IRC log started Sun Jan 17 00:00:01 1999
[msg(TUNES)] permlog 1999.0117
ωνω SignOff _QZ: #TUNES (BRiX [http://www.qzx.com/brix] :: )
ωνω mti [medtib@ppp-195-36-197-223.pop.club-internet.fr] has joined #tunes
<mti> hello
05:00am
ωνω SignOff mti: #TUNES (G3/583MHz :))
ωνω Netjoined: moore.openprojects.net koontz.openprojects.net
ωνω miguel [miguel@tirania.nuclecu.unam.mx] has joined #TUNES
ωνω NetSplit: moore.openprojects.net split from forward.openprojects.net [07:04am]
ωνω BitchX: Press Ctrl-F to see who left Ctrl-E to change to [moore.openprojects.net]
ωνω Netjoined: moore.openprojects.net forward.openprojects.net
ωνω miguel [miguel@tirania.nuclecu.unam.mx] has joined #TUNES
ωνω smkl [sami@ppp81.dial-in.verkkotieto.com] has joined #tunes
ωνω Fare [rideaufr@esmeralda.enst.fr] has joined #Tunes
<smkl> hi Fare
<Fare> lo smkl!
09:10am
ωνω tcn [tcn@bespin.ml.org] has joined #tunes
ωνω SignOff tcn: #TUNES (tcn has no reason)
ωνω _QZ [brand@p0wer.qzx.com] has joined #tunes
<smkl> hello _QZ
<_QZ> hello
10:10am
<_QZ> today is a good day, i got my celeron 300A to run stable at 504mhz
<_QZ> now im gonna test my luck and make the modifications to 2 300A's so they can do smp
<Fare> what's hardware power worth without proper software directions?
<Fare> QZ: what about Brix?
<Fare> does it manage keyboard correctly?
ωνω hcf [nef@escher.sdi.agate.net] has joined #tunes
<_QZ> Fare: how would it not manage keyboard correctly?
10:20am
<Fare> QZ: dunno, like, fail to function on some machines, while working on others.
<Fare> QZ: what's the current status of Brix?
<_QZ> its good
<_QZ> my key function should be stable
<_QZ> i wrapped it in a dos program and have sent it out to many ppl that needed keyboard code
<_QZ> no one has complained about it
<_QZ> ok time to put on lots of warm clothes and drop the room temp 50 degrees so i can power up my 300A at 504mhz :)
10:30am
<Fare> QZ: great!
ωνω ultima_ [ultima@user-37kbb1u.dialup.mindspring.com] has joined #TUnES
10:40am
<Fare> ult: are you still maintaining review?
<Fare> ult: what's the current status of what you're doing?
<ultima_> Fare: The current status is that I'm looking for human slaves to help me with something
ωνω SignOff smkl: #TUNES (argh)
<hcf> ultima_: im doing the os section
<ultima_> Fare: I need to convert all the entries into plaintext and check all the links
<ultima_> hcf *nod*
<hcf> doing as in link checking
<ultima_> Fare: I've already noticed a substantial number of incorrect links and such =(
<hcf> ultima_: i already hav a substantial number of correct links
<ultima_> hcf: clap clap clap =)
<Fare> ult: do you follow CVS updates?
<hcf> in the past 2months, i redid my own list o links, which has current links for a lot of what the review pages link to
<Fare> why plaintext???
<ultima_> Fare: Because i'm incorporating it all into a database so that it will be autogenerated and have a uniform look&feel
<Fare> we DO need hyperlinks as well as <UL> and <DL> stuff!
<ultima_> Fare: The database contains the name, the brief description, the long description, and related links, and that information is used to generate the project. I'm working on html, tex, and sgml output formats
<ultima_> Fare: Why do you need stuff like <UL> and <DL> in the data?
<Fare> ult: what about things more elaborate, like the LISP entry?
<ultima_> Fare: Well, thats the difficulty i'm having, you used too many non-uniform ideas
<Fare> ult: it has lots of sublinks in subcategories, plus comments, criticism, etc
<Fare> ult: I had non-uniform knowledge of languages with non-uniform interest!
<ultima_> Fare: Yeah, and it shows!
<Fare> you just can't decreet that things are uniform when they are not, and cannot be!
<ultima_> Fare: I'm not saying the languages should be, but the descriptions need to be
<Fare> why need they?
<ultima_> Fare: Readability, I want it to look rather like an annotated bibliography of sorts
<Fare> the format should be adapted to the contents!
<ultima_> Fare: Well, your 'contents' have a tendancy of being rather biased
ωνω NetSplit: moore.openprojects.net split from koontz.openprojects.net [10:48am]
ωνω BitchX: Press Ctrl-F to see who left Ctrl-E to change to [moore.openprojects.net]
<Fare> putting lots of structured text (as in the lisp entry) in a raw unstructured plaintext is DULL, and makes it *less* readable!
<Fare> of course they are biased!
<ultima_> Fare: I never said it was raw unstructured plaintext
<Fare> bias is good if it's done with an openmind, and considers information a posteriori
<Fare> rather than dogma a priori
<ultima_> Fare: html is a form of plaintext that contains markups. I dont want to use html markups because that makes it too difficult to generate .tex and such files
10:50am
<Fare> ultima: what about texinfo?
<ultima_> Fare: .tex can be converted to texinfo
<Fare> that's the opposite: .texinfo can be used as .tex!
<hcf> why is the generation of '.tex and such files' necessary?
<ultima_> Fare: Whatever. I dont know texinfo, I know latex and html. Those are the output formats currently planned for. If you know of a easy way to generate texinfo, I would happily use it
<ultima_> hcf: Well, right now its limited to html, and for one thing, I dont like the way html prints out
ωνω Major [major@orion.jimco-fwt.com] has joined #Tunes
<ultima_> hcf: I'd like to have html for online browsing, and alternative formats for printouts
π Major/#Tunes ponders.
<ultima_> hcf: Also, tex can be converted into pdf easily, and pdf is a quite nice online and offline viewing format
<Fare> QZX: the brix source is unavailable???
<_QZ> no really?
<ultima_> Fare: He ripped most of it from linux so he refuses to distribute it
<hcf> ultima_: latex sounds fine, whats the prob?
<ultima_> hcf: Uh, I honestly dont know. From this end, it all works out quite well!
<ultima_> hcf: Fare just doesnt like it because html gives him a woody or something
<Fare> Major: for introductory lisp, well, there's the TeachScheme project
<hcf> ultima_: do u hav a sample of html output?
<Major> hmmmm
<Fare> and/or stuff in the c.l.l faq, and/or stuff from the pages pointed at by the LISP Review entry...
<ultima_> hcf: Not yet. I need to figure out exactly what kind of tags need to be in the database text so that I can output correct-looking entries
<Fare> Major: do you wanna learn Scheme? CommonLISP? any lisp dialect?
<Fare> need it be online?
<Major> Fare, hehe, your rerences "sound" similar to clm in that clm explains who something is and it's functionality, but it doesn't explain relationship to the rest of the language
<hcf> ultima_: html tags?
<hcf> ultima_: rnt u going latex->html?
<Fare> ultima: what about starting with texinfo, and adding tags as needed?
<Major> Fare, well .. I had two lisp books at one time .. but aaa .. friend .. of mine sorta appears to have them now
<ultima_> Fare: i dont know texinfo
<Fare> Major: take em back :)
<ultima_> hcf: I was considering that
<Major> Fare, well .. he is a 4 hour drive from here
π Major/#Tunes ponders
<ultima_> Major what books
<Fare> Major: is he a one minute phone from where you are?
<Fare> Major: anyway, look around Review/Languages.html#LISP
<Fare> and also http://www.cs.rice.edu/CS/PLT/Teaching/material.shtml
<Major> ultima_, one was dark green .. thin .. paper back .. I wanna say it was Common Lisp .. but I dunno for certain on the title .. I have an odd memory when it comes to names .. the other was almost fleshen colored .. very thick
<ultima_> Major hehe
<Major> ultima_, I plan on getting replacements but today isnn't that day and I was kinda lookin for something to read and well .. our conversations on the subject sparked my mind again
11:00am
<Fare> one was heavy, with black symbols on white paper, and the other was not light, with white space carved out of a black substratum
<Major> Fare, huh?
π Fare/#Tunes whistles innocently
<Major> Fare, noo .. actually .. one of them was fairly light .. it wasn't thick at all
<Major> bah
<Major> the fleshen one wasn't fleshen .. it was god .. it was like an abstract art work on the cover
π Major/#Tunes grumbles that his memory works oddly like it does
<Fare> anyway, you can also order books online from amazon. The cll and cls FAQ describe the best books...
<ultima_> Major: I know that book
<ultima_> Major: It was almost like a landscape on the cover
<Major> also .. can someone explain to me why the only public domain implementation of Logo I have been able to find is UCB Logo .. which is Pascal based .. but MIT's AI lab shows Logo as beeing a spawned child of the Lisp family
<Major> ultima_, nod .. sorta fleshen in tone .. but with odd l;andscape .. thick one
<Fare> ain't there a C port of UCB logo?
<ultima_> Major: Soft cover
<Major> ultima_, nod
<Fare> anyway, there's a debian or redhat binary package for it that you can install!
<Major> ultima_, was the one of the two Cosmo appears to have gotten from me :(
<ultima_> Major: Thats a good book, old though
<Fare> btw, can anyone explain to me the whole arrow stuff? I admit I'm 2 months late in the discussion...
<Major> hmm
<Major> actually
<Major> Logo was scheme based?
π Major/#Tunes ponders .. 
<Fare> no, Logo appeared much before Scheme.
<Major> hmmm
<Fare> IIRC
<ultima_> Fare: DO you know texinfo?
<Fare> a bit
<Major> Logo was started by S. Papert?  MIT .. 
<Fare> Major: sure
<Major> according to the teachscheme page his first book on computers and children and such was 1981 .. 
<Major> and TeachScheme claims to be the highschool continutation of that project
<Major> in a sense
<Fare> well, actually, at about the same time in the late 70's
11:10am
<Fare> oh, maybe I'm wrong
π Major/#Tunes remembers Logo .. was the first language he really liked .. 4th - 5th grade
<ultima_> major hahahahahaha
<Fare> I never had a teacher in logo, and only did trivial things with the turtle
<Major> ultima_, well .. I had played with basic before and had done some stupid ness .. but never got used to it
<Fare> then I learnt applesoft basic
<Fare> then 6502 asm
<Fare> then BASICA and 8086 asm
<Fare> then HP RPL
<Fare> and Turbo Pascal
<Major> Fare, hehe .. my Logo teacher wasn't a teacher.. well .. not for Logo .. we had a logo machine that sat in the back of the room that you got to play with if you did your homework ahead of schedual and stuff
<Fare> then a tiny bit of FORTH, and C
<Major> Fare, the teacher was a Math teacher .. he had CP .. Mr. Smith I think was his name actually
<Fare> then C++ (old flavor)
<ultima_> Fare: Lamaer.
<Fare> then caml-light, then sh, then Scheme, then Perl
<Fare> ult: sure
<Fare> ult: that's how it happens when you're self-taught -- you take a lot of time before you find the right track
<ultima_> Fare; Turbo Pascal?!?!?!
<Major> hmm
<ultima_> Fare BTW do you ever do any coding?
<ultima_> or do you just write papers and stuff?
<Fare> ult: sometimes, I code
<Major> Turbo Pascal is sorta usefull for struct's and such .. but .. bleh
<ultima_> and when are you gonna translate those papers to english?
<Fare> Major: hey, I was 14, what do you expect?
<Major> Fare, hehe ..
<Major> havn't played with Forth myself
<Fare> (or maybe rather 15?)
<ultima_> Forth is queer
<Fare> I never played a lot with FORTH.
<Major> hmm .. what was I doing at 14 and 15
<Major> oh .. yah .. I had gotten into electronics
<ultima_> I'd rather just type in assembler than code in forth
<Fare> FORTH has a very powerful high-level language structure, on top of a bunch of very low-level primitives!
