work up some business plans on the idea of allowing
companies to outsource open source development or investment. We
already having companies that have a core part of their business
plan opensource development. What we could see are some companies
that serve as outsources for the complicated parts of opensource
development that complexify the relationship between companies that
don't want to have to rearrange their entire business just to make
it work with the inrush of the freesoftware community. Then such
companies could focus the efforts that I know non free software
companies should and will want to take full advantage of this
revolution thats happening. Unfortunately this could not be an
easy task. it could often require putting pricetags on feature
improvements and such in open source development. But could still
work to promote the serious side of OS investment that companies
will need to start making.

OpenNic

UnRated Net: OpenDNS

I think if we could offer just a generally better scheme than what
InterNic does, we could convince enough people serving DNS to patch
their copies of BIND to request from both NIC and then OpenNIC if
not found. Or just OpenNic as they also include Nic's info. Who
is in charge of the bind and other named sources? There is no reason
for nic service to be so bad. Thought I'm still in the minor
opinion that the idea of specific TLD's is outdated anyways. Now
if we can only figure how to do this with pacbell.

X-Men kicked ass.

[description of a GUI method idea left out cause it was too long]

14 Jul 2000

Fri, 14 Jul 2000 08:25:04 GMT

BBQ at my place on saturday!
BBQ at Dragons!

I've been looking at plex86, the open-source vmware clone.

I believe that something like this has a great deal of potential.
We've needed something like this for a long time on x86 systems.
Vmware is unfortunately too restrictive as it has almost no growth
potential in the small company that owns it.

Other architectures and OS' have similar packages such as Mac-On-Linux
and Sheepshaver. Both for the ppc architecture. But the Intel
processor architecture is so badly designed that this hasn't been
possible before. At least not easily. This could be especially
useful if we are stuck with Intel as the popular commodity system
for quite a while.

I was thinking about was the idea of io pass-through for hardware.
Lawton mentions this briefly in a discussion on the difficulties
of developing this kind of software on an platform such as this.
But I dont' think anyones given this possibility nearly as much
credit it deserves.

First off this could be an alternative to getting all of that
complex hardware emulation to work right.

Enabling you to use the original driver for the extra hardware
present in your system rather than having to write up a complete
emulator for that hardware. Having extra video cards, io cards,
and hard drives, and taking full advantage of the fact that your
host OS isn't currently driving them.

Now of course its just not this simple. But io pass-through might
be easier than hardware emulation and reduplication in the host
OS.

The main advantage, at least in the beginning, could be simplified,
easier development.

Duplicating hardware for stable testing purposes without having to
worry about:
1) the unstable code messing up the current running system
2) translating the meaning of what the guest OS
wants to do onto the host OS' hardware abstraction
i.e. blocking devices, file-systems and such.

After I had considered this for a while I found that Lawton was
the fellow responsible for bochs which is effectively a collection
of simple hardware emulation. So this obviously gives him a head
start on the other track.

With SMP becoming more popular there has to be a lot of advantage
that could be taken from having a dedicated second processor to
play with for a guest OS. At least some optimizations in all the
code scanning that is necessary in plex to overcome the limitations
of the Intel processor architecture.

I realize that the feasibility of these ideas are very low. But
its believable enough to get me to look at the code myself and see
if theres something I can try quickly. And anything that gets
someone to help with an important OS project is good. Right?

Of course and then theres the standard idea of a plug-in-architecture
for what hardware you want. Both emulated and io pass-through.
But this isn't something you think about until you have a good
working system already in place.

I'm always looking for good books suggestions. The best suggestions
are books that shaped the way you think or program. Some of the
most recent suggestions I've received are

Literate Programming
Alice's Adventures in Wonderland
The Mythical
Man-Month
The Practice of Programming
The Art of Programming
A Pattern Language: Towns, Buildings, Construction
Godel, Escher,Bach: An Eternal Golden Braid
Software Engineering Economics
Peopleware: Productive Projects and Teams
Computer Architecture: A Quantitative Approach
Computation: Finite and Infinite Machines
The Elements of Style
More Programming Pearls
How Buildings Learn:What Happens After They're Built Writing
Efficient Programs
Donald Knuth's 'Selected Papers' series
The Cluetrain Manifesto
Introductory Psychology Through Science Fiction
The Practice of Programming
Code Complete
Rapid Development
Writing Solid Code
The Pragmatic Programmer
Wicked Problems, Righteous Solutions
Extreme Programming
Object oriented perl
Structure and Interpretation of Computer Programs
The Evolution of Useful Things

More suggestions are always welcome I've recently purchased
Programming Pearls, Cathedral & The Bazzar and several others.

There dont' seem to be any good places for complex & organized
discussions about the current issues of software licenses and
licensing. This concerns me greatly because I think these issues
are whats going to shape a large part of the future of computers,
business and otherwise. For instance its obvious that the future
of business is going to be in implimentation and not in protocol.
But there are so many companies trying every which way they can to
keep the old checks in place so that they can continue to make
money of the protocols they have been for 50 years. That its actually
starting to hurt others. I was thinking of starting by setting up
or locating a bulletin board to be dedicated to this kind of
discussion. And then follow this up by working on a system in which
people can develope licenses and documents in groups just as source
code is developed. Obviously there are alot of issues here though.
And these kinds of things have been done to death in the past But
what I'd try to focus on is the discussion rather than the actual
changing of the work in progress.

"System for the Collaborative Development of Low Volitility Documents"

Using peer review to engage the evolution of such documents might
allow us some better understanding of the future we're about to
face in Intellectual Property.

The system would be a repository for such documents as well as a
complex way of orienting a bulletin board type system such that
messages can be directed towards the document as a whole or almost
any specific part of the document, regardless of how large or small
that might be. The difficulties are in how to arrange in an intuitive
manner such a cross referencing situation.

