IRC log started Sun Oct  3 00:00:00 1999
[msg(TUNES)] permlog 1999.1003
12:10am
<eihrul> lar1: translate them into english... it's obvious
<eihrul> lar1: "i want i am going to the movies"
<eihrul> lar1: "i want to go to the movies"
<eihrul> now babelfish might surely say the first :)
12:30am
-:- SignOff eihrul: #TUNES (Ping timeout for eihrul[usr5-ppp11.lvdi.net])
-:- eihrul [eihrul@usr5-ppp11.lvdi.net] has joined #tunes
-:- SignOff eihrul: #TUNES (Ping timeout for eihrul[usr5-ppp11.lvdi.net])
-:- eihrul [eihrul@usr5-ppp11.lvdi.net] has joined #tunes
<lar1> Heh, I suppose yer right eihrul
<lar1> makes mucho sense
<lar1> Hcf:  Do you read Abi's logs?
<lar1> I gotta go
<lar1> I am mucho tired
-:- SignOff lar1: #TUNES (Uhhhhh no?)
-:- SignOff hcf: #TUNES (Leaving)
12:50am
-:- eihrul [eihrul@usr5-ppp11.lvdi.net] has left #tunes []
-:- SignOff air: #TUNES (BRiX [http://www.qzx.com/brix] :: sleep)
-:- smoke [smoke@14dyn26.delft.casema.net] has joined #tunes
-:- SignOff Tril: #TUNES (Ping timeout for Tril[bespin.dhs.org])
-:- Tril [dem@bespin.dhs.org] has joined #TUNES
-:- mode/#Tunes [+o Tril] by ChanServ
-:- water [water@tnt-9-246.tscnet.net] has joined #tunes
<water> hey
<smoke> hello !
<water> what's up, smoke?
<water> want to discuss something?
<smoke> i was just discussing the arrow system a bit with an 'oldschool bbs' person
<water> k
09:20am
<smoke> yes, what about?
<water> what did you discuss with that person?
-:- wedge [wedge@lossi149.uku.fi] has joined #tunes
<water> hi wedge
<wedge> hi
<smoke> water: applying an arrow alike idea to create a bulletin board system
<water> (shoot any wamp rats lately? :)
<water> ok
<water> any interesting results?
<smoke> water: not yet :(
<water> what did you come up with, then?
<smoke> water: i didn't get one thing right - does the arrow system have arrows pointing to actual data ?
<water> smoke: arrows _are_ actual data
<smoke> water: or is it just arrows pointing to arrows pointing to arrows, ad infinitum?
<water> smoke: it's the same thing. the arrows _form_ the data
<smoke> okay i got more than one thing not right then :)
<water> it's like that for most people
<smoke> how would that interface with say, ascii stored data?
<water> btw, arrows can be stored in the usual datat structures
<water> well, you have to have a way to derive the "arrow content" of a format
<water> s/format/scheme of representing info (or data)
<smoke> hm
<water> when you do, the interpretation into the arrow environment of the data can be implicit
<smoke> would it be hard to give an example on how to store say a number?
<water> er.. you don't "store" numbers, you refer to them... they have a priori existence
<water> you store an arrow from the "variable" to the number
<smoke> hm. but this variable is stored as a collection of arrows then?
<water> the entire number system will exist in arrow like a programming library
<smoke> or could it just be a memory-slot in a computer?
<water> bleh. forget variables
<water> forget "memory slots"
<smoke> hehe.
<smoke> okay, forgotten.
<water> think abstractly... computer=finite-state machine (with access to infinite-state devices)
<wedge> hm, data equals to transitions in this state machine? 
<water> a finite state machine has a set of allowable states and a set of transitions that can be performed on that machine
09:30am
<water> think of it as a graph of arrows
<water> states being nodes, and arrows being transitions
<smoke> but i shall not think of a value associated with a node?
<water> the idea is that you select the next datum in the sequence that specifies the data
<water> the node _is_ the value
<smoke> aha. so if i have 26 nodes somewhere, and attach a typewriter to those nodes directly, i could make a typewriter inside an arrow system?
<water> keep in mind that until i get the theory completely worked out, i can't give you complete answers
<water> yeah, sort of
<smoke> that is, those nodes could make a portal from the latin alphabet to the involved arrow system?
<water> yep
<smoke> did you experiment with implementations already?
<water> i have something on the tunes site for working with basic arrow logic, and my own internal versions with more structure
<smoke> i'll keep that in mind -- i'm just looking for how to think less vague about it :)
<water> it's written in smalltalk
<smoke> ouch.
<smoke> is there a nice smalltalk implementation for linux?
<water> yep. squeak
<water> (which is what my code is for)
<water> abi: squeak?
<abi> squeak is a cool language descended from Smalltalk, at http://squeak.cs.uiuc.edu/ or at http://www.squeak.org/ or The open source mouse that roars!
<water> thank you, abi
<abi> my pleasure water
<smoke> grin.
<smoke> bummer.. it's apl'ed
<water> "apl"?
<smoke> ehm i meant the apple license
<water> no
<water> the squeak image contains it's own source code
<water> you _can_ reuse it
<water> besides, disney is the only company supporting it
<smoke> confusing.
09:40am
<water> well, it descends from apple's smalltalk, which of course is the same as the xerox version
<water> in fact, it still has a version of the old GUI model
<water> but hardly anyone uses it now
<smoke> did you experiment with storing 'text' in an arrow system?
<water> not yet
<water> i'm still working out the graph and ontology systems
<smoke> ah. i have this idea that text compression will be either straightforward or completely nontrivial in an arrow system
<water> "completely nontrivial"?
<water> you would not care about text in arrow
<smoke> would not?
<smoke> hm. allow me to explain my thoughts a little:
<water> arrow would be used for storing info that the software could directly understand and re-use in arbitrary ways
<water> english text (not built from concepts, but from strings of characters) would not be as useful as arrow info
<smoke> i thought it'd be nice to store messages on a bbs system in an arrow system to make it easy to link information to other information and files easily
<water> you _could_, but that's not the best way to do it
<water> any good hypertext system could do that
<smoke> wouldn't be ? my thoughts were that with a decent trained neural network it could well be used, or even translated to a compact form
<water> the arrow system allows something more, and should be used that way
<water> keep in mind that there is an entire philosophy that the paper demonstrates
<smoke> of course i want more than hyper/text/
<smoke> ehm.
<water> ok
<water> i won't deny that using nn's with arrow would help with info-gathering
<water> but someone with an arrow system shouldn't preferentially use a text-based info-interface
<smoke> no?
09:50am
<water> they shouldn't have to type in comments for their code (for example), because the info should  be directly integrated with the system
<water> code=info
-:- SignOff wedge: #TUNES (Ping timeout for wedge[lossi149.uku.fi])
<water> i'm not saying that naming things is bad, i'm saying that the current way of doing it is terrible
<smoke> how then would people add information ?
<water> the interfaces would record their selections of meaning, not their character strings as such
<water> btw, interfaces = ontologies
<smoke> hm
<water> i don't know how to explain it formally except by saying that arrow interfaces will be "noiseless" ones
<smoke> so you cannot talk to a system?
<water> yes, you could talk to it
<water> but it would not store "what you say"
<smoke> ah i see
<smoke> then i have mis-explained my idea 
<water> it would store the info derived from "what you say" according to the ontology spec
<smoke> okay
<water> oh?
<smoke> that's about what i was thinking
<water> ok
<smoke> but imho it wouldn't harm to store the original messages too
<water> i disagree, since the messages could be factored into info and noise, and the noise stored and discarded according to user-preference, and the info used by the system
<smoke> uhuh
<water> hopefully, all the useful info could be derived from the noise
<water> hmm... time for some arrow info-theory
<smoke> so you think the idea of building a bbs on top of an arrow system is not very useful?
