IRC log started Mon Jul 5 00:00:01 1999 [msg(TUNES)] permlog 1999.0705 -:- SignOff _QZ: #TUNES (BRiX [http://www.qzx.com/brix] :: sleep) -:- smoke [smoke@7dyn184.delft.casema.net] has joined #tunes -:- Cavey [cavey@passion.gimmie.tudelft.nl] has joined #tunes -:- Cavey [cavey@passion.gimmie.tudelft.nl] has left #tunes [] -:- SignOff Fare: #TUNES (Connection reset by pear) -:- hcf [nef@me-portland-us1009.javanet.com] has joined #tunes abi: seen fare? hcf: no idea abi: seen fare 05:40am -:- hcf has changed the topic on channel #tunes to: Free Reflective Computing System -:- tcn [tcn@cci-209150250101.clarityconnect.net] has joined #tunes >>> tcn [tcn@cci-209150250101.clarityconnect.net] requested PING 931184864 91183 from #tunes -:- Fare [rideau@quatramaran.ens.fr] has joined #Tunes hey fare what's up? 07:40am tcn!!! well, still haven't begun the compiler but have upgraded my router from RH5.0 to debian2.1 tcn: news from the IDE driver? tcn: your docs says vm86 is working. tcn: could we use the BIOS to read the HD? tcn: also, why not make retro LILO-friendly? (not hard to do; even tunes 0.0.0.25 did it) ok, i'm here but I'm there it's easier to use port I/O than BIOS, for IDe. Fare: iv got some urls for u based on keywords from implement.ps hcf: great! tcn: ok. but what about writing? http://www.isi.edu/software-sciences/wile/Popart/popart.html, http://www.cse.ogi.edu/PacSoft/projects/FM/, http://www.plattsburgh.edu/faculty/plazaja/research/lp/index.html, http://www.mjolner.dk/warehouse/MPS.html http://www.ispc.com/D-ProcessNotation.htm, http://www.limsi.fr/Individu/martin/ijcai/article.html, http://www.brics.dk/NS/94/6/NWPT6/NS/NWPT6-NS/NWPT6-NS.html, http://www.gr.opengroup.org/adl/ Fare: maybe today.. it's a holiday in the us http://www.brics.dk/FormalMethods/AS/Workshop-99/KutterPierantonio.html, http://www.brics.dk/FormalMethods/AS/Workshop-99/MouraMenezes.html, http://research.microsoft.com/Users/simonpj/Papers/papers.html -:- binEng [Anders@dialup210-4-54.swipnet.se] has joined #tunes hehe.. it's too hot to do anything else, anyway 85 fahrenheit already -:- AlonzoTG [Alonzo@216-164-231-48.s302.tnt7.lnh.md.dialup.rcn.com] has joined #tunes tcn: *yesterday* was a holiday in the US, isn't it? why today? Farenheit 451 ? Fare: yesterday was sunday, so we get work off today :) everything's closed ok, it's 30 celsius here, and it'll be 35-40 in a few hours. fucken hot eh? 07:50am under shadow, or under the sun? tcn: where in the US are you? in shadow. in New York -:- SignOff AlonzoTG: #TUNES (Have Nice Day :)) binEng: here?\ why, how hot is it in france? hcf: yup binEng: must be nice & cool in sweden :) binEng: in case u care, http://www.python.org/doc/essays/metaclasses/ tcn: some 25 degrees C hcf: what *is* that? tcn: it's cool here in Paris binEng: python metaclasses damn.. linux needs kernel support for Quake? classes of classes? Python has made my understand how functional lambdas are bE: by lack of it? :) Fare: no, Python supports some kind of lambda... dunno if it's the real one, but anyway :) oh this is nice, python's lang comparison page doesnt even mention lambda hcf: for a long time, python didn't have lambdas, I was told. but if it now does, it may at last qualify as a real language (unlike C++ and Java 1.0) Fare: they compare python to langs that have lambda w/o mentioning it How's a real lambda supposed to work? 08:00am the comparisons suck, theyre talk mostly about syntax I was told, however, that it was long possible to emulate lambdas in Python in an incorrect way (with ascending and/or descending funarg problem); if the current "lambda" is still that broken thing, then python isn't a real language yet (even Perl has real lambdas) How's a real lambda supposed to work? I maybe