<Major> sorta bailed from coding for a while .. 14 up through 11th grade I buried my nose in electronics 
π Major/#Tunes shrugs
<Fare> ult: FORTH is mucho better than asm!
<Major> got bored with silicon I guess and jumped back into coding
<ultima_> Fare: What universe are you from?? =)
<Major> Fare, you know .. APL is very powerfull and all those fun buzzwords .. but I wont learn it
<ultima_> maj go learn ada
π Major/#Tunes shrugs.
<Major> bah
<Major> no thanx .. Kysh decided ada was the best language ever
<Major> I was gonna play with it .. but you know .. Kysh talking about how kewl ada was sorta scared the hell outa me
<Fare> FORTH is kind of a functional programming language grown upon a very low-level programming model, with lots of low-level legacy
<Major> Fare, sorta sounds like C
<Major> Fare, or taht description does
<Fare> APL is a functional programming language with a terse syntax
<Major> hehe
<ultima_> Fare: So is assembler =P
<_QZ> well those dinky celeron heatsink/fans dont work at 504mhz, unless u live outside in a cold cold area
<Major> APL  == "A Programming Language"
<Fare> Major: C is sure not a functional programming language!!!
<Major> Fare, define "functional"
<ultima_> Major: He means object oriented
<Major> ahh
π Major/#Tunes apologizes then
<ultima_> like C++
<ultima_> Fare is really into C++ and stuff
π Major/#Tunes though "functional" as in "it functions for me"
<Fare> Major: 
<Fare> <UL>
<Fare> <LI>For a great book, see the <A HREF="#SICP">SICP</A>
<Fare> <LI>Prof Leavens' page about
<Fare>    <A HREF="http://www.cs.iastate.edu/~leavens/teaching-prog-lang/home.html"
<Fare> </UL>
<Fare> <P>
<Fare> <LI><H4>Bibliographies</H4>
<Major> s/\(though\)/\1t/g
<Fare> <UL>
<Fare> <LI>Mark Leone's list of
<hcf> Fare: thx for phrasing stuff for abi's consumption, whether purposeful or not
<Fare> <LI>The <A HREF="http://hypatia.dcs.qmw.ac.uk/SEL-HPC/"
<Fare> <LI><A HREF="http://liinwww.ira.uka.de/bibliography/Contributing.html"
<Fare> <LI><A HREF="http://fermivista.math.jussieu.fr/"
<Fare> </UL>
<Fare> </UL>
<Fare> <P>
<Fare> <BR>
<Fare> <H3>Some "paradigms" of programming</H3>
<Fare> <UL>
<Fare> <LI><H4>Functional Programming</H4>
<ultima_> abi what is forth?
<abi> somebody said forth was queer
<ultima_> Yeah I thought so...
ωνω mode/#TUNES [+o Fare] by ChanServ
ωνω Fare was kicked off #tunes by Fare (flood!)
<Major> heheeh
ωνω Fare [rideaufr@esmeralda.enst.fr] has joined #Tunes
π ultima_/#TUnES chuckle
<ultima_> abi, karma for ultima
<Fare> abi: forth is at www.forth.org
<abi> ...but forth is queer...
<Major> hahaha
<hcf> Fare: use 'is also'
<Fare> abi: forget forth
<abi> Fare: I forgot forth
<Fare> abi: forth is at www.forth.org
<ultima_> abi, forth is also queer
<abi> okay, ultima_.
<Fare> abi: forth
<abi> it has been said that forth is at http://www.forth.org or queer
11:20am
<ultima_> abi, forget forth is also queer
<abi> ultima_, I didn't have anything matching forth is also queer
<ultima_> abi, forget forth is queer
<abi> ultima_, I didn't have anything matching forth is queer
<Fare> abi: ultima sucks
<abi> ultima_: my karma functions hav been disabled
<ultima_> abi who is ultima
<abi> i think ultima is god
<ultima_> Smart bot.
<Fare> abi: you suck!
<ultima_> abi who is fare
<abi> fare is probably utter <censored>
<abi> fare: u first
<ultima_> BAHAHAHAHAHAH
<ultima_> I'm sorry, I love that bot =)
<ultima_> anyway Fare
<ultima_> have you translated any of your papers to English?
ωνω Major [major@orion.jimco-fwt.com] has left #Tunes []
<abi> ultima_: i think ur pretty snazzy too
ωνω miguel [miguel@tirania.nuclecu.unam.mx] has joined #TUNES
<Fare> miguel?
<Fare> eres tu?
<Fare> ult: well, there's the one at ~fare/articles/ll99/
<Fare> ult: and the one in bespin:~fare/fare/lambdaND.tex
<Fare> miguel: eres un bot, o el real MdI ?
<Fare> oh, Idle 1 day 17 hours 35 minutes 58 seconds
<Fare> 
11:30am
<Fare> damn mouse
<ultima_> Fare heh wha do you guys think of clinton over here?
11:40am
<Fare> s/n//g
<ultima_> ?
<Fare> no, nothing!
<ultima_> heh
π ultima_/#TUnES heard frenchers were making fun on the US
<Fare> I tend to agree with what Salman Rushdie says of the affair
<ultima_> whats he say?
<Fare> instead of wasting billions on a stupid affair of pseudo-morality, the US should rather look at the morality of their own large-scale behavior towards, e.g. third-world country, or Iraq, or whatever.
<ultima_> heh
<ultima_> whats wrong with bombing Iraq
<Fare> He says that if the US senate wants to do a christian gesture for Y2K, they could, for instance, void the huge 3rd world debt!
<ultima_> heh
<ultima_> dumb christian ideas
<ultima_> y would ANYONE do a dumb thing like that
<ultima_> We need u third world countries owing us money
<ultima_> so that u stay 3rd world
<ultima_> duh
<Fare> it was just a suggestion about a real christian gesture, instead of hypocritical petal order.
<Fare> s/petal/petty moral/
<ultima_> dude
<ultima_> its not a moral thing
<ultima_> its not about him getting head in the oval office
<ultima_> its about him lieing under an oath
<ultima_> and obstructing justice
<ultima_> he attempted to stop due process
<ultima_> and *that* is why he's gettin his butt whipped
<Fare> it's not that we like clito particularly
<Fare> but the whole affair is a sign of petty hypocrisis on both sides.
<Fare> it's sad that the "most powerful country" in the world, the "cops of the world" behave so despicably
<ultima_> yeah
<ultima_> that is kinda sad
<ultima_> its a power struggle though
<Fare> a *hypocritical* power struggle
11:50am
<ultima_> Is there any other?
<Fare> I hope there is. I'm persuaded democracy could be mucho better.
<ultima_> heh
<ultima_> So what if the IQ of the average US citizen is below their shoe size...
<ultima_> So what if we are heading toward a censored society where people are protected from knowledge by stricter and stricter laws
<ultima_> So what if eventually we will be slugs 
<Fare> that's the anti-thesis of (classical) liberal democracy
<ultima_> Fare heh
<ultima_> Well at least our food is good =)
<Fare> like, MacDonald's?
<Fare> ;->
<ultima_> Yep
<ultima_> Not to mention the wonderful chitterlings
<ultima_> and moms meatloaf
<ultima_> and oh yeah stew
<Fare> where's the barf bag?
<ultima_> We left em in france, they would need em more =)
<ultima_> What kind of stuff do u guys eat?
<Fare> I think I'll stick with french cuisine
<ultima_> what kind of stuff do you eat?
<Fare> lots of wildly different things. From oeuf-sur-le-plat to boeuf bourguignon.
<ultima_> i have no idea what any of those are
<Fare> ultima: there are lots of cuisine recipee pages on the 'net, if you wanna try!
<Fare> what I prefer of course, is french pastry
<ultima_> ah, what kind of pastries?
<ultima_> I love italian deserts
12:00pm
<Fare> lots of kinds of soft cakes
<Fare> or just brioche, croissant and derivative pastry.
<ultima_> pets de soeurs
<ultima_> oo that looks kinda good
<Fare> nope, pets de nonnes
<ultima_> Whats that?
<Fare> there's a XL version of these -- the religieuse.
<Fare> s/soeur/nonne/
<Fare> a "nonne" is a *religious* sister
<Fare> "soeur" is just sister
<ultima_> oh
<ultima_> i see
<Fare> "religieuse" is another name for a nun
<ultima_> whats that have anything to do with it?
<ultima_> Baeckeofe, heh, frenchers eat stew too
<ultima_> French people eat horsemeat??
<Fare> a "religieuse" is a big ball of paste stuffed with cream, with an icing on top, and another smaller similar ball on top
<ultima_> oooh
<Fare> we may occasionally eat horsemeat, but it's rather more expensive and not as tender as beef.
<Fare> when you keep only the smaller ball, it gives a "pet de nonne"
<ultima_> hrm, so pet de nonne is same thing only a bit smaller?
<Fare> I can assure you that cooking this kind of pastry is not for newbies.
<Fare> the "religieuse" is both balls
<Fare> the "pet de nonne" is lots of smaller balls; usually served at buffets.
<Fare> whereas the "religieuse" would be served at a 4'oclock tea or such, between friends.
<ultima_> oooh
<ultima_> ok
<Fare> (friends, well, let's say, a smaller assembly)
<Fare> since we're not british, the tea might be replaced with coffee, or some soft drink
<ultima_> sounds nice
<ultima_> I like the way french people work
<ultima_> its not like here
<ultima_> here u are a slave
<ultima_> i mean, it explains our high productivity, but people dont go out and have a nice lunch or antyihg
12:10pm
<ultima_> you go to McDonalds and get a bigmac for lunch
<Fare> uh, you don't get pastry everyday, of course. And don't expect *good* pastry at work lunch
<ultima_> Of course;)
<ultima_> But I assume at least you eat something edible
<Fare> or even at restaurants, least they are very expensive.
<hcf> anyone know what happened to BatOS?
<Fare> but you can have pastry of whatever quality you're ready to pay for at shops down the street.
<ultima_> mmhmm
<Fare> hcf: no idea
<Fare> Novi Marek left university.
<Fare> Maybe he was hired by the BETA people?
<hcf> most current email addr i could find, novy@kiv.zcu.cz
<Fare> hcf: does mail reach him?
<hcf> `vrfy` has the addr is fine
<hcf> s/has/says/
<Fare> what's vrfy?
<abi> vrfy is ftp://ftp.nikhef.nl/pub/network/vrfy.tar.Z
<Fare> abi: what is it for?
<abi> fare: no idea
<hcf> verifying email addrs
<Fare> how does it work?
<Fare> is there something in the [E]SMTP protocol that allows for verification of validity of an address?
<hcf> something like that
<Fare> 137!
<Fare> abi: I love you ... NOT!
<hcf> if links r dead, is someone gonna follow possible leads (email addrs etc) to see what happened or just forget them?
<Fare> hcf: as you like!
<hcf> i dont wanna do it
<Fare> hcf: I enjoin you to send mail to novy marek.
ωνω SignOff ultima_: #TUNES (Ping timeout for ultima_[user-37kbb1u.dialup.mindspring.com])
12:20pm
<Fare> 130!
π Fare/#Tunes is away dinner
ωνω SignOff Fare: #TUNES (Connection reset by pear)
12:30pm
ωνω NetSplit: norton.openprojects.net split from carter.openprojects.net [12:57pm]
ωνω BitchX: Press Ctrl-F to see who left Ctrl-E to change to [norton.openprojects.net]
ωνω Netjoined: norton.openprojects.net carter.openprojects.net
ωνω _QZ [brand@p0wer.qzx.com] has joined #TUNES
<hcf> _QZ: http://www.sci.tamucc.edu/~pmichaud/toast/
<_QZ> i have seen that before
<hcf> i thot so
<hcf> found the link on danh(dead qnx guy)'s homepage
<_QZ> uhh
01:20pm
<_QZ> hey do u have any docs that list all family/model/steepings for all cpus?