<water> no, it would be very useful
<water> but you would completely have to do away with the idea of a bbs as such
<smoke> but if only the 'useful' info were stored, then the social feeling would get lost, no?
<smoke> yes of course it's not much like an original bbs, agreed.
<water> well, i mean that the philosophies are "ideally" incompatible
<water> but you _can_ store the social feeling as info
<smoke> lol :)))
<water> keep in mind that the ontology idea is ultimately flexible
10:00am
<water> i'm serious
<smoke> sorry, i know you are serious - it's just that you speak so freely about very weird ideas. :)
<water> info theory applies to everything, not just data in computers
<water> i mean, not just "what the programmer calls data"
<smoke> hm yes.
<water> if you had a video-stream for a "video phone" conversation, the arrow system could abstract info from that, too
<smoke> yeah that would be nice
<smoke> that's what i somehow meant by text compression
<smoke> one could write text and only the important facets would have to be stored
<water> the point is that with the arrow system, you can control any information that comes into contact with it, no matter what kind of info it is or what the source
<smoke> but that would surely need some decent neural nets, no? for example to understand 2d representations of 3d images in case of the video phone conversation
<water> well, that means for the computer to do it in real time
<water> you could tell the system to store the data, and then later on do the info-abstraction yourself
<water> and use that to teach the system how to do it in real-time
<water> since the system would take info from your process of info-abstraction
<smoke> do you see neural nets as subsets of arrow systems, or do you prefer to consider them outside the arrow systems?
<water> neural nets could be applied and understood by arrow. the theory of the should definitely find a home within arrow
<water> i.e. nn theory is very useful info for arrow
<smoke> aha.
<smoke> that's good :)
<water> but since people can do what nn's do, a person could do the same thing in arrow and arrow would still get the info
* water/#tunes notes to self to include this example in the "optimization" section of the intro draft
<smoke> so, to get back to the bbs a bit; would it be feasible to have a system on, say, a 486 that stores data as arrows with a messagebase and chat-sessions as input ?
<water> yes, it would be feasible
<smoke> i can't get a clear image of the size of an arrow system
<water> ah
-:- tmf [s720@rovmaur.ii.uib.no] has joined #tunes
<water> by "storing data as actual arrows" do you mean lisp cons cells?
<water> hi tmf
<tmf> hi
<smoke> personally i'd think that about 10 links for each message would make for a nice hypertext system; yet i see more use in a system that can store about 1000 links.. 
10:10am
<smoke> (most of those links would be context generated of course)
<water> well, if you can derive the links by a "formula", then the storage req's for the links go down
<water> and the "formula" consists of info about those links
<water> this is why the notion of graph exists
<smoke> the storage isn't the biggest problem i think.. 100-1000 links for each item wouldn't at all be a problem.. with some garbage collecting a resulting 50-100 links would be enough
* water/#tunes thinks that smoke is missing some small clue here
<smoke> uh oh.
<water> you're really close to the big picture, but it sounds like you're not considering some aspect
<smoke> which aspect would that be ?
<water> the fact that most times, you will be dealing with infinite numbers of arrows
<water> (i think)
<water> (i mean, i think that's what you're missing)
<smoke> eek. infinite numbers?
<water> yeah, arbitrarily so. it's in the intro draft
<smoke> i surely missed that one. i thought one could just number each node and store that as a 32bit int
<smoke> -*- mode: naive -*-
<water> i mean, you could never put a limit on the number of arrows in the system. any infinity would never be large enough
<smoke> hm
<water> arrow ~= lisp
<water> :)
<water> i think that you were <actually> considering using lisp for bbs's, not arrow
<water> look in the "garbage collection" section of the draft
-:- SignOff smoke: #TUNES (Ping timeout for smoke[14dyn26.delft.casema.net])
<water> yeah, i should really say something there about how nn's allow seemingly "uncalculable processes to become calculable
<water> tmf: how goes it?
10:20am
<tmf> water, nn's seems close to arrows
-:- Miellaby [sgarden@Rennes-Villejean-1-154.club-internet.fr] has joined #TUNES
<water> miellaby!
<Miellaby> hi :)
<water> it's been too long
<water> tmf: huh?
<water> tmf: i hardly think so
<tmf> ok:)
<water> what's up, miel?
-:- smoke [smoke@14dyn26.delft.casema.net] has joined #tunes
-:- eihrul [eihrul@usr5-ppp185.lvdi.net] has joined #tunes
<Miellaby> well.. I'm asking for some news from Tunes's crew.
<water> wb guys
<eihrul> water: i finished reading your paper
<smoke> if i were to implement what i have in mind atm, i would probably use c++
<smoke> ? did i leave?
<water> miel: hmm... i guess i'm the representative for now
<water> smoke: yeah, ping timeout
<smoke> oh bummer
<water> smoke: chekc the logs
<smoke> water: (repeating), i wasn't thinking of using lisp at all. If i were to implement what i have in mind atm, i would probably use c++
<smoke> water: mostly it's that i want a system to store information which would be gathered from irc and email sessions.. and that data should be made available in a way less stupid as the www does it. :)
<water> smoke: well, i'd say that adapting the ideas of irc and email to arrow would be MUCH more useful
<water> from eihrul: 1. how do you decide the bootstrap ontology?
<water> from eihrul: 2. how do you introduce information into the system?
<tmf> ( good questions )
<water> for (2), a preliminary answer is in the "user interface" section of the draft
<water> as for the bootstrap ontology, i'm working on that right now
<eihrul> preliminary, but not definitive
<water> there is a lot of theory to work through
<eihrul> it just points out certain pieces of information as noise
<water> eihrul: sure
<eihrul> i would assume information is put into the systems as arrows itself...
<water> but arrows are abstractions
<water> they don't have to be data structures
<eihrul> yes...
<water> well, consider the example squeak code
<eihrul> but how much information does a 'mouse click' introduce
<water> ahh
<eihrul> would it modify the arrow structures or would it introduce a model of the mouse click....
<water> very abstractly, it could be the selection of a modality shift
<eihrul> if it just introduced arrow structures without giving information of the mouse click
<water> more concretely, you could be drawing an arrow, and selecting an endpoint
<water> (sic)
<eihrul> then you would have some small kernel doing that
<eihrul> which would be bad, no? :)
<water> er.. not a "kernel"
<eihrul> or you would atleast have to make that part of it reflective itself
<water> the ontology interprets the info
<eihrul> so that it could be changed
<water> right
10:30am
<water> well, the reflection is implicit for arrow
<water> info is info is meta-info is subject
<Miellaby> 'arrow' : what is it ?
<water> the amount of info a mouse click introduces depends on the ontology used to view the event
<water> abi arrow?
<abi> arrow is a homo-iconic information representation system intended to support programming and human languages in the way that Tunes should. at http://www.tunes.org/papers/Arrow/, http://www.tunes.org/~water/
<eihrul> where was the squeak code again?
<smoke> eihrul: www.squeak.org/
<smoke> oh sorry i see you mean the arrow squeak code.
* Miellaby/#TUNES thinking .oO ((what a get-up !))
<eihrul> yes
<water> http://www.tunes.org/~water/Abstract-Arrows.st
<water> you file it into a Squeak image
<water> which allows you to view it in a browser
<water> eihrul: maybe you should refine your question to, say, how the executable code might work, maybe?
<eihrul> i think that question is implicit in the whole arrow idea :)
<water> well, right now i implement it in smalltalk
<water> but the implementation isn't nearly complete
10:40am
<water> the idea is that the software in memory responsible for handling hardware input buffers would pass off the data to a sort of parser that would handle the info deterministically, allocating arrows in various graphs
<water> the graphs contain info which could potentially be used to generate the code that handles those input buffers
<water> make sense?