I guess I'd use angryprogrammer.com

I'm using barkingweasel.org for my own site. Frikkin network
solutions won't expire the registration for barkignweasel.com
regardless of how uninterested the original owner is in renewing
it. So I can't snatch it up for myself. Bastards!

Looking into zope for a solution.

I'm still baffled by the paradox that is Pacific Bell. In what
other industry can there exist a company, which every one of its
customer hates profusely. Every single one!

I didn't get to bed until 8am. Why do computers have to be so damned
fun.

Have to look more into functional languages like haskel and caml.
Also have to learn objective c serious and look into sather. I'm
extremely interested in low level programming languages with the
high level features of memory management.

Need to get GTKtalog running on a machine so I can make indexes of
all these damned burnt cd's laying around. I can never find anything.
Wish it would run on freebsd well. I got it to compile and start,
but I had to hack the code heavily and it would still need some
code to drive the cdrom. Should still be remedial but time consuming.
Also to really make full use of it I'll have to come up with some
way to have it recursively work on particular files. Decompressing
files and working on those contents as if it was a directory and
then repeating the process on any compress files in found. (i.e.
treating compressed files as directories with a great deal of meta
data )

Gnustep is such a great thing. And they've made such great progres.
I'm boggled by the fact that I don't hear anyone talk about it or
anyone use it.

Looking at the FAQ for sather, another interesting looking language
I haven't had the time to get much into yet I started to ponder
how we're defining languages now. There are some umteen million
(gives or take a few) languages now. There seems to be several
different parts to what a language is, and syntax just one of them.
Not only that but the other parts seem to be completely independent
of the design of the syntax. This would leave me to believe that
theoretically you could redesign languages so that the syntax was
personal. Or specifically that it should be possible for one to
describe what 'features' of syntax he/she prefers and thus design
ones own 'look' of a language. Example syntax features would be
the much debated tab based block definition vs curly bracket ( or
some other identifier ) based block delimiting. If taken far enough
one could specify a syntax as strict as python or java or as loose
and maliable as perl, or for that matter any variation in between.

A further stretch of the imagination takes us to this point. Lets
saw we have language syntax features that we consider as shorcuts,
simplifications or some such. That could theoretically be broken
down into longer but much more explicit code. (btw, such features
seem to be the core of perl syntax design) Then if we could define
the algorithms that would expand uses of these features to the
longer but more explicit code, we would have a basis to result any
particular use of these features down to a common denominator.

This makes no sense so let me try to explain this again. Shortcut
operators. Those funky pipes and ampersands that let us do pseudo
boolean execution of statements but let us get away with only
executing the second statement (or third or forth etc) if the first
one returned what we were expecting it to. These are extremely
simple (and short) version of what we could write logically with
a group of if then's. But we have this shortcut for the soul reason
that we use them often. Lets say you were to enjoy using this
particular feature. So you set in your language prefs or at the
top of your source (or Make) file your code will use this syntactic
sugar. Then someone else comes along who thinks this complexifies
code making it hard to understand and prefers not to use these
handy operators. Rather than forcing him/her to deal with your
travesty of what real programming should be he can merely specify
a small command that will go through his/her copy of the source
and expand those handy dandy pipe dreams into something that is
not quite so short and simple but that he/she can understand more
readily and won't find an eye-sore. Take this and apply it to what
makes up most popular language syntax and you have a programming
language that can look like almost any other type of language you
want it to. Allowing each individual programmer to program in the
way he or she decides best, but that can still be devolved into
some common denominator language specification that everyone can
understand. The volume of default arguments in perl alone that
could be exposed is particularly horrifying.

Well, I probably don't know what I'm talking about, but thats not
going to stop me from considering it further.

Something I'm sure about is the fact that this kind of slicing and
dicing of bits could not be done with the actual non syntactical
language features such as memory managment, typing, garbage
collection, and objects withall the different and harried ways of
implementing and manipulating them. But! And thats a might big
but you have there Mr Stillwell. If one could manage to do so with
a few core language features then one would have an interesting
candidate for a self optimizing languages. This goes even further
though add that pseudo native binary, pseudo intrepreter idea that
seems to be really popular nowdays and you could end up with a
language that not only is all things to all people but actually go
through being every stage in the gambit from extremely high level
(prototyping and such) to very low level ( effectively c ) all in
the development cycle of one project. Thereby simplifying development
as you start with an interpreted extre featured and rather easy to
debug (yet slow) interpreted language while your still prototyping
and nailing down what your trying to write in the first place and
then slowly make your way down (still in the same code base) to a
low level no frills c end result. Heck at that point if you'd
already nailed down your features you could keep track of the
transitions in a new wave version of cvs and go back up in the
train of revisions to something more high level when you need to
add a new feature or something like that. Doing horizontal code
changes early in the life of the code long after the code is even
finished.

Anyways, thats enough theoretical language design. heres some
links

My wondering why I let my hair grow long.
http://adsl-63-195-163-198.dsl.s nfc21.pacbell.net/~dragonfax/webcam32.jpg

The Perl Filesystem
horrors!

Powerpuff Girls

13 Jul 2000

Thu, 13 Jul 2000 14:49:52 GMT

I've actually started submitting diary entries.
Lets home I can come up with something to put here.

I setup webdav on barkingweasel.org and angryprogrammer.com.
The idea is if its easier to post content there, then I might
actually get around to making some.

I also wrote up this quick little script to let me post
my advogato diary from the command line. Lets hope it work. =)

13 Jul 2000

Thu, 13 Jul 2000 14:41:21 GMT

I can name three cheese.