<hcf> thers the pc chip list
<hcf> which u probly know of
<_QZ> no i dont
<hcf> http://infopad.eecs.berkeley.edu/CIC/
<hcf> thats not the pc chip list
<hcf> ftp://rtfm.mit.edu/pub/usenet-by-hierarchy/comp/sys/ibm/pc/hardware/chips/*
<hcf> IBM-PC Hardware and Software hacks @ http://people.ne.mediaone.net/fstewart/ibm-pc.html
<hcf> </end>
<_QZ> hehe
01:30pm
<_QZ> thanks
<hcf> need more?
ωνω BlueEye [root@bjas1-p30.telepac.pt] has joined #tunes
ωνω SignOff BlueEye: #TUNES (Leaving)
<_QZ> nope
<_QZ> not yet atleast
01:40pm
ωνω GMOL [gmol@h24-64-168-191.mt.wave.shaw.ca] has joined #tunes
<GMOL> ANyone here?
<hcf> semi
<GMOL> :)  What did you think of the What I Think Tunes is thing....would you say your ideas agree/disagree?
>>> GMOL [gmol@h24-64-168-191.mt.wave.shaw.ca] requested USERINFO  from TUNES
01:50pm
ωνω SignOff GMOL: #TUNES (Read error to GMOL[h24-64-168-191.mt.wave.shaw.ca]: Connection reset by peer)
ωνω Tril [dem@bespin.ml.org] has joined #TUNES
<Tril> Word's grammar correction cracks me up.  I typed this sentence: "He dies as a man, with honor and respect, stronger than any other man in the execution room."  Word asks me to add the word "does" to the end of the sentence.
<hcf> hoy Tril
<Tril> ehh
<hcf> how about using #tunein for irc meetings?
<Tril> have we had another meeting?
<hcf> no
<Tril> what would be the reason for that
<hcf> to help handle off topic chatter
<Tril> if you want to, sure
03:30pm
<hcf> whats happening w/ the site, anything?
<Tril> I think I'm going to do it, since nobody else seems to be
<hcf> whers beholder lately?
<Tril> first I'll finish this projects thing. If you want to test it out, visit the url in the topic. You can create an account and edit info, login and logout, but that's it so far.
<Tril> dunno
<hcf> these account differ from bespin?
<hcf> s/nt/nts/
<hcf> ccess Denied. Either your password is incorrect or you have logged out
<hcf> already 
<Tril> yeah, these accounts are only for the subprojects database. You need a user id in the projects program to be involved in any of the projects in it.
<hcf> s/cce/Acce/
<TUNES> do you have cookies turned off
<Tril> argh
<hcf> yep :P
03:40pm
<Tril> "login" and "logout" are equal to "set cookie" and "unset cookie" :)
<Tril> anyone should be able to participate in a project, which means I plan on a lot more people signing up on that web page than I am going to give shell accounts to.
<hcf> what about ppl who dont do cookies?
<Tril> not allowed. The only way to do that is to have passwords visible in the html source, which is a security risk (others could go to your cache and get access to your member id)
π _QZ/#tunes is GONE ( leaving )
<Tril> if necessary I can set up a "lynx" telnet account so people can use the system
03:50pm
<Tril> cookie paranoia is dumb - even if you turn off cookies there is plenty of information admins can get from you when you visit a site.
<Tril> you should instead be paranoid to be online at all.
<Tril> there are some weird people here.
<Tril> :)
<hcf> ok, what info did u get when i hit ur projects page?
<Tril> just the ip and the path you requested, although if I increased log detail I could get stuff like browser,referer,email
<hcf> i use a proxy
<hcf> cant get my browser
<hcf> referer is turned off
<Tril> that's odd, when i use my isp's proxy, my browser type goes through just fine
<hcf> maybe it does
04:00pm
<hcf> but so what
ωνω Myrmidon [elysium@dm5-39.slc.aros.net] has joined #tunes
π hcf/#tunes is away 17/01/99 19:08:33 EDT : ...
ωνω Tril has changed the topic on channel #TUNES to: http://tunes.org/ - This is the channel of the TUNES reflective computing model project
ωνω Beholder [beholder@ppp-073.m2-2.sub.ican.net] has joined #tunes
04:10pm
<Tril> aww B hcf was just wondering where you've been..
<Beholder> My ISP canceled my account, and decided not to tell me :( !
<Tril> hmm, were you violating the terms of service?
<Tril> looks like you have it back
<Beholder> Yes, I was.  I changed my adress.  Still would have been nice if they would have said something first :)
ωνω ultima [ultima@user-38lc7ep.dialup.mindspring.com] has joined #TuNeS
π ultima/#TuNeS whistles some tunes
<Tril> your snail mail address? so you couldn't be billed?
<ultima> Tril
<ultima> sup
<Tril> Toons?
<Beholder> I get billed on a credit card, so I really don't know why they need to know where I live.
<ultima> Beholder: Why, so they can verify that you are who you say you are
<ultima> Beh, who wants your addy?
<Beholder> Ult: That is probably the best guess.  They had good cust support people though, and made it out to be a huge mistake on their part :)
<Beholder> ult:  My ISP
<ultima> Beholder; Had a feeling, what was the mistake?
<Tril> <Beholder> My ISP canceled my account, and decided not to tell me :( ! 
<ultima> Oh. Suck.
<Beholder> ult:  They cut me off without contacting me by phone or e-mail
<ultima> Beholder: Their reason
<ultima> ?
<Beholder> ult: Change of address
<Beholder> I guess they really need to know where I live :)
<Beholder> But this stuff is not important at the moment, I did not get any mail from Tunes or UniOS in the last day or two, I'm going to go check the archives, but was wondering if there was anything important to look for?
<Tril> did you get the one where Maneesh said if there is no progress by July we should shut down the project?
<ultima> Tril well what is getting done?
<ultima> We need some code
<Beholder> Tril: Nope, sounds important though.  I'll take a look at it.
04:20pm
<Tril> ultima: I wrote some code, see http://projects.tunes.org/.  OK, it's perl code and it's not exactly part of TUNES, but it's code! :)
<ultima> Tril we are not doing anything!
<ultima> Write some REAL code
<ultima> Write a kernel that can boot
<Tril> that's too easy
<Tril> it would't be useful. there already are plenty of kerensl that can boot
<ultima> sure it would
<ultima> it would give u a development platform
<ultima> Tril I'm gonna hurt you
<ultima> your lame login prog uses cookies!
<Beholder> Tril:  Could you have a basic kernel that is expandable enought to incorporate Tunes ideas?
<Tril> B: that is kind of what core was trying to make with Clementine.
<Beholder> Tril:  Did it work?
<Tril> Do you know anything about Clementine?
<ultima> Tril what is tunes
<Tril> have you read the "No Kernel" document by Emmanuel Marty?
<Tril> ultima: a frigging abstract model of ideas(information)
<ultima> uh huh
<ultima> That sounds simple enough
<Tril> Good!
<ultima> Then, why is it vapourware?
ωνω ultima is now known as ultima_
<Beholder> Tril:  Sorry, I have heard very little about Clementine
<Tril> ultima_: Because it's not just any model of information, it is THE model.
<Tril> or, more accurately, all models.
<ultima_> Tril I doubt that
<ultima_> How is it a/the/all models of information?
<ultima_> Tril9~
<ultima_> ?
04:30pm
<Beholder> Actually, couldn't the Tunes model be developed on any platform, and not necessarily need a new OS?
ωνω SignOff ultima_: #TUNES (Ping timeout for ultima_[user-38lc7ep.dialup.mindspring.com])
<Tril> the No Kernel is an operating system design based on everything being modular binary components.  It boots up and loads components, and that's about it so far. It will be released to the public soon.
<Tril> Clementine is the os being designed with the no kernel model.
<Beholder> Interesting
<Tril> Emmanuel Marty thought of the no kernel model and programmed most of clementine. He wants TUNES to do whatever it takes to modify his system to work for tunes!
<Tril> I don't know how we're going to do that, yet...
<Beholder> Tunes, like UniOS, is a high level definition.  You can make it run on anything if you define it properly I guess
<Tril> not yet it isn't!!!!!! :)
04:40pm
<Tril> it's supposed to be
<Beholder> So once Tunes is defined, it should be able to squeeze into any given framework.  Am I right?
<Tril> other way around, I hope, instead.
<Tril> without squeezing necessary, either.
<Tril> any given framework should be able to fit into tunes. that's a main goal.
<Tril> i'm not sure exactly what you mean  tunes fit into any given framework, though.
<Beholder> I mean you can have the tunes final "applications" and enviroment being able to run under any given OS.  This is assuming that Tunes is a high level definition.  Not that you would want to run it under anything but a new OS designed for it, but the possiblity of all High level framworks is that they can be ported.
<Tril> yes, that is a theory, which remains to be seen :)
<Tril> but what does "high level" mean?
<Tril> to you
<Beholder> High Level to me, means that the specification is generic enough to be coded in any given language, on any given platform.  Psedo code I guess.   The oppitsite of saying "We have to make Tunes run in Rung Level 0, in this protected mode, and to speed things up we utilized these certian registers".  High level also means non platform confoming
04:50pm
<Tril> let's criticize together this page: http://tunes.org/papers/Glossary/#abstraction level
<Tril> well, I think it is pretty good. If you want to add to it please do!
<Tril> (no, I didn't write any glossary entries yet, Fare did them all)
<Beholder> Hmm... didn't really understand it. 
<Tril> a program is low-level if it introduces NOISE in the solution to the problem.
<Beholder> That makes sense.  This is the metaprogramming thing right?
<Tril> what do you mean
<Beholder> You make simple programs that "describe" the final effect, and the compiler makes the real source code for you?
<Tril> don't all programming languages do that?
<Tril> compiling IS metaprogramming.
<Tril> metaprogramming is any program which inputs or outputs another program.
<Beholder> Ahh, understood now :)
<Beholder> What do you define as NOISE?
<Tril> noise is extra information in the solution (which is a program) to the problem
<Tril> "...program is...low-level...if it handles not the objects that appear in the problem, but representations of them in some space external to the problem..."
05:00pm
<Tril> In other words, solving problems in their own problem space = high level = noise-free
<Tril> solving problems in the wrong problem space = low-level = noisy = not useful, but expedient.
<Beholder> What part of a typical C++ program code would you consider noise?
<Tril> the syntax of C++, any instructions which could be executed in parallel (but can't because the language only lets you build programs out of sequences), any "hacks" or "ad-hoc" code added to work around problems in underlying libraries or language limitations,
<Tril> any manual optimizations done such as creating data structures in memory in the code just for performance, manual persistence (saving/loading information)
<Tril> inputting information from the user,
<Beholder> So what you wish to do is abstract the solution so it's more human level?
<Tril> accessing hardware directly
<Tril> yes, well, at the level of the problem, instead of at the level of how the problem is best represented on the computer
<Tril> i mean on a case by case basis, each problem. I dont know what human level is exactly, it's hard to define.
<Beholder> I agree with your goals in totallity, however were you aware this was possible to develop without an OS?  (as far as I know)
<Tril> tunes is not an OS, I know that.  Do you mean something in particular?
<Tril> ie..  "are you aware this is possible by doing..."
05:10pm
<Tril> i agree with you tunes does not need an OS, since it is just a philosophy, and any environment could adhere to it.  But the entire point is
<Beholder> Umm, I think it would be possible to make the Tunes system similar to the Java environment.  Total control over it's own resources, and leaves the OS alone.  (just as a thought)  That way you could have all the development power of the Tunes sytem, and be able to port it to any OS that can handle it the most "expediently"
<Tril> yes there are domain specific abstractions, for each problem.. that's what we have now. But they're all in different languages.  We need all domainspecific abstractions to be supported by one general domain abstraction system, so the specific abstractions can be defined in terms of how they relate to all the others (so shared concepts between domains can be reused!!)
<Tril> no, your idea is fine, 
<Tril> tunes doesnt need to be the entire system, in order to work, or to interoperate with other tunes', or whatever you are thinking
<Beholder> However, I think that an OS designed from the ground up, to support a specific type of environment, would logically be more "expident"
<Beholder> I'm not trying to state this information to be argumentive, or repetitive, I'm just trying to further my own understanding of the proposed system.  The more I hear, the better it gets :)
<Tril> OSes designed from the ground up to support specific environments is ALL WE HAVE.