<eihrul> that much i've understood
<water> the actual implementation could be changed, since 'allocating arrows in graphs' is a very abstract idea
<water> any data structure which could be interpreted into arrows could store arrow info
<water> in fact, the code generated by arrow stores arrow info, but extracting that info would be very difficult
<water> which means that actual code wouldn't be kept around unless directly re-usable
<water> actually, the squeak vm code performs exactly that role right now
<water> any more questions?
<eihrul> in all of this i still see a small kernel, no matter how minute, of privileged code :)
<water> then you have no imagination
<water> abi: slk?
<abi> somebody said slk was the safe language no-kernel research project at http://www.cs.cornell.edu/slk/
<eihrul> do i, or do you?
<water> well, you'll admit that linux can replace it's kernel, right?
<eihrul> yes, and so could arrows
10:50am
<water> i'm sqying that this could be done at run-time
<water> i.e. without re-booting
<eihrul> it could
<water> there of course is a real difference between code that actually runs on the various machine processors, and data which is not
<water> but then, even code contains data
<water> i.e. binary code contains numbers, and addresses and such
<eihrul> i guess perhaps there would be no kernel, eventually
<water> i'm saying it's possible
<water> especially since memory-protection and task-switching can be done safely at the software level
<water> as long as you have a language (or system of langs) which enforces certain safety constraints
<water> well, i will be afk for the next 15 minutes
* water/#tunes is away
-:- ultima [ultima@user-38lcn4c.dialup.mindspring.com] has joined #Tunes
<eihrul> lo
<ultima> Heya
<Miellaby> hello
<eihrul> how goes the hardware project?
<ultima> s'ok
11:00am
<ultima> I'm pretty much condemned to ColdFire, but I'd kill to be able to use MIPS.
<eihrul> there's worse.... 8086
<ultima> not by much ;)
<ultima> ColdFire is completely adequate for my purposes, I'd just rather do it on mips.
<ultima> Instructions like beq just give me a woodie.
-:- Jacob [Jimmy@ppp204-17-109-202.online.sh.cn] has joined #Tunes
<eihrul> je is more finger-friendly
<ultima> eihrul: MIPS instructions are kewl though. beq is like beq operand1, operand2, address
<ultima> if operand1 == operand2, branch to address
<ultima> all in one instruction
<eihrul> how many cycles though?
<ultima> not sure.
<ultima> Shit shit shit
<eihrul> eh?
<ultima> $4k for a MIPS eval board
<eihrul> used?
<ultima> new
<eihrul> i meant why not look for....
<ultima> *mumble*
11:10am
<ultima> eihrul...you can get a fucking used SGI for less than these boards!
<eihrul> so why not?
<ultima> hmm
-:- SignOff Jacob: #TUNES (Leaving)
<ultima> la de da
<water> ultima!
* water/#tunes is back
<ultima> hey
<ultima> water, I don't suppose you know where I can get a cheap MIPS evaluation board?
<water> i don't suppose the old conversation thread still has any momentum
<water> hmm... various pda's use mips
<ultima> evaluation board
<ultima> I don't want a PDA...
<water> i know, but that's all i know about cheap mips hardware
<water> smoke, eihrul: still there?
11:20am
<eihrul> yes
<water> still want to talk?
<ultima> talk is cheap
<water> so is smalltalk! :)
<eihrul> but so are not mips
<water> lol
<Miellaby> ls
<eihrul> somewhat foggy as to how arrows transform into executeable code
<Miellaby> hups!
<water> eihrul: well, you can specify algorithms in arrow for any finite-state machine
<eihrul> you would need some translator that could take, say, a bitstream modeled by arrows and converted it into an actual one
<water> yep, which would be executable code generated to implement an arrow spec for it
<water> i honestly haven't implemented that yet, but i have a pretty good idea
<water> the basic idea is that the bootstrap system provides access to the binary storage device, and you have to model this situation and link it to the model (i think)
<smoke> re.
<eihrul> and the ultimate goal is to one day have arrow-native hardware? :)
<water> no
<water> hardly
<eihrul> well, it'd be nice
<water> it wouldn't do anything for me
<eihrul> it would theoretically happen if arrows became widely used, though
-:- air [brand@p0wer.qzx.com] has joined #tunes
<water> yeah, but at that point, i'd be moving on to my next projects
-:- SignOff ultima: #TUNES (Write error to ultima[user-38lcn4c.dialup.mindspring.com], closing link)
<Miellaby> bye:)
-:- SignOff Miellaby: #TUNES (Leaving)
11:30am
<water> bleh
<water> oh well
<water> smoke: ?
<water> anyone?
<eihrul> waiting for squeak to download
<water> oh
<water> there's a way to strip the image of unwanted code (safely) once you start using it
<water> you can cut down the size by about 50%, but for some reason they don't offer that version on the site
<eihrul> they could atleast gzip it
11:40am
<water> yeah, i'll post that suggestion to their mlist
<water> done
* eihrul/#tunes struggles with Netscape.
<water> just remember that the arrows code from the tunes site is very minimal. i run experiments on code on my own system
<eihrul> so why not post such?
<water> because it's usually some weird experimental stuff that i change around a lot
<water> i put code on the site to explain things, not make them more difficult
<eihrul> it can still be helpful if nothing else to clarify ideas about the arrow system by showing various things you're attempting with it
11:50am
<water> i disagree, because i've changed the code very many times
<water> i've re-implemented the entire thing about 5 times
12:00pm
-:- SignOff eihrul: #TUNES ([x]chat)
<smoke> smalltalk is a weird language
12:10pm
<water> what's weird about it?
<smoke> hm. it's me i think. i don't understand it.
<water> it's the closest thing to tunes that i've seen, as far as providing an environment that can be played with at any level
<water> what's so difficult? you send objects messages, that result in executing code
<water> and the code just consists of more messages or VM primitives
<smoke> there are so many characters that i can't place.. like ^ ! : and _
<water> ^ represents the return instruction
<water> _ represents assignment (shortcut for :=)
<smoke> hm
<water> i don't usually use !, so i'm looking it up
<smoke> almost every line ends in ! !
<water> oh, that
<water> that's the FileOut syntax
<water> don't think of it as code
<water> you never have to use it when you program, anyway
<smoke> i should use the browser?
<water> well, it helps a lot
<water> maybe i should have exported the class diagrams
<smoke> i haven't installed squeak yet.. i only have 5mb of free diskspace here
<smoke> (and on top of that i'm trying to install octave, which needs a lot of diskspace too)
<water> ouch
<water> what's octave?
<smoke> hm
12:20pm
<smoke> it's a math language for numerical computation
<water> oh
<water> i use MathMorphs under squeak
<smoke> i've been using matlab and maple, and was looking for an opensource replacement..
<water> MathMorphs :)
<smoke> hehe, you really like smalltalk, no? :)
<water> actually, i like squeak more than smalltalk
<water> i wish that they'd turn it into self
<water> they're doing it gradually
<smoke> ehm. what's self?
<smoke> oh wait that's sun's language. sorry
<water> it's a language like smalltalk, but without classes
<water> yeah
<water> well, not any more
<water> i mean, sun doesn't support it, and it's being ported to a bunch of architectures
<smoke> i have my backgrounds in assembler and C, so i feel very uncomfortable with all these 'newschool' languages..
<water> newschool?!?
<water> smalltalk is older than c
<water> and definitely older than c++
<smoke> i feel uncomfortable with c++ too :)
<water> hell, self's 11 years old now. it just doesn't run on anything useful
<smoke> funny fact, smalltalk and C both originate from 1972 :)
<water> smoke: you should :)
<smoke> the projects i'm currently working on are mostly about taking a simple machine to the max
<water> like forth
<smoke> hm i don't know forth
<water> http://www.forth.org/
<water> it's pretty much a high-level macro-assembler
<water> if that means anything
<smoke> sounds scary :)
<water> it runs in tiny memory spaces, and is very fast
<water> and handles i/o and just every kind of looping construct
<smoke> and its drawbacks? :)
12:30pm
<smoke> *eek* gazing at www.forth.org example code - the syntax to start with
<water> the memory model doesn't work well on large machines
<water> the keywords are short
<water> and it's not encouraged or recommended to re-use code: better to build your own specialized code
<smoke> well i guess that's not what i'm looking for :)
<water> well, code re-use in less than 64k is not the usual kind of thing
<smoke> on one side i want to write very fast code, but i'd like to have the big parts maintainable
<water> maintain"able" or easy to maintain?
<smoke> hehe :) easy to maintain
<water> that's a joke
<water> what kind of maintainability are you looking for? compared to?
<water> oh yeah, i forgot to mention that "_" in squeak is displayed as a left-arrow for assignment "<-"
<smoke> ah. i really ought to install squeak and see what it brings me :)
12:40pm
<water> MathMorphs has to be downloaded separately, if that's what interests you
<smoke> i'd like maintainability that will free mee from having to think about designing c++ classes until headache strikes
<water> well, then smalltalk would be the answer, unless you need executable speed, in which case you should get a smalltalk compiler
<smoke> i do need the speed :)
<smoke> i think i'll probably end up writing a lot specific code in C anyway
<smoke> is there a decent opensource smalltalk compiler available?
<water> hmm
<water> gnu smalltalk
<smoke> hm. i'll have a look at it then..
<water> but maybe they don't have a compiler
<water> well, squeak can convert code to c
<smoke> i have to go now.. thanks for the information
<water> there's already a project for dynamically-compiling squeak code to make primitives
<water> ok
<water> brb
12:50pm
-:- bdb [bdb@bdb.rezel.enst.fr] has joined #Tunes
<water> hello, bdb
<bdb> hi
<water> another frenchman
<bdb> oui
<bdb> :)
<bdb> just being interested by the webpage
<water> ah
<bdb> browsing does lead you to some strange places !
<water> we can answer questions for you
<water> hehe
<bdb> is TUNES available for my PC ? :-)
<water> :(
<water> no
<bdb> I was just kidding
<bdb> seems very interesting though.
<water> yeah, but our current problem is that Fare won't decide on a Lisp to use
<bdb> hehe :-)
<water> note: Fare and Tril are away
<bdb> I would suggect O'Caml ;-)
<water> ocaml is pretty good
<bdb> being a french, i was taught it at school by its inventors, and they were good at advertising it :)
<water> smoke: keep in mind that much squeak code must be "filed in" after being downloaded separately
<water> oh
<water> this explains why Fare likes it so much
<bdb> hehe :)
<water> it has taken me a while to understand him at all
<bdb> In fact, the browse that took me here was Inria -> ICFTP -> ENS Team -> Faré -> Tunes :)
<water> ah
<bdb> are you involved in the project ?
<water> unfortunately, it seems Tunes _does_ belong to Fare, though he believes otherwise
<water> sort of
<water> i discovered it on the web about 4 years ago, but kept separate until a year ago
<bdb> are there many similar os'es ?
<water> i work on an even stranger project called Arrow
<bdb> (in the building, i mean)
<water> there's UniOS, i guess
<water> but most aren't like tunes as  far as abstractness
<bdb> hmm
<bdb> abstractness is _cool_
<water> abi: arrow?
<abi> hmmm... arrow is a homo-iconic information representation system intended to support programming and human languages in the way that Tunes should. at http://www.tunes.org/papers/Arrow/, http://www.tunes.org/~water/
<water> s/representation/manipulation
<bdb> ...
<water> don't ask
<water> even Fare doesn't get it
<bdb> you're all kind of theoretical people here, hmm ? :-)
<water> well, i am
<water> and fare is
01:00pm
<water> maybe tril and tmf are too, in some ways
<bdb> do you know what they are having me learn at school at the moment ? the lap-b serial line protocol 
<bdb> :)
<water> but the rest are coders
<bdb> i see
<water> eek
<bdb> :)
<bdb> well, not only that, hopefully !
<water> that characterizes why i dropped out, anyway
<water> i couldn't afford to take classes like that while working on my project
<bdb> i could write your lap-b driver if you need it at some point :)
<water> er...
<water> we'll let you know
<water> :)
<bdb> :))
<water> right now, i'm concentrating on completing the arrow theory
<bdb> tell me when the book ships :)
<water> well, there's an intro on the tunes site
<water> (draft form, since the system isn't ready for release)
<bdb> yes, i'm sort of having a look at it.
<water> yeah, there eventually will have to be a book
<bdb> is tunes currently designed to run over unix ?
<water> well, the prototypes we've written run on linux
<water> well, posix
<bdb> ok
<bdb> i was thinking you probably hadn't started with assembly code :)
<water> god, no
<water> :)
<bdb> that's how os start, mostly !
<water> no, we go for lisp or scheme
<bdb> i see.
<water> well, we shouldn't write the os unless we can write the compiler
<water> since it's supposed to be reflective
<bdb> anyway, caml has too much keywords to be aesthetic -> prefer lisp...
<bdb> (or schemeà
<bdb> )
<water> yeah, it makes it more complex to reflect upon
<water> eval & apply for ever!
<water> :)
<bdb> :)
<water> actually, i'm also a smalltalk & self fan
<bdb> the scheme compiler in scheme is quite straightforward, i was told ?
<water> wait... brb
<bdb> i mean, interpreter
01:10pm
<bdb> bye all, gotta go !
-:- SignOff bdb: #TUNES ([x]chat)
<water> back
<water> darn
01:20pm
<water> cool! the Self/R page is starting to show results
01:30pm
-:- NetSplit: irc.linux.com split from lewis.openprojects.net [01:41pm]
-:- BitchX+Deb1an: Press Ctrl-F to see who left Ctrl-E to change to [irc.linux.com]
-:- Netjoined: irc.linux.com lewis.openprojects.net
-:- Tril [dem@bespin.dhs.org] has joined #Tunes
-:- abi [nef@206.63.100.13] has joined #Tunes
-:- Fare [fare@quatramaran.ens.fr] has joined #Tunes
-:- water [water@tnt-9-246.tscnet.net] has joined #Tunes
-:- tmf [s720@rovmaur.ii.uib.no] has joined #Tunes
-:- smoke [smoke@14dyn26.delft.casema.net] has joined #Tunes
-:- air [brand@p0wer.qzx.com] has joined #Tunes
-:- ServerMode/#Tunes [+o Tril] by irc.linux.com
<water> who's messing with the logger?
<water> is anyone here?
02:00pm
* smoke/#tunes 's here every now and then.
<water> k
02:10pm
-:- water [water@tnt-9-246.tscnet.net] has left #tunes []
-:- SignOff smoke: #TUNES (Zz.)
-:- hcf [nef@me-portland-us725.javanet.com] has joined #tunes
-:- SignOff tmf: #TUNES (Leaving)
-:- Downix [down@d-gnaps-134.ici.net] has joined #tunes
<air> Downix: hey
-:- SignOff Downix: #TUNES (Ping timeout for Downix[d-gnaps-134.ici.net])
03:20pm
-:- Downix [down@d-gnaps-71.ici.net] has joined #tunes
<air> any1 used infodisc cd-rw's before?
<Downix> not I
03:30pm
<air> 25pk for $13
<air> u heard of cmc magnetics cd-r discs?
<Downix> nope
<air> $7 for a 20pk
<air> do u use cd-r's at all?