<Beholder> Now, I think I actually understand where the project is coming from, and possibly where it is going :)
<Tril> that's why we want to do something different.
<Tril> Well, that's good. :)
05:20pm
<Beholder> I can see now, where the UniOS system (in my mind) differs from Tunes slightly.  The focus of UniOS is currently on developing the specification for a Universal portable object format, that can be used in any enviroment, and any language.  Unless I am wrong, and this is one of the domain you speak about?
<Tril> Not just OSes, but llanguages, and programs, and libraries, too.
<Tril> well, i believe it is impossible for one static format to be universal.
<Tril> requirements are always changing, so eventually any format will be obsolete.
<Tril> Actually, Emmanuel Marty's XCOM is a portable binary object format, that can be used in any environment and any language.
<Tril> tunes is more an expressive human-readable object construction system
<Beholder> There are a finite number of things that can be done in a program.  Simplified they are storage, loops, decisions, threads,  simple math, and object interaction (as far as I see)
<Tril> wayyyy too many things, simplify it a little more!
<Beholder> How would you simplify it?
<Tril> a "pure functional" approach does not require storage.  Loops can be encoded as recursion, like in FP.  what is a thread?  simple math, well isnt that just a particular program you can write, not a language feature..?   object interaction= all of the above could be called object interaction, you hve to be more specific.
<Beholder> Storage means the ability to hold numbers in memory or register... not disk based :)  You need numbers.
<Beholder> Thread is the "new" programming concept to allow parallel processing within a program
<Tril> no, i mean a pure function doesn't have variables.
05:30pm
ωνω SignOff miguel: #TUNES (Leaving)
<Tril> no assignment statements.
<Beholder> How do you code functions without any numbers?  I'd imagine it would be hard to make decisions.
<Tril> well dont worry about the numbers thing. functional languages all have support for numbers in them.
<Tril> (for efficiency)
<Tril> but if you want to be pure, then numbers are just constant functions which return themselves.
<Beholder> But on some level, you would eventually have to deal with the numbers.  And the object would have to store them.
<Tril> i dont want to argue about functional languages. you should go learn one if you are interested, and it's not directly on the topic we are discussing.
<Tril> so what are you trying to do with this list of things a program can do?
<Beholder> This list of functional things would be the basic generic object format.  Like a Java class file.  Architecture neutral application distribution. 
<Tril> so why not just use java
<Tril> I sort of had what you say before I joined the tunes project
<Beholder> That was another possiblity.  But the language is not free.  And Sun makes damn sure if you make anything using the Java lang, that it follows the Java specs... don't want that
<Tril> a couple basic operations that were available to write methods in.
05:40pm
<Beholder> My first idea was a freeBSD java OS, but Java as an environment is very limiting
ωνω SignOff hcf: #TUNES (Leaving)
<Beholder> Actually, I have done little research into alternatives, do you know anything that would fit the description, but it free?
<Tril> java is just another language. it isn't any more like what we want for tunes, than any other language. java isn't even very special as a language, anything done in java now could be done in lisp, which is 30 years old
<Tril> I think you should look at XCOM
<Beholder> I'm not looking for a language, just a distributable object format that encompases basic program functionality, and allows for threads
<Beholder> By object format, I mean something that is platform neutral
<Tril> I think clementine requires the compiler to implement security, though.
<Tril> heard of ANDF?
<Beholder> Nope
<Tril> abi, andf?
<abi> andf is http://slashdot.org/features/98/10/20/116240_F.shtml
<Tril> there's lots of good links on there..
<Beholder> Actually one of the UniOS members said DCOM was what we were looking for... uggghh.... 
05:50pm
<Tril> XCOM is similar to the idea of dcom, but with big improvements (I can't name them, I don't remember the limitations of DCOM)
<Beholder> Do you have a URL to XCOM?
<Tril> it was on the dolphin site, but that's down.. it will be on the nokernel.org site, but that's not yet up.. I'll send it to you. this is the xcom specification. is that ok?
<Beholder> Yep
<Beholder> How would a Tunes application be distributed?
<Tril> tunes doesnt have applications
<Tril> it has modules, which are the unit of distribution
<Tril> (distribution, as in a package, not distribution, as in distributed networks)
<Beholder> Ok, how does it distribute
<Beholder> modules
<Beholder> Yes getting code from programmer to user :)
<Tril> hmm
<Tril> there can be data formats
<Tril> just they will be easy to make , easy to generate automatically programs to use them, and to generate converters between versions automatically.
<Beholder> So what format are they distributed in?  Pure Tunes Code (text), abstracted object (like UniOS), or machine compiled?
<Tril> any format can be used to describe data formats, such as XML, or jiml's prism system
06:00pm
<Tril> i dont know that tunes uses text.. what is an abstracted object in this case
<Tril> probablyt not machine compiled, at least not without source included also.
<Beholder> Some variant of the XCOM model.  Unless I'm missing something.  I assume code is still written in plain text, compiled in some way, and stored for usage on the target system, to be brought up at the appropriate time?  (traditional computing model?)
ωνω jdl [jiml@ultra1.inconnect.com] has joined #TUNES
<jdl> Hi, folks
<Beholder> Hello :)
<jdl> I was just reading the log and I noticed you were discussing domain abstraction :)
<jdl> Have any of you had a chance to look at Prism yet (Tril?)
<Tril> i guess you are missing something.. code is not written in plain text (hi, jiml!) ..it's written by syntax-directed editors for the particular domain.  Compiling is well, if you read my mail I said compiling and interpreting both should be combined into one, "evaluator" which can do either, and not just flip-flop choice but a sliding scale, partially compiled/partially interpreted, or entirely one or the other. 
<Beholder> Yep, a little while back
<Tril> jdl: yeah we're just figuring out that the main principle of tunes is to solve problems in their own problem space, a lot like what you are doing. :)
<jdl> Prism does a lot of the things you were just talking about.
<jdl> Tril: Great!
<jdl> :)
<Tril> how easy is it to use prism and to see it doing these things?
<jdl> Hmm.... well, I think pretty easy, but can you be more specific?
<Tril> what is the interface like? text? graphics?  interactive or command-line?
<jdl> The main idea of Prism is that you capture domain abstractions as "metamodels" and then write small programs which translate models from one metamodel to another.
<jdl> Tril:  working...
<abi> i heard working was always a plus :)
<jdl> Tril: Prism doesn't specify a compiler interface.  Metamodels are syntax-independent and can use any language, textual, visual, or other.
<Tril> what does it run on and how hard to install.  But my main question is, if I download it, will I be able to SEE anything happen and understand what is going on
<jdl> The current Prism compiler uses a graphical interface.
<jdl> Tril: Java based, no automated install.
<Tril> abi, you are cool
<abi> ...but abi is the little girl borg with a big brain or #tunes' personal little whore or an infoslut...
<jdl> You'll need the JDK to install the current version of the Prism compiler.
<Tril> the entire jdk or just the libraries?
<jdl> Just the 'java' executable.  But...
<Tril> I have the JDK Runtime Environment, not the development tools
<jdl> Although Prism is fully specified, the compiler is not done.  <sheepish look>
06:10pm
<jdl> If anybody wants to help me finish it, I'm willing to let them.  :)
<Beholder> jdl: What needs to be done? 
<jdl> The architecture is mostly complete, the visual interface is totally done, but needs to be updated for a change to the Prism spec, file IO and a basic generic language parser are totally incomplete.
<jdl> More...
<jdl> The Prism meta-metamodel is completely architected and done but needs to be updated for the Prism change, and the Reflection metamodel interpreter is not yet started.  The interpreter is the next big step.
<Beholder> Wow, sounds like you've been doing a lot of work on it :)
<jdl> See http://www.teleport.com/~sphere/0009/1/index.html
<Beholder> I got a 404 on that one
<jdl> Boholder: Yep, a TON.  I've been working on this basically nonstop for about nine months, the compiler implementation about two months.
<jdl> Beholder: Hold on, chacking URL
<jdl> Whoops!  sphere/documents/0009...
<jdl> You can also download the current version from that URL.  It basically demonstrates the visual interface and meta-metamodel support.
<jdl> It doesn't do anything significant, though, because the interpreter's not in yet... :)
<Tril> jdl: have you read any of my conversation with maneesh about my abstract model?
<Beholder> Hmmm... programmed a RAD language to develop something easier then RAD :)
<jdl> Tril: Yes, that's why I popped on.  VERY similar to my Prism/PISE thoughts.
<jdl> Beholder: ?
<Beholder> Java is conciderd a Rapid Application Development language, whos purpose is to make programming easier and more logical.  The Tunes and Prisim system sound better :)
<jdl> Beholder: Hopefully!  :)  The idea behind Prism (and Tunes, I think) is to give every problem domain a specific RAD language.  Prism even allows you to mix languages. :)  (Well, it allows you to do so if you figure out how... still very...
06:20pm
<jdl> experimental stuff, but the tools are there.  I call it a "language laboratory.")
<Beholder> So what is the major benefit of these types of systems... actully better question:  Who would want it the most?
<jdl> Only in Prism, "languages" are syntax-independent, so I call them "metamodels."
<jdl> Beholder: Major benefit, as far as I'm concerned: more direct association to problem domain, therefore easier, quicker development and less errors.
<Tril> I am skeptical that you can have a completely generic model system while still based on limits of 32-bit numbers
<jdl> Tril: Yeah, Billy Tanksley pointed that out too.  The limitation has been removed. :)
<Tril> I didn't see whether you replied about it...
<jdl> Tril: I replaced the _pBit32 datatype with a _pStream datatype... See my first response on the mailing list, I'll get a URL...
<jdl> But it's basically an ordered set of _pBits.  I've already defined a metamodel which allows you to store infinite integers, given infinite capacity :)
<Tril> What is the status of info on the web page vs. info on the tunes mlist? is everything you posted on your web page?
<jdl> Tril: The Prism spec is out of date.  I'm updating it but haven't posted it yet.  Everything else is current.  The error in the Prism spec is that _pBits32 has been replaced with _pStream.
<jdl> Okay, info on _pStream is at http://www2.tunes.org/list/tunes/9901/msg00101.html, about 1/3 down.  Search for "Next paragraph..." and read my next block of text after the quoted text.
<Beholder> jdl:  I'm still a little clueless on some of the details of these types of systems.  What advantage does it offer to somone like a high performance game programmer? 
<jdl> Beholder: I believe that the disadvantage of domain abstraction is that you lose control over the details.  Thus, C++ isn't as efficient as assembly, but it's easier to program in.  (more)
<jdl> The advantage of Prism is that you could create a domain abstraction which gave you exactly the level of control you needed.  So you could abstract out the simple, repetitive stuff while still allowing you to zero in on important details.  (more)
<jdl> I can't give you an exact example, because I'm not a games programmer. :)  (more)
<jdl> In the short run, it's probably less efficient to create a new domain abstraction for a single program.  But I see them being sold, like class libraries.  (end)
<Beholder> jdl:  I see the advantage of the system in research fields and corporations
<Beholder> jdl: They need something like that :)  
<jdl> Beholder: There's more advantage for those who don't need tight control over the system, like games programmers do.
06:30pm
<jdl> The tradeoff is quicker development speed for less control, which often means slower, less efficient computer speed.
<jdl> But for a lot of folks, people time is more valuable than computer time.  :)
<Beholder> jdl:  Understood :)  It's a kick ass idea, but it could not serve as the basis of an OS that would be implemented in a home or industrial enviroment.  Typically both fields tend to need a very high degree of control over the target system.
<jdl> Beholder: Prism is not an OS.  :)  It's (at the moment) the realization of a development philosophy, and (in the long run, I hope) a development platform.
<jdl> That's the main difference between Tunes and Prism.  Prism is a little less ambitious, I think.  :)
<jdl> Of course, I could be totally wrong about Tunes' goals. :b
<Tril> no, that sounds about the same thing as tunes,, actually.
<Beholder> jdl:  Hey, if it can run on top of UniOS, then I'm happy :)  
<Tril> I would like to find out if prism can handle my type system ideas..