<Downix> yes
<Downix> Sony CDU-928
<air> what kinda discs do u use?
<Downix> usually Sony
<air> heh samsung has a 700meg cd-r disc
<Downix> impossible due to the fact that every CD-R has the same max
03:40pm
<Downix> to make one that's larger would make it incompatable
<Downix> they are all 700+ meg unformatted, but formatting them removes a lot of that memory
<air> this is 700meg
<air> 80min audio
<air> very few support them
<air> mine doesnt
<Downix> mine doesn't I'll guarantee
<air> u probably need a new cdr drive to do 700meg
<Downix> I'm not buying one just to add 20 meg
<air> 50meg
<Downix> nopt worth it in the end
-:- eihrul [eihrul@usr5-ppp185.lvdi.net] has joined #tunes
<air> can u burn a cdrw to be a cdr?
<air> so it will work on most all cdroms?
<air> the damn cdrw's are $.50ea and the cdr's are over a buck
03:50pm
<Downix> nope air
<Downix> a cdrw is a cdrw
<Downix> a CD-R is a CD-ROM
<air> damn
<Downix> ok
-:- SignOff Downix: #TUNES ([BX] Dr. Kavorkian would KILL to use BitchX. Shouldn't you?)
04:00pm
-:- NetSplit: irc.linux.com split from lewis.openprojects.net [05:07pm]
-:- BitchX+Deb1an: Press Ctrl-F to see who left Ctrl-E to change to [irc.linux.com]
-:- Netjoined: irc.linux.com lewis.openprojects.net
-:- Tril [dem@bespin.dhs.org] has joined #Tunes
-:- abi [nef@206.63.100.13] has joined #Tunes
-:- Fare [fare@quatramaran.ens.fr] has joined #Tunes
-:- air [brand@p0wer.qzx.com] has joined #Tunes
-:- hcf [nef@me-portland-us725.javanet.com] has joined #Tunes
-:- eihrul [eihrul@usr5-ppp185.lvdi.net] has joined #Tunes
-:- ServerMode/#Tunes [+o Tril] by irc.linux.com
-:- Kaufmann [Kaufmann@200.224.105.148] has joined #tunes
<Kaufmann> Hey!
<eihrul> lo
05:20pm
-:- SignOff hcf: #TUNES (Leaving)
-:- water [water@tnt-9-232.tscnet.net] has joined #tunes
<eihrul> wb
<water> thx
<water> got ?'s
<water> i mean, do you have questions?
<water> ok, maybe not
<water> anyone want to talk?
<eihrul> aside from the bootstrap ontology... not much
<water> i guess it would be the "arrows and graphs" ontology, if it is one
05:40pm
-:- SignOff Kaufmann: #TUNES ("Jesus would make a good main character for a Seinfeld spinoff" - Amber)
<water> why do you ask?
<water> actually, my reflective arrow logic (with textual input) would probably be the first ontology
<eihrul> you have to have some way to make the system active... it can't just be completely passive data
<water> yeah, i'm working on that... the arrow logic part will address it
<water> Be right back
05:50pm
!Hyrinux:*! Abusing my wallops priviledge ... I'd like to ask if anyone is up for a few rounds in quake3 on a decent server...
-:- Kaufmann [Kaufmann@dial608.infolink.com.br] has joined #tunes
!irq:*! hang on
<water> back
<water> hey kauf, why'd ya leave?
<water> eihrul: reflective arrow logic quantifiers can introduce new arrow graphs
<water> and switch arrow contexts and, hopefully, ontologies
-:- SignOff Kaufmann: #TUNES (Ping timeout for Kaufmann[dial608.infolink.com.br])
<eihrul> indeed, but you need to start somewhere :)
<water> yes, which means i need "arrow primitives"
<water> or perhaps better, "graph primitives"
<water> now, i have classes for arrows and graphs
<water> but i haven't looked into how to make smalltalk-proper primitives
* water/#tunes looks into this
<water> eihrul: do you have squeak yet?
06:00pm
<eihrul> yes
<water> what do you think of it?
<eihrul> too naive with it yet to have such an opinion :)
<water> well, if you have questions, just fire away
<water> btw, i work in the Morphic UI, mostly
<eihrul> how do you run a given smalltalk file (i.e. your arrow code)
<water> you have to import the file...
<water> open the desktop menu, the select "open..." and "file list"
<water> then you browse until you find the right *.st or *.cs file, right-click on it, and select "fileIn"
<water> that results in the system parsing the file and creating the classes and objects in the file
<water> you then open a browser to look at them, or use a workspace to invoke them
06:10pm
<eihrul> no 'fileIn'
<water> what's the file's extension?
<water> if it's not *.st (say, *.txt), then you have to rename it to remove the extension
<eihrul> arrow.st
<water> hmm
<water> you right-clicked on the file name in the file list?
<eihrul> yes
<eihrul> ah
<eihrul> second mouse button
<eihrul> so not right, not left, but middle :)
<water> oh, you have all three mouse buttons
<water> all three are used for squeak
* water/#tunes uses Alt-right for middle
<eihrul> two button mice are not quite x11 friendly
<water> yeah, so i've found with my linux-box
<eihrul> know of any good smalltalk tutorials? :)
<water> hmm...
06:20pm
<water> i know http://www.squeak.org/ has links to a few
* eihrul/#tunes sees he will not go far with Squeak unless he learns smalltalk.
<water> well, the syntax can be seen from the browser
<eihrul> i can't even find the imported arrow objects :)
<water> er... once you select a method for viewing
<water> scroll down the list in the first pane
<water> at the bottom, you'll find "Abstract-Arrows"
<water> that's the category of my classes
<eihrul> not there... i'll try importing again
<water> what's the last entry on the list, then?
<water> actually, click on the first pane (middle button?) and select "update"
<eihrul> no luck
<water> darn it, i should have sent you my image... it has syntax-highlighting and is already set up for arrow
<water> is it working now?
<eihrul> i tried importing/updating a few times, but no... am reading a tutorial right now
<water> k
06:30pm
<eihrul> can you redefine operators in smalltalk as messages or are they just treated as primitives?
<water> "operators"?
<water> oh, arithmetic
<eihrul> yes
<water> they _are_ messages
<eihrul> that's what i wanted to know :)
<water> look into their code within the Numeric collections to see how they wrap around primitives
<water> yes, there are unlimited-length integers and naturals
<water> and rationals and reals ....
<water> and complexes, etc
<eihrul> nice
<water> yeah
<water> mathmorphs are even better
<eihrul> so if i wanted to play around with smalltalk, i just put code in a workspace then do what to run it?
<water> select the entire text, then click on the selection and pick "doIt" from the menu
<water> there's actually a system for programming graphically, but it's still a prototype system
06:40pm
* eihrul/#tunes is not very proficient with mice, anyway.
<water> you will be ;)
<water> actually, there are shortcut keys everywhere
* eihrul/#tunes got a trackball yesterday, so maybe not.
<water> well, that's what i have as well
<eihrul> much easier to target things with
<water> yeah
<eihrul> how do i print a value?
<water> hmm... like show 4+3?
<eihrul> yes
<water> select "print it" instead of "do it", silly :)
<eihrul> i just saw that, but there was an object you could send a message to to print a value
<water> what object was that?
<eihrul> cannot remember the name, i believe it started with a T
06:50pm
<water> transcript?
<eihrul> yes
<water> you can open a transcript window from the desktop
<water> i don't work with them very much, except when i'm toying with handwriting-recognition
<eihrul> Transcript show: 'foo'
<water> oh ok
<water> haven't worked with it
<water> ah.. i see
* eihrul/#tunes revels in the short-cut keys.