<jdl> Tril, you mean N-space?
<Tril> it would also be cool if it can run on clementine.
π Beholder/#tunes is AFK
<jdl> Prism is a compiler SPECIFICATION, so if you can right a compiler for it, you can run it wherever you want.  ;)
<jdl> s/right/write  :b
<Tril> Well, the entire intentional system, where things are built from the top down using constraints, instead of from the bottom up using primitive variable types in structures.
<jdl> Tril: hmm... okay, I don't know much about your system.
<jdl> But...
<jdl> I would like to find out if it could be modeled in Prism as well.
<Tril> since you have a spec, it's easier for me to read yours and compare them myself..
<Tril> i'll do that soon
<jdl> My goal is for Prism to be able to model any computing system, i.e., have no built-in limitations.
<Tril> i need to go ...last bus.. I'll log back on in about 45 mins.  do you want to meet?
<jdl> One question: Is your intentional system a metamodel, a language?  Do you create programs in it?  If not, it's probably not appropriate for Prism.
<jdl> Sure.  Maybe we can work out how to do your system in Prism together -- you provide info on yours, I'll provide on mine.  :)
<Tril> I don't know. I will think about that.
π Tril/#TUNES is away: (AFK) [BX-MsgLog Off]
π Beholder/#tunes has returned
<jdl> welcome back... Tril had to go.
<Beholder> jdl: I'm not too sure if we've met yet, but I'm the project coordinator for Tunes and UniOS (not Tunes, but similar).  I've read some of your posts to the list on the prisim system, but until now did not understand much of it :) heheh
06:40pm
<jdl> That's why I'm here.  :)  Who are you on the list?
<Beholder> I'm beholder :)  But I have not posted much to the list (actually only 1 message :)  My job tends to keep me to doing the background work
<jdl> That's okay. :)  Do you have any other Q's about Prism?
<Beholder> Yep, who are you developing it for, and why?
<jdl> Good question.  Hmm...
<jdl> I'd have to say I'm developing it for myself, because I hate computers.  :/
<Beholder> I don't mean to pry, I'm just a curios person :)
<jdl> Seriously, I find computers incredibly obnoxious to use.  I want them to be less, um, crappy.  ;)
<Beholder> jdl: LOL :)  Thats the best answer I could imagine :)
<jdl> That's why I created the Sphere project.  See http://www.teleport.com/~sphere/documents/0000/4/index.html  (don't worry, it's VERY short.)
<Beholder> My resoning was:  You can't make computers less crappy until they have a better OS :)
<jdl> Sphere started out as an OS project.  :)  I got sidetracked, though, and I like my current approach better now (of course, otherwise I wouldn't be doing it this way ;)
<Beholder> I agree, you found something that was easily attainable (relatively :) , that could further your project goal.  Not only that, you are almost done! :)
<jdl> The end of the beginning, at least.
<Beholder> Yes, but the first step is always the hardest
<jdl> Absolutely.  Prism has come a LONG LONG way.  I was pretty excited when I finally put up the web page.
<jdl> Of course, hardly anybody responded.  ;p
<Beholder> And you did it all in Java :)  Damnit, us OS projects have to worry about developing hardware drivers, kernels, IPC code, GUI's... it's quite a huge step :)
ωνω hcf [nef@escher.sdi.agate.net] has joined #tunes
<Beholder> Takes forever to make an OS... 
<jdl> Exactly.  All those things aren't really revolutionary, but you have to do it in order to make the interesting parts work.
06:50pm
<jdl> That's what happened nine months ago when I started Prism, I started questioning my goals very critically -- why an OS, why an OO OS, etc.
<jdl> Prism and the current goals of the Sphere project were the result.
<Beholder> Yes, they are just well researched, tested, concepts, put together in a (more) logical way than what we currently have :)
<Beholder> The OS thing of course,  not the "super-langs" :)
<jdl> :)
<jdl> So, you know anybody who would like to help finish coding the Prism compiler?
<jdl> (hcf, same to you..?)
<Beholder> Not personally, but I have a few project members who might be interested
<hcf> not i
<Beholder> I think it's interesting but don't have time to code anything at the moment :(  (school, work, OS Projects)
<jdl> If you would drop them a line, that would be great.  Time is always a problem  :b
<jdl> Beholder: exactly.  :)
<Beholder> I will post a message in the UniOS only mailing list, they might be interested.
<jdl> Please do.  It would help a lot for someone not related to Prism was convinced that it was possible.  :)
<jdl> Hmm.m. that was a lousy sentence. :)
ωνω Myrmidon [elysium@dm5-39.slc.aros.net] has left #tunes []
<Beholder> jdl:  Can't even make proper english and you want to create a language?  heheheheeh :)  (note: I know my grammer is bad, but I'm only making an OS :) hehehe
<jdl> LOL   I'm in deep trouble, aren't I!  :)
<jdl> Fortunately, I'm just saying how people can move their lips -- they get to create the language.  :)
<jdl> language(s)
<Beholder> So do the sematics of the language map to prisim functions?
<jdl> The analogy about lips is accurate... I'll explain:
<jdl> Prism defines certain data types: _pBit (bit), _pStream (ordered collection), _pMap (indexed collection)  (more)
07:00pm
<jdl> Language creators define groupings of these data types and assign semantics to those groupings.  (more)
<jdl> (I wish I could give you an example, but IRC isn't good for ASCII art.)  (more)
<Beholder> I think I have a general idea, but doesn't creating all the inital sematics end up being time consuming?
<jdl> So Prism defines no functions, just basic data types for languages.  Those languages are used to create programs -- actally, MODELS of programs -- and those models are translated into machine code.
<jdl> (end)
<jdl> To answer your question...
<jdl> Yes, if you only do it for one program, it would be a waste of time.
<jdl> But a single metamodel could be used over and over for programs in a single domain.
<jdl> For example...
<jdl> Making Java was definitely time consuming.  But the initial investment has paid off for Sun and everybody who uses it (hopefully! :) )
<jdl> Prism also has features which make it easier to define a metamodel than a traditional language.
<jdl> (end)
<Beholder> So it's a language, to create languages, that can be used to solve specific problems.  (my own laymens definition)
<Beholder> sorry, maybe specific types of problems would be more accurate?
<jdl> Kind of.  Closest example:  It's a language factory and compiler-compiler, to create languages and compilers, that can be used to solve specific types of problems (as you said).
<jdl> The Prism meta-metamodel (_pBit, _pStream, _pMap) is the language "factory" -- the meta-metamodel.  Languages are metamodels.
<Beholder> So you create languages and compilers, which are used to code specific problems, which are ALL done in a unified format, and eventually run like any other progrm
<jdl> If I understand you, yes.
<Beholder> There is a meta-compiler, and a real compiler correct?
07:10pm
<jdl> Ummm... yes, I think: There is a "Reflection" metamodel/language which is used to code compilers.  The meta-compiler is the Prism compiler -- it interprets Reflection models.
ωνω ultima_ [ultima@user-38lc72m.dialup.mindspring.com] has joined #Tunes
<jdl> (BTW, if you can come up with a better name than "Reflection" let me have it! :) )
<Beholder> The only other thing I can think that makes more sense is "self-modification" but it's longer to write
<jdl> So, does all that make sense?  Or am I tilting at windmills?  :)
ωνω SignOff ultima_: #TUNES (Leaving)
π Beholder/#tunes is AFK (be back in a sec)
07:20pm
ωνω ultima_ [ultima@user-37kb9uu.dialup.mindspring.com] has joined #tunes
π Beholder/#tunes is back
<Beholder> jdl:  I think you have a good concept, if I uderstand it correctly :)
<jdl> Thanks  :)
<jdl> There's a lot of work left to do, even once I get the compiler finished.
<jdl> What I really want is to mix multiple metamodels/languages to create a single program.
<ultima_> jdl what compiler
<jdl> ultima: Prism compiler.  See for yourself -- http://www.teleport.com/~sphere/documents/0009/1/index.html
<Beholder> utl:  He's the prisim guy from the mailing list
<ultima_> Beholder oh
<jdl> If you know Java and want to help, I'd be happy to take it.  :)
07:30pm
<ultima_> Sooooooo...whats prism?
<Beholder> ult: As I understand it, it's kinda like tunes
<jdl> Short answer: a compiler-compiler.
<jdl> I've got to go.... Tril, if you're reading this, I'll be back by 8:00 PST.
ωνω SignOff jdl: #TUNES (Leaving)
<Beholder> jdl:  Bye, thanks for the info :)
07:40pm
π Tril/#TUNES is back from the dead. Gone 1 hrs 13 min 24 secs
π _QZ/#tunes is BACK [away time = 4h:6m:57s]
π Beholder/#tunes is gone
ωνω SignOff Beholder: #TUNES (Read error to Beholder[ppp-073.m2-2.sub.ican.net]: Connection reset by peer)
<hcf> whats the url for fare's linux asm howto?
π Tril/#TUNES summons the daemon of Jim Little to appear online
<ultima_> Its on the LDP
<hcf> ultima_: i know, but isnt it on fare's hp also?
<Tril> http://metalab.unc.edu/pub/Linux/docs/HOWTO/Assembly-HOWTO
<ultima_> yeah maybe
<ultima_> ~fare?
ωνω jdl [jiml@ultra1.inconnect.com] has joined #tunes
<jdl> Hi, Tril... sorry I'm late
<hcf> Tril: http://www.tunes.org/~fare/ ?
<Tril> wb jdl only 6 mins np
<Tril> are you familiar with AOP?
<jdl> Me?  yes
<Tril> BTW, you can now use http://tunes.org/..that's the same as www and www2 (all go to bespin).  Sweety is www1 (we're sort of giving up on it, it goes down too much)
<Tril> jdl: just checking, I'm reading these prism essays on the list and I couldnt help notice how closely Prism resembles an aspect weaver :)
<Tril> you are probably more familiar with AOP than I am, though.
<Tril> but this message by David Jeske says exactlry that :)
<jdl> Tril: Others have said the same.  :)  But I see AOP as "weaving a carpet" and Prism as "sewing a quilt" if that makes any sense.  Prism's mixing of metamodels much looser (and flexible, I hope) than AOP's.
08:10pm
<jdl> I looked at AOP a year or two ago, haven't kept up on it.
<Tril> i don't think that would be too hard, Fare says aop is nothing new but just one example of metaprogramming.
<Tril> you read Fare's LL99 article on Metaprogramming?
<jdl> Yep, it has some good insights.
<jdl> I like the explicit definition of metaprogramming, something I hadn't though to do.  However, I think he (and just about everybody else) errors by assuming that the syntax and semantics of a language are inextricably tied together.
<Tril> hahahahahaa
<jdl> I see syntax as a communications medium, unimportant other than it's needed to specify semantics.  The important part -- the MODEL -- is embodied in semantics.
ωνω lar1 [Larman@1Cust252.tnt22.sfo3.da.uu.net] has joined #tunes
<lar1> Hi!
<Tril> hey lar1 :)
<jdl> Hola!
<lar1> I have been B-U-I-S-Y!
<Tril> lar1: you wanted to help out and do stuff? ultima needs slaves, he says
<lar1> Slaves eh?
<hcf> lar1: link check the review's languages page
<Tril> jdl: how do you specify semantics? you have to use syntax right? So which one you see just dpends on your point of view, doesn't it?
<jdl> As long as slaves are being passed out, I could use a Java programmer or two  :)
<lar1> hcf: where?
<hcf> lar1: http://www.tunes.org/Review/Languages.html
<lar1> Thankx
<jdl> Tril: That's just it!  You have to use a language to convey semantics, but the semantics exist independently of the language!  So Prism defines data structures which can be given semantics, and then any language may be used to describe those struct
<jdl> ures.
<lar1> Anyone program for the Ti-86?
<Tril> is that the one that comes after the ti-92?
<jdl> lar1: I programmed for the TS1000, does that count?  :)
<lar1> Hmmm, no
<Tril> sinclair?