<water> of course... the same kind of code is what the handwriting recognizer uses
<water> which means it could be plugged into anything, potentially
<air> hmm stupid bochs wont let me press a key and step into the irq
* eihrul/#tunes wonders if you could boot Squeak as a stand-alone environment.
<eihrul> perhaps wiring it to OSkit?
<water> well, they've considered doing it
<eihrul> they did it with kaffe
<water> but they probably won't until the native-code generation project is done
<water> which will be a few months
* eihrul/#tunes ponders where BRiX is supposed to fit in with a native-code generating squeak.
<water> it could
<eihrul> disclaimer: sarcasm given above
<water> about booting it stand-alone?
<water> or about Brix?
<water> oh n/m
07:00pm
<eihrul> there is an irc interface in smalltalk?
<eihrul> er squeak?
<water> yep
<water> you'll have to get into morphic
<eihrul> even emacs doesn't have that :)
<water> from the desktop, open "project (morphic)"
<eihrul> done
<water> click on it to enter it
<water> now from the desktop open IRC chat
<water> btw, morphic is 10 times more powerful than the other UI you were using, but it's still in development
<eihrul> is there any help file in here? :)
<water> and if you see these dots pop up around a window, ignore them (click outside somewhere)
<eihrul> er, doh, everything in here uses the middle button 
<water> hmm... help for morphic...
<water> er... open a browser and read the comments for the source code
<water> :)
<eihrul> is there any way to make it cover the whole screen?
<water> squeak?
<abi> squeak is probably a cool language descended from Smalltalk, at http://squeak.cs.uiuc.edu/ or at http://www.squeak.org/ or The open source mouse that roars!
<water> yeah
<water> doh!
<eihrul> even if i resize the window it stays the same size
<water> when you resize, click on the desktop, "restore display"
<water> that's a VM bug that hasn't been fixed yet
<water> click desktop, then "appearance", then "full screen on" to make squeak full-screen
-:- eihrul-squeak [eihrul@usr5-ppp185.lvdi.net] has joined #Tunes
<water> lol
<water> hello again
<eihrul-squeak> so how do i resize the window? :)
07:10pm
<water> the squeak window or the irc window?
<eihrul-squeak> the whole squeak window
<water> just resize it like normal, and select "restore display"
<water> oh wait... you're in X11, aren't you?
<eihrul-squeak> there, worked...
<eihrul-squeak> you can use squeak outside of x11?
<water> hmm... i think that there's a version for it
<water> there IS one for bare dos
<water> and i know that you can emulate dos
<air> how the flip do i get outta this morphic thing
<water> lol
<water> air in squeak?!?
<eihrul-squeak> morphic is nice..
<water> what is the world coming to?
<water> select "previous project" from the desktop
<water> use "jump to project..." in the general sense
<eihrul-squeak> if i ever do decide to finish my kernel, squeak would make an excellent environment for it methinks
<water> yeah, it has a lot of potential
-:- SignOff eihrul: #TUNES ([x]chat)
<water> it still needs some improvement before the average person can really use it to its potential
<eihrul-squeak> /nick eihrul
<eihrul-squeak> ack
-:- SignOff eihrul-squeak: #TUNES (Read error to eihrul-squeak[usr5-ppp185.lvdi.net]: EOF from client)
-:- water-squeak [username@tnt-9-232.tscnet.net] has joined #Tunes
-:- eihrul [eihrul@usr5-ppp185.lvdi.net] has joined #Tunes
<water-squeak> testing, one two three
<eihrul> irc is bit clumsy, but otherwise...
<water-squeak> i can feed back bug reports to the squeak mlist
<eihrul> is there a squeak-terminal?
<water> yeah, but you have to download it
<eihrul> where at?
07:20pm
<water-squeak> check out the squeak site... there's a connection to a Swiki (smalltalk wiki)
<eihrul> if i could just port squeak to GGI, i'll never have to use X11 again :)
<water-squeak> it has links to all of the add-ons that people have made
<water-squeak> which reminds me, there's a special Swiki interface for squeak
<eihrul> in it now
<water-squeak> hmm... updating "self #nick" doesn't work
<water-squeak> i'll have to mention that
-:- water-squeak [username@tnt-9-232.tscnet.net] has left #Tunes []
<water> btw, there's a well-known performance lag with ports that squeak has, and a few coders are working on it
<eihrul> can't find squeak-term
<water> hmm
<water> i remember someone mentioning a t3270
<eihrul> found syntax highlighting, tho
<water> make sure to get the 2nd version
<water> the 1st won't work after image version 1.8 or so
07:30pm
<eihrul> squeak telnet
<water> telnet is under "goodies"
<water> i have almost all of the goodies in my image, so it's quite bloated
<eihrul> 6.4 GB hd's remedy that
<water> hehe
-:- air-squeak [brand@p0wer.qzx.com] has joined #tunes
<water> yeah, but my "bloated" image is about 6.3mb
<water> in fact, i have a squeak image ready for my Casio E-105 when it arrives
<eihrul> the image i got started at 5MB
<water> stripped down of course
<water> yeah, well i don't use speech-recognition very often, which would boost image size by a mb or two
<water> :)
-:- hcf [nef@me-portland-us135.javanet.com] has joined #tunes
<water> hcf!
<hcf> wat!
<water> i'm sharing the gospel of squeak
<water> :)
<water> i even got air to check it out
<eihrul> praise the vm :)
<water> eihrul: have you checked out the 3d scripting stuff? it uses a vm plug-in
<eihrul> nope
07:40pm
<water> it's in one of the "play with me" windows in the original MVC project
<water> it has a tutorial with it
<eihrul> dern... can't find TelnetMachine
<eihrul> even though i filed it in...
<water> hmm
<water> did you update the "class categories" pane?
<eihrul> i just tried to open it in a workspace
<water> oh
<water> it's probably not that simple
<water> open a browser and look through the source to see what messages it responds to
<eihrul> i can't find it in browser and the class isn't found either when i try to instantiate a new one
<water> weird
<water> i can't figure out how to run it
<eihrul> neither can i :)
<air> this irc chat is dumb
<water> i mean, the class exists, but i can't get it to open
07:50pm
<water> air: the squeak client or the content? :)
<eihrul> i tried 'find class' and i got nothing
<air> squeak client
<water> air: yeah. why don't you re-write it?
<air-squeak> testing
<water> :)
<air> its nothing more than a telnet client
<water> well, it does have some features, like user and channel lists
<water> and message windows
<air> oh wow
<water> eihrul: "TelnetMachine open"
<water> it works fine
<eihrul> "Unknown variable: Telnet Machine"
<eihrul> er w/o the space
<eihrul> it isn't loading the class at all
<water> oh wait
<water> what os/gui?
<eihrul> debian/x11
<water> the files have apple-style cr's
<water> there's a way to change the fileIn so that it handles them correctly
<eihrul> so i have to strip them out?
-:- SignOff air-squeak: #TUNES (Read error to air-squeak[p0wer.qzx.com]: EOF from client)
<water> no. i'm looking up the class variable that you have to change
<air> hmm nice quick message
<air> quit
<water> keep in mind that code gets improved in squeak extremely quickly
<water> there are usually about 5 bugfixes per day and 3 enhancements
<eihrul> stripped them out
<water> stripped what out?
<water> oh
<water> n/m
<water> actually, the answer is in the faq
<eihrul> stripping didn't help
<eihrul> will check faq
08:00pm
<water> http://minnow.cc.gatech.edu/squeak.565
<eihrul> found it
<eihrul> it doesn't say what to change though
<eihrul> er, n/m
<water> System-Files/FileStream(class)/concrete classes/concreteStream
<eihrul> found... <crossing fingers>
<eihrul> how do i commit the changes?