<jdl> Timex-Sinclair 1000, yup.  My first computer.  :)
<Tril> I have one, but it's broken. the cable that connects the keyboard to the circuitboard is cut.
<jdl> Ditto
<Tril> never used one. i just collected it. my first computer was a Ti-99/4a
<Tril> then the C=64 and then to pcs
<lar1> My first computer was some big TI thing...
08:20pm
<Tril> jdl: i'm going to reply to the first part of what you said, then i'll need to ask about the second part
<Tril> jdl: a syntactic structure has semantic "meaning", but the meaning depends on the context-- on the environment it's interpreted in.
<Tril> then you asid prism lets you define structures, and specify semantics for them (interpretations)...
<Tril> that is ok so far.. but what do you mean "any language may be used to describe those structures"? 
<Tril> defining "valid" structures and mapping them to meaning is what a language is.
<Tril> or, does.   (whichever?)
<Tril> s/asid/said/
<jdl> Sorry about that..
<jdl> To answer your questions...
<jdl> Mathematically, as I understand it, a language does not define meaning, only syntax.  That's what I meant when I said "language."
<jdl> What you're talking about, I would call a "metamodel."
<Tril> an interpretation for a language.
<jdl> ?
<Tril> "mathematically", a language is a possibly infinite set of finite-length strings of characters taken from a given alphabet (set of characters).
<jdl> Precisely.  Although there are also attempts at formalizing visual languages, such as maps.
<Tril> an interpretation is a mapping from every expression(string) in the language, into some other domain (which is defined as "lower-level").
08:30pm
<Tril> the interpretation gives a semantic meaning to each expression in the language.
<ultima_> Tril hrmmm
<jdl> In my opinion, that's backwards.  (more)
<_QZ> hehe, i was able to mount a second heatsink/fan on the back of my celery
<ultima_> Tril: So, now we know what a language is =)
<jdl> I agree that that's the way it works.  (more)
<_QZ> homemade celery sandwhich
<ultima_> _QZ: Heh.
<jdl> But in Prism, you define the semantics of your metamodel first, typically in natural language (but you could use a formalized semantics language).  (more)
<Tril> defining semantics first is good..
<jdl> Then you create programs which translate models from one metamodel to another.  (more)
<jdl> These translation programs should not change the semantics of the model, although they could perhaps add to them (e.g., translating to machine code adds specific timing semantics).  (end)
<Tril> hmm....
<Tril> let me read some more of these letters.
<jdl> Which one are you on?  Some are better than others. :)
<jdl> "Prism Mechanics, Part 2" is the best one to read.
<Tril> yes, that one.
<jdl> Be aware that I've changed _pBits32 to _pStream (as mentioned earlier).
08:40pm
<Tril> jdl: you take the 3 primitive types, and build "arrangements" of them. How can you claim that is not syntactic (structural)?
<hcf> Tril: does LispOS hav an official hp?
<Tril> hcf: No, just their mailing list. It's not one project, but a jumble of people with different projects (sound familiar?)
<Tril> there used to be a page of links to lispos-related stuff, but it went down.
<jdl> Tril: You can think of it as syntactic if you want.  Maybe calling it a "semantic" meta-metamodel is just hubris.  However, data structures are conceptual, not physical, which is why I chose that word... and in a metamodel, data structures ...
<jdl> ... correspond directly to semantics.
08:50pm
<Tril> No, data structures are given semantics BY the metamodel. 
<Tril> by themselves they are just syntax
<jdl> Tril: Okay, granted.
<Tril> not like it makes much difference. let's not argue over terminology..
<Tril> can prism programs invoke the prism compiler at runtime?
<jdl> Good.  You're right, of course, but there is a big difference between the syntax of C and the "syntax" of a Prism metamodel.  There's a more direct (one-to-one) mapping in Prism metamodels, at least as I see it.
<Tril> i.e. do they run in a "prism environment", like LISP programs?
<jdl> Umm...
<Tril> who defines the one-to-one mapping?
<jdl> The Prism "compiler" is more accurately a virtual machine.
<jdl> The programmer who creates the metamodel.
<jdl> The Prism virtual machine executes "Reflection" models... models which manipulate the Prism compiler.
<lar1> Does anyone here use Netware?
<jdl> Umm... that probably doesn't make any sense.
<Tril> lar1: My school does.
<ultima_> jdl why did you write prism in java?
<ultima_> a virtual machine inside of a virtual machine
<ultima_> when will prism be able to produce native binaries of itself and thus be truly reflective?
<Tril> jdl: yeah it makes sense. the question is, is the reflection domain reuqired, or optional?
<jdl> ultima: Prism is a compiler specification -- I wrote the current Prism compiler in Java for speed of development.
<jdl> Tril: just a sec...
<lar1> Am I better off using Linux + Samba on my 386/40 or Dos 6.2 + Win 3.1 + Netware 1.1?
<Tril> lar1: what are you going to network with?
<jdl> ultima: Not for a while... that would be pretty hard.  It is a goal, however, at which point it would be very similar to Tunes, I think.
<lar1> My 3 win95 computers
<ultima_> I see...
<jdl> Tril: The reflection domain is required to translate models from one metamodel to another.  It's not required to define metamodels.
<Tril> lar1: Netware interacts VERY poorly with any Windows. You should avoid it at all costs.
<Tril> lar1: (i meant, you should avoid mixing Windows 3.1/95/NT with Netware)
<jdl> s/reflection domain/reflection metamodel
<jdl> Tril: The reflection metamodel is how you write your translation programs.
<Tril> jdl:I think you missed the point of my question. I'm wondering about the timing. What output does prism produce and what kind of environment does the output(a program) require?
<lar1> Tril:  Ok
<jdl> Tril: Whatever output you want.  A Prism "compiler" doesn't actually compile.  It interprets Reflection models which do the actual compilation.
<lar1> Will Xwindows run useablly on a 386dx/40?
<Tril> lar1: No, but the more RAM you have, the better.
<lar1> Tril:  I have 16meg...
<jdl> Tril: My choice of words -- "Prism compiler" and "reflection metamodel" isn't the greatest.  ;)
<ultima_> X server will
<ultima_> but X with apps on the local machine wont
<Tril> lar1: X is ok.. but don't run Netscape or any big apps with it
<jdl> s/words/names
09:00pm
<lar1> Tril:  Ah, ok.
<Tril> lar1: It will work, just very slowly.
<lar1> Will Liunux boot from the slave drive on my secondary IDE channel?
<jdl> Tril: A Prism compiler may output machine code, just as a C compiler can -- if you create the right Reflection model to make it do so.
<Tril> lar1: go ask in linpeople I'm not an expert on lilo and we are talking about tunes here :)
<Tril> err, I mean prism
<lar1> Tril:  Oh, yah... hehe
<jdl> :)
<lar1> Can I plug my SCSI hardrive into my sound card's SCSI port?
09:10pm
<hcf> lar1: 42
<lar1> hcf:  What?
<hcf> lar1: the answer is 42
<lar1> Ah, I see
<Tril> lar1: how old is the drive
<jdl> :) lar1: I wouldn't recommend it.  That "SCSI" port is pretty limited, I think.
<lar1> Tril:  I dunno, it is at least 4 years old... a 250Mb from a mac
<Tril> lar1:yueah it's for CDroms, hard drives would run pretty slow.
<jdl> Tril: Did you see my response to your last question?
<Tril> jdl:are you asking if I paid attention or if you are desynched?
<jdl> Umm... dunna about desynched (?) but I didn't see a response from you...
<jdl> s/dunna/dunno (don't know)
<Tril> I see it.  but how does that work?
<jdl> Be more specific...?
<Tril> "..if you create the right reflection model to make it do so" -- how do you do that?
<_QZ> SUCK 
<Tril> i kind of didnt answer because I'm reading more of these mails in an attempt to understand...
<_QZ> i just spent 30 minutes hooking 2 heatsinks onto the wrong celeron
<jdl> Tril: That's what I thought, just wanted to make sure you were still here... :)
ωνω GMOL [gmol@h24-64-168-191.mt.wave.shaw.ca] has joined #tunes
<lar1> _qz:  hehe that sucks!
<jdl> Tril: The Reflection metamodel is defined in "Prism Mechanics, Part 3" if you want to read about it there... or I can answer questions.
<jdl> In a nutshell, Reflection models control the Prism compiler and can read and modify other models loaded into the compiler.  Reflection models are also Turing-complete, which is how they can translate models into machine code.
<jdl> Tril: Does that help?
<Tril> i don't know... :)
09:20pm
<jdl> :) I'm not always very clear.  :b  What can I clarify?
<GMOL> jdl One thing that I don't like (despite the fact that I don't understand it fully yet:) is that the programmer has to make sure domains can interface with each other properly...in the spirit of reflective metprogramming, I feel that this is a task that should be left to the computer...no?
<Tril> that's sort of what my type system is for..helping to automate translations between models
<jdl> gmol: That's a difficulty I haven't resolved yet.  :b  Ideally, I want to be able to say, "If you do these things when you create your metamodel, it will automatically interface with any other metamodel."  But I'm not there yet.
<jdl> gmol: Who are you on the list?
<hcf> _QZ: http://www.xs4all.nl/~ganswijk/chipdir/
<GMOL> Maneesh Yadav
<lar1> Gotta go... bye
ωνω SignOff lar1: #TUNES (Leaving)
<Tril> jdl: I don't understand the system abstraction / domain abstraction  terms
<Tril> I think of a "domain" as the input or output of a function.
<jdl> Tril: Are you looking at the "Prism Glossary" essay?
<Tril> how do each relate to procedural abstraction, or creating a symbolic name for a procedure?
<Tril> I think that is a system abstraction, but not sure.
<Tril> no, I'm not to that yet
<jdl> Tril: Yes, procedural abstraction (aka lambda abstraction, yes?) is an example of system abstraction.
<Tril> going to look at it
<Tril> yeah i was thinking of lambda-ish abstractions..
<jdl> Tril: It's a glossary, doesn't assume any knowledge.  I think it will help.  :)
<GMOL> that's what I was just looknig at, jdl I still don't what a domain has to do with a metamodel, could you elaborate?
<jdl> Okay, let me see... 
<Tril> here's my question..dunno if it relates to gmol's..
<jdl> First, before I wrote the glossary, I mistakenly used the word "domain" to mean "metamodel."  So that may be what you're thinking of.
<Tril> a domain abstraction is supposed to "simplify" a domain? Which direction would that be, to a regular compiler? Is ASM "simpler" than BASIC, or is it the other way around?
<GMOL> crap I have a quiz in inorganic chem tommorow, must study, later all
ωνω GMOL [gmol@h24-64-168-191.mt.wave.shaw.ca] has left #tunes []
<jdl> gmol: You achieve domain abstraction by creating a metamodel, just like you achieve system abstraction by writing a function... drat, he's gone.
<jdl> oh well
<Tril> Hmm, guess Canada doesn't have Martin Luther King Jr. day
<jdl> :)  Let me try to answer your question...
09:30pm
<jdl> First, I would say that BASIC is a domain abstraction of ASM.  BUT...
<jdl> technically I would say that any substition of one domain for another was a domain "abstraction."
<jdl> Fare is better at defining these things than I am, maybe you could get him to do a nice formal definition.  ;)
<Tril> translating from BASIC to ASM is performing the action of "concretion" :-)
<jdl> excretion?  :)
<Tril> no you must be thinking of the Win32 API.
<jdl> Ah, my mistake.  That would be it.  ;)
<jdl> back in a few
<Tril> ok when you get back, this idea of semantic models really annoys me. I agree with using abstract concepts instead of symbols to convey meaning, but then you go and say data structures are abstract concepts, but they're just syntactic structures, too!
<Tril> it sounds like you are attaching semantics to the syntax of "Data structures" too closely--I mean inseparably. But I believe syntax and semantics should be kept separate, (so you can assign new syntax to the same semantics)
<jdl> ok, I'm back
<jdl> Hmmm... maybe so.
09:40pm
<jdl> But that's the best I can do.  (shrug)
<Tril> i dont think that can be avoided. anyway, it's not too important. here's my next q:
<Tril> semantics is meaning, or a mapping FROM some syntax TO other syntax.