<water> trust me :)
<water> alt-s
<water> or select "accept" from the pane's meny
<water> menu
<water> btw, notice that the "?" box in the class pane will bring up class overall documentation
<eihrul> agg, still not working
<water> but you _stripped_ the cr's
<water> the cr's have to be there
<eihrul> but i output to diff filenames
<water> oh
<water> did the syntax-highlighting file in correctly?
<eihrul> didn't install that yet
08:10pm
<eihrul> squeak doesn't like this page very much
<water> what page?
<eihrul> the page the syntax highlighting was on
<water> oh
<eihrul> syntax highlighting is not loading
<water> hmm... and you changed FileStream #concreteStream to ^CrLfStream?
<eihrul> yes
<water> ugh
08:20pm
<water> don't know... maybe you downloaded in text mode instead of binary
<eihrul> nope, the carriage returns are still there
<water> that's really strange
<eihrul> could it be a bug in this version?
<water> maybe in your vm, but the image can't be the problem
<eihrul> i'm using 2.4c
<eihrul> should i be using a later version?
<water> there isn't one for linux
<water> i have 2.5, but it hasn't been ported yet
08:30pm
<eihrul> i can read the file in from a workspace
<water> cool
<eihrul> is there any way to take that and load it in?
<water> i don't know
<water> select all and "do it"
<eihrul> i think it worked...
<eihrul> it said "loading TelnetMachine.cs", haven't seen that ::)
<water> hehe
<water> yeah, that's what you should have seen earlier
<eihrul> i just did (FileStream oldFileNamed: 'TelnetMachine.cs') fileIn
<water> hmm
<eihrul> that's odd
<eihrul> that it works if i load file in workspace, but not from browser
<eihrul> easier to load from workspace anyway, though
<water> yeah
<water> well, check out the arrow code, if you could
<eihrul> the image will maintain all loaded classes, even if i exit, right?
<water> you have to "save" before exiting or "save and quit"
<eihrul> done
<eihrul> how do i get syntax highlighting working?
<water> it's that way for safety, since you won't discover a bug in a method until you actually use it
<water> just file it in and open a new browser
<eihrul> did
<water> check out a message
<eihrul> just monochrome
<water> (in the 4th pane)
<water> hmm
<water> hold shift down while de-selecting and re-selecting the mmessage once
08:40pm
<water> then de-select and re-select again
<water> (without shift)
<eihrul> nope
<water> damn
<water> you opened the browser after filing in?
<eihrul> yes
<water> :P
<water> why not try arrow? :)
<eihrul> arrows are loading, atleast
<water> they don't do anything by themselves, yet
<water> so no bugs so far :)
<eihrul> what is foo? 
<eihrul> or atleast in the context of an Arrow
<water> "SyntaxHighlightingParser doSyntaxHighlighting"
<water> just the state of a reference
<eihrul> mm... color
<water> cool
<water> yeah, it makes the syntax more clear for newbies
<water> make sure to notice the class hierarchy
<water> AbstractObject -> Arrow -> ArrowGraph -> ArrowWorld
08:50pm
<water> AbstractObject inherits from Object itself
<water> this allows me to treat a graph as the arrow that forms the parent node of its meta-graph
<eihrul> don't you get the same effect by subclassing ArrowGraph from Arrow?
-:- td [x@1Cust22.tnt3.wilmington.nc.da.uu.net] has joined #tunes
<water> that's exactly what i did
<water> hi td
<eihrul> then what is AbstractObject for?
<td> hey
<water> to allow the naming of objects.. it will provide the default user ontology for system objects
<water> it's only variable is "uiName"
<eihrul> /me foo
<eihrul> hrmm
<water> ?
<eihrul> not many features
<water> btw, reference is a class instead of a variable, because variables can't refer to their containers, whereas and arrow's references must be able to refer to their arrows
<water> nope. not yet
<water> features should stay in the objects as much as possible, not the classes
09:00pm
<water> i should put an ArrowWorld into a MorphicWrapper to make it into an actual project
<eihrul> i was talking about IRC-Chat specifically :)
<water> oh
<water> :)
<water> you have to disconnect and reconnect after doing "self nick: foo" to change nicks
<eihrul> wow, squeak browser even handles images
<water> gifs, mostly
<eihrul> no more netscape :)
<water> there's a vrml viewer, but it doesn't work with the web browser
<water> it's got a long way to go, though
<water> there's a java system and xml parser in the works, too
<eihrul> java squeak? shudder....
<water> well, it won't be part of the standard image
09:10pm
<water> the central squeak development has to do with native-code-generation and JIT compilation
<eihrul> must go do homework, bye
-:- SignOff eihrul: #TUNES (Read error to eihrul[usr5-ppp185.lvdi.net]: EOF from client)
<water> hmm... anyone else out there?
09:20pm
<air> ya i am
<air> i am sooooooooo close to having a keyboard driver in brix
<water> you mean the kb doesn't work yet?
<air> no
<water> well, that's re-asssuring
<air> it will in like 5 minutes
<air> i currently have a text mode screen driver that will support text windows in a few days
<water> ah
<air> also have a speaker driver
<air> :)
<water> so what's wrong with kb input?
<air> well the driver works fine
<air> its getting it to run
<water> oh
<air> but i have found several stupid typos i made in the kernel that were the cause
<water> hehe
<water> someday, i will have an os kernel where "typos" would be impossible to make
<air> its making bochs segfault right now when i press several keys
<air> and bochs kinda dont give any log file when that happens so i have no clue what happened and im forced to manually step thru it until it crashed
<air> crashes
09:30pm
<air> hmm i think tril should junk his os-help/ideas mailing lists and just add a pointer to the pmode-l mailing list
<water> hehe
<air> dont u agree?
<water> i don't look at that list at all
09:40pm
-:- SignOff Fare: #TUNES (Ping timeout for Fare[quatramaran.ens.fr])
-:- Fare [fare@quatramaran.ens.fr] has joined #Tunes
-:- lar1 [LarMan@1Cust5.tnt31.sfo3.da.uu.net] has joined #tunes
<lar1> Hey
<water> hey lar
<lar1> Hey water
<water> you old enough to drive yet?
<water> :)
<lar1> Me?  Nope
<water> yeah, i know
<lar1> 1 3/4 years left :)
<water> btw, don't come to me with your ideas and expect me to be interested until you've learned quite a bit more
10:00pm
<lar1> Water:  Who was that directed to?
<water> hmm. maybe it was addressed to the 14-year old people in the audience
<lar1> What should I learn more about then?
<water> anything... compiler design, language implementation, language paradigms, ...
<water> look into actual language implementations and learn how to use some of them
<water> learn about advantages and disadvantages of as many languages as possible before touting a language idea as "the solution" for problem X
<lar1> Are you telling me this because of my idea for that compiler?
<water> yes
<lar1> Last night I decided to drop the project simply because I don't know enough about optimizing, and I am not willing to learn it all right now
<water> i can point out a few really good books about optimizing
<lar1> I don't want to learn that right now
<water> well, optimizing assembly using software is probably the hardest way to optimize
<lar1> I want to learn more about OS stuff
<water> eek
<water> what's the hardest program you ever wrote?
<lar1> Mmmm, I don't know
<water> just an example
<lar1> Mandelbrot set generator?  I haven't done alot of really hard programming.  I know lots of how it works in theory...
10:10pm
* lar1/#tunes invisions water in a histerical laughter
<water> what is it you want to know about oses? how they work or how to code them?
<water> no, i'm not laughing, i'm trying to find out how to get you on a good track
<lar1> I figure I need to grok how they work before I can grok how to code them
<water> (although now that you mention it, "bwa hahahahaaa!!! :)
<lar1> Because all the time someone in here is talking about somthing and I go, oh crap, better add that to the design!