<Tril> note you can only map to other syntax. There is nothing else to map to.
<Tril> At every level in teh system, there is just syntax.
<Tril> at the base of the system, you have machine instructions, which are still syntactic...
<jdl> Except in your head, and even that could be considered syntax -- the syntax of firing neurons.
<Tril> in my opinion, meaning is just the relation of something to everything else. And a relation is just syntax
<jdl> Hmm... I'm not sure if I agree.  But I don't know how to disagree either... this seems more like a cognitive psychology issue.
<jdl> Look up "semiotics" some time. :)
<hcf> abi: webster for semiotics
<abi> semiotics may be sought at http://work.ucsd.edu:5141/cgi-bin/http_webster?semiotics
<jdl> I think you'd find it fascinating.  I just started reading up on it myself.  It's the study of signs and meaning.
<hcf> "a philosophical theory of the functions of"
<jdl> signs == syntax
<hcf> its not ology, thus not 'the study of'
<jdl> whatever
<jdl> My dictionary defines semiotics as "of or relating to semantics".
<Tril> writing a function,you say, is system abstraction, or taking a group of objects and replacing them by one single object. Right?
<jdl> yep
<Tril> isn't a function also a domain abstraction, between the set of possible arguments and the possible results?
<Tril> a domain abstraction is a mapping between two domains , right? (you called them metamodels?)
<jdl> Yes, and yes!  If the function is applied at compile time (to make this more concrete) then I would call it a domain abstraction.
<jdl> I think.
<jdl> Tough questions... :)
<jdl> Why compile time?  I dunno.
09:50pm
<Tril> the domain abstraction is of a higher abstraction level than the system abstraction.  Why? Because the source code for performing a mapping isn't the mapping itself. The mapping is the COMPLETE SET of all associations between elements of the domain and elements of the range.
<Tril> which you only allude to with some source code
<jdl> Hmmm... I think you better say that again, in one syllable words this time ;)
<Tril> it's like saying "2x+1". you don't mean literally, multiply the input by 2 and add 1. You mean, the set of all possible sequences of operations that have the SAME RESULT as multiplying the input by 2, and then adding one...of which there are an infinite number!
<jdl> Okay, so you're saying that the literal arithmetic is a system abstraction, and the second thing is a domain abstraction.   Yeah, I'd go for that.
<Tril> 2x+1 is just shorthand, and is used because it is the "minimal" expression having the same meaning.  Bigger expressions are "2x+5-4", "4x/2+1-0-0-0", and so on
<jdl> Where is this leading?
<Tril> I was going to say that there is a distinction between these two views of a function.
<hcf> stenography
<Tril> In one context (in the system), it must be treated as "the set of all functions semantically equivalent to the function modelled by this expression"
<jdl> Okay.  I have to admit, I have trouble seeing domain abstraction at this level.  I think I understand what you're saying, but perhaps because I can't see a practical application, it's a little confusing.
<jdl> That makes more sense.
<jdl> I think.
<Tril> The reason this is important is that is what a specification is
<jdl> specification?
<Tril> are you familiar with my "hierarchy of specification/implementation pairs"?
<jdl> No, sorry.
10:00pm
<Tril> a specification is just an expression in some language, and an implementation is a corresponding expression in some other language that the first language is a domain abstraction for.
<jdl> I think I just understood something that you said earlier.  2x+1 = 5 (when x=2) is a system abstraction.  2x+1 == 2x+(5-4) is a domain abstraction.  Yes?
<jdl> Are you waiting for me to respond?
<Tril> no. forget the system/domain abstraction here.  2x+1=5 when x=2 is an extensional evaluation of the expression 2x+1, that is, getting its value=extension.   The set of all expressions equivalent to it (i.e. the meaning of the operation) is the intensional interpretation of the same expression. (I think)
<jdl> I don't understand.
<Tril> 2x+1 only = 5 when you evaluate 2x+1.
<jdl> Okay
<Tril> 2x+1 = the set of all functions that do the same thing, only when you interpret 2x+1 (examine its semantics)
<jdl> Okay.
<Tril> the extension of an expression is its value (in this case, 5)
<jdl> Is that your term?
<Tril> the intension of the expression is its semantics (in this case, a function of x to integers)
<jdl> Okay.
<Tril> these are a couple of terms I am trying to make very clear, because they are important in my model.
<Tril> whether for use in this conversation, or my own benefit later, who knows. :)
<jdl> Are they standard terms, or did you define them?
<Tril> they are actual terms, which I am trying to understand :)
<jdl> I definitely see a relationship between your work and mine, although I can't say exactly WHAT it is. :)
<Tril> now.. what exactly IS the semantics of 2x+1? well, the complete, explicit, semantics are a set of associations from integers to intgers (a mapping between 2 infinite sets, which is also infinite)
<Tril> Therefore, by saying 2x+1 and interpreting it intensionally you DEFINE AN INFINITE SET.
<jdl> Okay, I like that.
<Tril> If you want to prove that something is equivalent to 2x+1, you just show that their intension results in the same infinite set.  (ugh)
<jdl> The semantics of 2x+1 are 1->3, 2->5, 3->7, ad nauseum.
10:10pm
<jdl> Okay
<Tril> now where was I going to go ... :)
<jdl> :)
<jdl> Like I said, I see a relationship, but WHAT relationship I don't know.  :)
<Tril> all right, more theories to try out on you :)
<jdl> Hold on... something to say...
<jdl> Perhaps the relationship is this:
<jdl> Prism is not Tunes.  But Prism could be Tunes if it were an "OS," or a fully-reflective runtime system.  (more)
<jdl> The piece that's missing for Prism to be Tunes is a way to make it fully-reflective.  (more)
<jdl> That piece, I think is a means of formalizing semantics so that ... ? Oh I don't know.  Go on... (end)
<Tril> i think i'll ask more questions about prism instead
<Tril> so everything is system abstractions or domain abstractions?
<jdl> That would be good, because I have to go soon...
<jdl> Ummm... no, not in Prism.
<jdl> System abstraction and domain abstraction is part of the philosophy, PISE.
<jdl> Prism is an attempt to realize the philosophy.
<jdl> Maybe I'm being weasel-worded.
<jdl> Assume I said yes, let's see where that goes.  :)
<Tril> ok..you asked if my model was a metamodel..I said I didn't know. well thats just because I misread what you saisd after that,
<Tril> that "if not, it won't work very weill in prism", or something. I thought you said "if so, it won't work very well in prism" :-)
<jdl> :)
<jdl> So is it?
<Tril> you asked if I write programs in it..I'm not sure. I think first I'm thinking of a framework to define domains, or spaces which can be the inputs or outputs of functions.
<jdl> Let me put it this way:
<jdl> Can you use it to model physical things or abstract concepts?  If yes, it's a metamodel.
<Tril> let me just say yes..i'm sure it is, but I can't say exactly how right now..
10:20pm
<jdl> In that case, I think it can by modeled as a metamodel with Prism's meta-metamodel.  ;) (Say that three times fast!)
<Tril> metametametametametametamodel!
<jdl> LOL
<jdl> metamucil?
<jdl> Okay, so what exactly is your ____?
<Tril> ok sorry modem just hungup on me
<jdl> 'sokay
<Tril> Ahh...let's start trying to implement my idea in prism
<Tril> start with a "predicate". do you knwo what that is?
<jdl> Define it for me.
<jdl> Actually...
<jdl> it would help if you told me what your idea's "reason for being" was.  Oh yeah, give me a name, too.  :)
<Tril> it's supposed to be a unified typesystem. that means it's job is for any object and any operation, in the entire system whatsoever, it will decide whether or not it is OK to do it (perform the operation on the object).
<jdl> Could you give me an example?
<Tril> therefore it is responsible for the security (i.e. that everything does what it is supposed to) in the system.
<Tril> example: I want to divide by zero. I have a division operator, and two arguments (say they're both zero), and I ask the question, "Is this a valid expression?"  It is the job of the typesystem to answer Yes or No.
<jdl> Hmm.... I don't think that's a metamodel... but...
<jdl> I think it's PART of a metamodel.
<Tril> Since everything is an object, you are technically able to form expressions consisting of any object in any relation to any other object. Therefore some mechanism needs to exist which tells you what is allowed, and what is not.
<jdl> Granted.
<jdl> I think that's part of the semantics of the metamodel.
<jdl> In other words...
<Tril> In order for the typesystem to function, it requires an extensive "metamodel", or framework of information about objects.
<jdl> A metamodel should say that the expression 4/0 is not a valid model.
<Tril> it's going to make USE of a metamodel, anyway... and, to be reflective, the model of this typesystem will be IN the metamodel, but that is not until later..
10:30pm
<jdl> slow down  :)
<jdl> I need to think about this for a moment.
<Tril> I'm more used to running ahead and then jumping back :)
<Tril> but ok.
<jdl> Let's back up for a second.  How will this universal typesystem be used?
<Tril> Anyway, the information required, is an INTEGRATED model of everything in the system. So, ok, I guess the typesystem is the (meta?)model.
<Tril> it's kind of like your domain abstractions, except that all domain abstractions have to be related into the master typesystem somehow.
<Tril> Otherwise they can't even be construed to exist.
<jdl> Sorry, but none of what you just said made sense.  :)
<jdl> (to me ;) )
<Tril> How will it be used?
<Tril> It will be used for everything. It is the entire system and everything you can do in it.
<jdl> ?
<jdl> just a sec...
<Tril> I just call it a universal typesystem, because that's what I think tunes is :)
<jdl> Give me a use case. :)  You mentioned dividing by zero... step through that scenario, starting with where the division by zero comes from.
<Tril> but more simply, you can't really do that much. you can build expressions, and "move" them around. In essence there's only one thing. building expressions is the same as moving them around.
<Tril> ok. the example
<Tril> i have an expression editor, in which i move objects around (independent of any meaning, this is just a structure editor)
<Tril> and combine them to form expressions.
<jdl> okay
<Tril> "moving them around" is just bringing them into my current expression, and then rearranging the expression in different ways.
<jdl> okay
<Tril> the units are obviously objects, and since they have no meaning, there is no difference between them (except their symbol, which the editor doesn't care about)
<jdl> So they're just draggable blocks of pixels.
<Tril> so i have a couple of "0" objects, and a "/" object
<jdl> okay
<Tril> well, in a gui, they appear that way, sure.
<jdl> what are they actually, then?
<jdl> Just faceless "objects" with names the computer doesn't care about?
<Tril> we'll get to that, later :)  that question was very difficult for me to answer.
<jdl> okay... I have a feeling it's an important question, then.
10:40pm
<jdl> But go on.
<Tril> the "/" object has an interface defined (I guess that's the term  most everybody is familiar with)
<Tril> the interface says it is a binary operator.
<jdl> okay
<Tril> so in the editor you can see two slots on the "/" object, or, at least two arrows pointing at it (or something, yo uget the picture)
<jdl> got it
<Tril> assuming the editor is just looking at the syntax (it is, for now), all it knwos is there are 2 arguments. and it lets  yo place objects in the position for the arguments.
<jdl> okay, so you form "0/0".
<Tril> so finally when I have painstakingly (sarcasm) dragged the arguments in place I have a formed expression in front of me, an abstract syntax tree. (heard of that before? :)
<jdl> sure.  It's a semantic metamodel.  ;)
<jdl> s/metamodel/meta-metamodel
<Tril> then i'm going to hit the button (big, and red) that says "GO!"
<Tril> or should it be green.
<jdl> make it polkadotted... go on.
<Tril> anyway that says to evaluate it. If it is invalid, it can't be evaluated. So right before it is evaluated, it is typechecked.
<jdl> And that's where your UTC comes in.
<Tril> obviously we want an error to come up, and the expression to be unchanged, and nothing to happen. (except the error)
<jdl> go on
<Tril> now for the scope of this.. in tunes everything you do (any computation) is represented by an expression. Any action is achieved by "evaluating an expression". 
<jdl> Okay, you can stop.  I understand.
<Tril> And, since tunes is an inhernetly secure system, all these potential evaluations muts be typechecked...
<Tril> so you can see the importance of a typesystem.
<Tril> (i stopped)
<jdl> Okay, let me ramble for a bit here...
<Tril> sure! :)
<jdl> First, Prism can't help you.  :)  But wait, there's more...
<Tril> lol
<jdl> Tunes (and by extension, the UTC) is a model... a running system.
<jdl> Prism provides no help in defining models.
<jdl> HOWEVER.
10:50pm
<jdl> I think by making your UTC part of Tunes (and thus reflective)
<jdl> you're imposing a large cognitive burden on yourself.
<jdl> I.e., it's hard enough to think up a UTC, let alone one that works in an unspecified reflective system.
<Tril> wait, a model is not a running system. what did you mean?
<jdl> gimme a second, I'm on a roll here.  :)
<Tril> that's fine
<jdl> I think the goal of a UTC is a good one, but perhaps you should think about how to make it work with metamodels before making it work with reflective models.
<jdl> That would break the problem into two disparate parts: 
<jdl> 1) UTC
<jdl> 2) UTC in a reflective system.
<jdl> Personally, I have no idea how to make it work at all.  ;)
<jdl> That's it.  To answer your question, I shouldn't have said model.  Tunes is a running system (or at least it will be).
<Tril> anything else?
<jdl> done
<Tril> so what you meant was "tunes is a running system. Prism provides no help in defining running systems"?
<jdl> I think so.  :)  Reflection confuses the bajeezus out of me.  :)
<Tril> i'm not even sure what you mena, because if I have a model for my system (an idea of it), prism should be able to manipulate and implement my model.
<jdl> Okay, what I meant was that Prism can't manipulate your model until you define a metamodel and implement your model within the metamodel.
<jdl> And of course, it has to be a Prism metamodel. (more)
<Tril> And what you implied was that TUNES is its own metamodel (and therefore reflective), so prism can't help there.
<jdl> I think so, yes.
<Tril> But you can say that for any language. We already know current languages are bad for writing reflective systems, that's why only a reflective system is good for writing reflectibe systems :)
<jdl> Okay, how's this:
<jdl> Before you can use Prism to define your UTC, you need to define a metamodel for reflective systems.  :)
<Tril> By definition of reflection, a reflective system can express itself. Therefore we conclude that reflective systems should be ideal for implementing reflective systems!
<Tril> (at least ones in some way similar to themself)
<Tril> selves
11:00pm
<jdl> A reflective system is ideal for implementing a reflective system identical to itself.  That, I'll give you.  <big grin>
π Tril/#TUNES freaks... isn't that what Brian said, that he has no idea how to give a specification for reflection in general?  :-)
<jdl> Here's how I think you can use Prism:
<Tril> I think you use "metamodel" sometimes when you mean "model".  Either that, or I'm missing something.
<jdl> Start small.  Define a metamodel for reflective systems that is minimally reflective, and see where that takes you.
<jdl> As for metamodel/model -- it's possible, but I haven't noticed me doing that.  :)
<jdl> Sometimes I interchange model and system -- the system is what runs, the model represents the system.
<jdl> system == program
<Tril> well, could it be that you always think of compiler compilers, so everything in prism is a metamodel (a model of a model) , because prism manipulates models..
<jdl> A metamodel isn't a model of a model.  That may be the confusion.
<Tril> jdl:in reflection it's even more easy to interchange those, when your system is acting on its own model :)
<Tril> sure it is. you said define a metamodel, then define your model in it.  
<Tril> the "model" is a language, isn't it?
<jdl> Whoops!  No.
<Tril> or you mean the model is a particular program's source.
<jdl> Examples:
<jdl> Source code is a model.
<jdl> C is a metamodel.
<jdl> Phrase-structure grammar (BNF) is a meta-metamodel.
<jdl> A metamodel defines a set of models.
<jdl> A meta-metamodel defines a set of metamodels.
<jdl> (done)
<Tril> metamodel=language (kinda). meta(2)model=grammar (yes)
<jdl> no
<Tril> model=expression
<jdl> metamodel = language/grammar
<jdl> meta-metamodel = type of grammar
<jdl> model = representation of running program
<jdl> program = electrical activity
<jdl> (done)
<Tril> uhh..off topic, but what do you do about multiple levels of models between the source and the electrical activity. there is the machine language, then assembler, which is at least 2 possible levels..and if your model is LISP you may translate from that to C giving 3 intermediate layers..
<jdl> Not off topic at all.
<jdl> machine language is a metamodel.
<jdl> Assembler is a metamodel.
<jdl> Assembler is a domain abstraction of machine language.
<jdl> LISP is a metamodel, and a domain abstraction.
<jdl> C is a metamodel.
<jdl> etc.
<jdl> (done)
<Tril> so the metamodel which is the domain abstraction of a particular language, is not necessarily the same language which is the meta-metamodel for that language.. ?
<Tril> since the meta-metamodel of Machine language is an instruction set definition
<Tril> (the opcodes)
<jdl> Ummm.. let's leave out meta-metamodels for now.
11:10pm
<jdl> Machine language is a metamodel, and that metamodel defines the semantics of sets of bits.  Part of those semantics is the instruction set definition.
<jdl> (done)
<Tril> the meta-metamodel of Asm (no, let's not, this has been bugging me for a while) is the syntax of asm (order of columns in an asm file: label: instruction: operands: comments
<Tril> the meta-metamodel of C is a phrase structure grammar or whatever
<jdl> Do you know what phrase-structure grammar is?  (How about BNF notation?)
<Tril> justasec..what is the meta-metamodel of lisp?  
<jdl> Phrase-structure grammar, I think.
<Tril> here's some confusing terms: LISP is a metamodel, so we shouldn't say "what is the meta-metamodel of lisp", we should say "what is the metamodel of lisp" since we want a model for lisp.
<Tril> the metamodel of lisp happens to be a meta-metamodel, in absolute terms, but not compared to lisp.
<jdl> Hmm... okay, I'll go with that.
<Tril> and how do you distinguish between the model of lisp, and the model used to model the model of lisp
<Tril> a model of lisp is a meta-model but is a particular expression IN PSG
<Tril> PSG is the metamodel of that expression?
<jdl> Wait...
<jdl> What do you mean, "A model of LISP?"  A LISP model -- i.e., a LISP program?
<Tril> no, I mean the grammar for lisp.
<jdl> Okay, yes and yes.
<Tril> the actual set of "productions" which generate the valid expressions in the lisp language.
<jdl> yes.
<Tril> I just call that the specification for lisp. 
<Tril> (formal specification, to be sure)
<jdl> Okay, so we could call it "the LISP metamodel" or "LISP's model".  (more)
<jdl> And LISP's metamodel is PSG, which happens to be a meta-metamodel in absolute terms.  Yes?  (done)
11:20pm
<Tril> PSG is a meta-metamodel compared to a lisp program.
<jdl> Exactly.
<Tril> I don't even want to get into what the lisp program is modelling. Let's just say it prints "Hello, World." and be done with it.
<jdl> It's modeling disgusting cheerfulness. :)
<Tril> I mean, if it were an implementation of TUNES, then it would be a meta(infinite)-model.
<jdl> I am my own grandpa!  (great song, BTW)  :)
<jdl> That's why reflection confuses the bajeezus out of me.  ;)
<Tril> anyway, my revelation was how to understand the difference between, a language used to model another language, and a language which is the domain abstraction of another language (i.e. it outputs programs in the second language)
<jdl> whoops, another terminology problem.
<Tril> what, the word "language"?
<Tril> :)
<jdl> No, the word "the"  ;)
<jdl> Seriously -- "domain abstraction"
<jdl> Why don't you define it.
<Tril> Oh, I see what you mean.
<jdl> Good, what do I mean?  :)  (Just so we're on the same track)
<Tril> if language A is "compiled to" language B, I can't say "language A is the D.A. of language B", instea dI have to refer to the trasnlation function from A to B, and say "function F is the domain abstraction of B, <something or other> A".
<Tril> where <something or other> is "from", or "resulting in", or possibly some other term neither of us has heard of.
<jdl> Okay, I misunderstood what you were saying the first time.  I wouldn't use the term "domain abstraction" at all in this case.  Example:
<jdl> Okay, not an example, just more explanation... I use domain abstraction to refer to actual models.  "This C program is a domain abstraction of this ASM program."  Make sense?
<jdl> (more coming)
<Tril> ok, I can see how that makes it simple, then what do you call the other thing (what I meant)?
<jdl> Well, technically any Turing equivalent metamodel can be translated into any other Turing equivalent metamodel.
11:30pm
<jdl> But...
<jdl> What you're saying, I think, is that language A's semantics are "higher-level" than language B's, so that we can translate A into B without losing any meaning, but we can't translate B into A without losing meaning.
π Tril/#TUNES intends to find a computatation which cannot be expressed by a Turing machine
<jdl> (done)
<Tril> well, what I was really referring to was the fact that *a compiler exists* from language A to B.
<Tril> regardless of information loss (although a compiler usually implies that)
<jdl> Okay.  Let's say that "A compiles to B" then.
<jdl> or "translates to".
<Tril> if program A (written in C) is a domain abstraction of program B (asm), 
<Tril> isn't it a model?
<jdl> Both programs are models.
<jdl> Urk.  Both source code listings are models.
<Tril> are yo writing something? I'm drawing a picture of all this.
<jdl> nope, just waiting for you :)
<Tril> ok, then hold on :)
<jdl> Wish IRC could show pictures.
11:40pm
ωνω SignOff ultima_: #TUNES (Ping timeout for ultima_[user-37kb9uu.dialup.mindspring.com])
<jdl> Tril: still alive?  :) What's your picture look like?
<Tril> uhh.. I started over. I'll ask a question before I finish it. let's see if i know what a PSG is
ωνω ultima_ [ultima@user-37kb9uu.dialup.mindspring.com] has joined #tunes
<jdl> The C grammar is at http://bit.csc.lsu.edu/tutorial/ten-commandments/ANSI-C-grammar-y.html
<jdl> Here's a simpler example:
<jdl> <simple sentence> = "A" <noun> <verb> "on the" <noun>
<jdl> <noun> = "cat" | "dog"
<jdl> <verb> = "sat" | "crapped"   ;)
<jdl> (done)
11:50pm
<Tril> do yo know the hierarchy of languages?  e.g. regular languages are a subset of deterministic context-free languages, which are a subset of context-free, are a subset of context-sensitive, which are a subset of unrestricted grammars (which is another term for phrase-structure grammars).
<Tril> oops
<Tril> context-sensitive languages are a subset of recursively enumerable languages (which are described by PSGs).
<Tril> regular languages are described by regular expressions; deterministic context-free languages are described by LR(1) grammars
<jdl> Actually, I wasn't aware of that.  To be honest, I was never confident that PSG was actually the right meta-metamodel.  In fact, I only intended to use it as an example to explain what Prism's meta-metamodel was, by means of comparison with...
<jdl> ...traditional languages.
<Tril> it says here recursively enumerable languages are Turing equivalent
<jdl> cool :)
<Tril> (the other more restricted languages are not)
<Tril> ALL the types of languages I listed are describable by grammars; with appropriate restrictions for each language (and no restrictions for rec.enum. lang, which is why they are called unrestricted grammars)
<jdl> So where are we?
<Tril> uh, I'm trying to find out how many dimensions we're in.
<Tril> there's the dimension of "I am a meta-model of you"
<jdl> okay
<Tril> there's the dimension of "I am a domain abstraction of you"
<jdl> ick
<Tril> and I want to draw that on a 2D piece of paper
<jdl> Leave out the domain abstraction dimension, trust me :)
<jdl> Domain abstraction is a philosophy, metamodels are the means... as I see it.
<Tril> huh?
<jdl> Think of system abstraction.  System abstraction is a way of simplifying software.  But you don't think of system abstraction -- it's a philosophy, something to do.  You think of functions, a means of system abstraction.
[msg(TUNES)] newlog 1999.0118
IRC log ended Mon Jan 18 00:00:01 1999