<water> so, you want to know how oses work in general?
<water> hcf?
<water> :)
<hcf> yes?
<lar1> I know how they work in general... I think
* hcf/#tunes is back
<water> can you help him with pointers about how oses work?
<lar1> I want to know enough to be sure that when I start up emacs I don't have to reimplement a whole bunca stuff because I didn't know about somthing
<water> lar1: sure
<water> abi osweb?
<abi> water: no idea
<water> www.osweb.8m.com , i think
<lar1> Is that a sure of sarcasm or a sure of aggreement?
<water> abi: osweb is at http://www.osweb.8m.com/
<abi> ...but osweb is at http://www.osweb.8m.com...
<water> hcf!
<water> geez
<hcf> beetcha :P
<water> :P
<water> :)
<lar1> Hey!  My OS is listed there
<water> wow.
<lar1> Umm
<lar1> what should I be looking for at this site?
<water> research projects, mostly
<lar1> Uhhh, where do I even start?  That is a pretty extensive list!
<water> hehe
<hcf> lar1: check out the nachos link(s) on tunes-rev-os
<water> yeah, that one's a good uK teacher
<lar1> Hmmm, eihrul has tought me to hate uKs...
<water> there's also the OS Ring, which has more HowTo info
<hcf> lar1: great, hate things u dont understand, good start
<water> abi: OSRing is at http://junior.apk.net/~qc/osring/index.html
<lar1> hcf:  I can only go by what I know
<lar1> hcf:  Don't uKs have lots of overhead?
10:20pm
<hcf> lar1: dont go by ppl's biased opinions, seek the truth
<lar1> How?  Every thing I learn will have a bias to it
<water> lar1: there are tech reports for all good research
<lar1> Now I am just lost
<water> where?
<lar1> My mind, I need to reboot it
<water> lol
<lar1> NeuroWare -- Build 14 starting... :)
<lar1> Where do I go from here?
<water> learn about different file-systems
<lar1> Ugh
<water> or memory management
<lar1> Ugh++
<water> or task-switching
<lar1> Yah!
<water> or thread support
<lar1> Yah!
<water> or even (dare i say) high-level language kernel-writing
<lar1> Isn't task switching easy?
<water> hehe
<water> oh yeah, it's a piece of cake ;)
<lar1> I had my task switcher all planned out
<water> how would it work?
<lar1> The timer would be set to quantum of each thread.  Every time the timer interupt would occur, the registers would be saved and the next thread looked up.  Its registers would be loaded and execution would resume.
* lar1/#tunes hides
<air> almost
<lar1> Registers being saved via software
10:30pm
<lar1> Tsses are too bulky
<air> u do need to check for sleep timers and load iobitmap tables for system threads
<lar1> air:  true
<lar1> And also if a thread needs v86 mode, I must switch to that
<lar1> water:  What do you think?
<water> i gave up that kind of os-writing a long time ago... it's too difficult and time-consuming for the value of the end-product
<water> (which makes me less of a coder, i suppose)
<lar1> water: How would _you_ do it?
<water> i _wouldn't_ :)
<lar1> So you don't switch tasks ever?
<water> serously, i'd put the functionality into the compiler or the language vm
<lar1> Huh?
<water> and leave switching within processes to the language
<lar1> Ouch!
<water> exactly
<lar1> That just hurt my head
<water> my idea of an os is not yours
<lar1> No, not at all
<water> (or most people's, for that matter)
<water> oses are lame
<lar1> How so?
<water> they require c or c++ or assembly programmers to work on them in secluded places for long periods of time in order to adapt to new circumstances
<water> or to just fix bugs
<lar1> How is your idea different in that respect?
<water> "lame" = hobbles around because programmers have to hold onto its legs and push them around to get it to move
<air> hmm 20 minutes to have keyboard support in brix or im gonna shoot myself
<lar1> air:  Whats to hard about keyboard support?
<lar1> s/to/so
<water> the language implementing the os should be simple enough for the user to work in, and dynamic enough to allow new kinds of semantics for the os
<lar1> You dont care about factors such as speed?
<water> yes, of course i do
<water> but speed is always relative to the task that something is designed for
<lar1> I see your point
<lar1> I want my os to do heavy gaming
<lar1> and anything the avgrage Joe would want to do
<lar1> on his home box
<water> hehe... for c-coding game programmers, of course
10:40pm
<water> anything that the c-coders let Joe do with their programs
<lar1> Whats that suppossed to mean?
<water> it means that if Joe isn't a c-coder, then he doesn't have a choice
<water> c-coders have power, and others don't
<air> lar1: there is nothing hard about keyboard support. its making my really nice event handlers fork threads for the keyboard driver and then making the keyboard driver fork a thread to the client that received the scancode
<lar1> Why don't forth coders have power?
<lar1> air:  Why so many forks just to deliver a scan code?!
<air> the event handlers are forking correctly now and i have found the bug that made the keyboard driver not fork the scancode
<water> they have a certain kind of power, but it is again limited by the language used
<air> lar1: thats the way brix works
<air> lar1: very flexible
<lar1> air:  Isn't that mucho slowo?
<air> no
<lar1> water:  But you are always limited my the language of the platform, right?
<lar1> water:  The opcodes of the chip stay the same...
<air> damnit, i bid $31ea for 8 1.2gig drives and was 3 of 49 earlier today and now im 31 of 42
<water> well, there _are_ fpga's, but i digress
<lar1> air:  fpga?  Is that like a bigass gal?
<lar1> air:  Ouch
<water> field programmable gate array
<air> auction closes in 14 hrs
<air> im gonna have to bid $34 each
<water> a processor whose (read-writable) microcode completely determines processor architecture
<lar1> water:  Like a whole bunch of generic gates and your code decides what each one of those gates become?
<water> yeah
<lar1> Why do you say nay to that?
<lar1> air:  What auction service do you use?
10:50pm
<hcf> abi: be water
<abi> nay
<water> hehe
<water> now, i'm a naysayer :)
<lar1> abi:  be lar1
<lar1> water:  Why don't you like fpgas?
<water> abi: be lar1
<water> but i _do_
<water> they're just not relevant
<hcf> lar1: add a saying to abi by using 'be lar1 is <reply> saying'
<lar1> hcf:  Ok
<lar1> abi: be lar1
<abi> be lar1 is AIEEEEEEEE!
<lar1> water:  How are they not relevant?
<water> we're discussing os principles
<lar1> water:  Doesn't the OS depend on the architecture, though?
<water> yes, but having an fpga cpu just allows the cpu to take over os functions
<lar1> Is that not reflection in a way?
<air> lar1: i use both ebay and onsale but im on onsale right now
11:00pm
-:- SignOff water: #TUNES (Ping timeout for water[tnt-9-232.tscnet.net])
<lar1> air:  Which do you perfer?
* lar1/#tunes has tried ebay and donsn't like it too much
<air> they both have good deals
<lar1> Onsale doesn't look like an auction place...
<lar1> Ohh, nevermind
<lar1> Heh, it would help if I clicked on the auction tab :)
<air> YES
<air> right after i tell the keyboard driver to resume the forked thread it will work
<lar1> Cool
<air> ok maybe just a little more debugging :(
<lar1> heh
11:10pm
<lar1> Whats with this rectuangular pixels in digital cameras crap?!
<lar1> I want _square pixels_!
11:20pm
-:- SignOff lar1: #TUNES (Leaving)
-:- SignOff hcf: #TUNES (Leaving)
-:- demoncrat [darius@shell.accesscom.com] has joined #tunes
<air> demoncrat: u got any 386's ?
<demoncrat> can't say I do
[msg(TUNES)] newlog 1999.1004
IRC log ended Mon Oct  4 00:00:00 1999