can tell if this qualifies... Fare: hey, how's your /etc stuff working? i'd like to see it.. bE: must close its lexical environment with indefinite extend s/d$/t/ tcn: which /etc ? you mean, my meta-/etc stuff at home? Fare: d$? bE: the ending d bE: man regex (or man sed or man perlre) Fare: I don't get that explaination... could you elaborate? bE: let f = \x -> \y -> x+y well, then (f 2) will be the function that increments by 2. and that must not depend on any dynamic variable x or y or another instance of f or anything ic... Fare: yeah does python pass the test? -:- AlonzoTG [Alonzo@216-164-242-134.s388.tnt10.lnh.md.dialup.rcn.com] has joined #tunes 08:10am I don't think so... but I'm not sure how to test it tcn: do you want the /etc for redhat or debian? Fare: debian tcn: ok. Wait. How do you run lambdas with values? (or whatever you do with lambdas...) (set! foo (lambda () (display "hello world\n"))) !HyrlikW:*! do we have anyone from @home online now on this network? !HyrlikW:*! please message me uh, gotta test this some other time... cya FUCK LINUX HARD1!!! -:- SignOff binEng: #TUNES ( ) I HATE IT I HATE IT I HATE IT I HATE IT I HATE IT I HATE IT I HATE IT I HATE IT atg: when's the last time you used it? -:- COMMAND [user3048@bubbette.cs.rice.edu] has joined #Tunes hi, command! command: do you know prof Felleisen? -:- SignOff COMMAND: #TUNES (Leaving) 08:20am just five minutes ago tcn: ok, it's in ~fare/lxconf.tb2 on bespin. Now, 1) not included in the packages is a special directory where I store my secrets (passwords, keys, etc) 2) not included are a few scripts that I wrote, in /usr/local/stow/fare/bin or in ~fare/bin (these are in bespin) LINUX IS EVIL!!! 3) I added a /trans hierarchy for transient system files (not included) ATG: indeed; but it's still the kindest we have 4) some stuff in there is "work in progress" and non-functional 5) do not hesitate to ask me about it allright, cool. I change the machine's identity with /conf/bin/$NEWHOSTNAME I change the machine's current network with /conf/bin/donet $NEWNETWORK force machines are in /conf/hosts/ networks are in /conf/nets/ there are also evil things done upon ppp launch, but must be non-functional due to bit rot since I went from redhat to debian 08:30am work in progress include /conf/NEW, that ought to become a scheme-based way to do things, and /conf/generic/video, that ought to help compress video configuration stuff 08:40am Fare: what kernel version are you using? 2.2.x x depends on the machine. Some machines still have 2.0.3x kernel configuration is in /conf/linux just wondering. I'm building 2.0.10 now -:- SignOff AlonzoTG: #TUNES (Ping timeout for AlonzoTG[216-164-242-134.s388.tnt10.lnh.md.dialup.rcn.com]) you mean 2.2.10, I guess my laptop is on -ac7 (required for pcmcia serial) yeah, 2.2.10 dammit, kernels shouldn't take an hour to compile when I got my p133 it compiled linux in 5 minutes -:- AlonzoTG [Alonzo@209-122-205-168.s168.tnt4.lnh.md.dialup.rcn.com] has joined #tunes tcn: when I got my 386dx33 w/ 8mb, kernel compilation took hours, too and it was only 0.99.15 or so hehe that's one reason I got my "new" machine a K7 600? 08:50am I used to compile big things on a 486-66 w/ 8MB at school, and download it.. * Fare/#Tunes hopes the K7 will lower the price of Alpha motherboards... heh no, my "new" one is still the p133 did you try overclock it? no tcn: are you bored with my /conf yet? Fare: actually I haven't unzipped it yet :) ok, here we go 09:00am hey, what would VM86 be useful for, if i'm not using the BIOS? Fare: any of those urls any good? hey, my kernel compiled in "only" 20 minutes! tcn: VM86 w/o the BIOS? not much perhaps to use VESA 1.2 drivers oh yeah, gotta do modules and even then VESA 2.0 is 32-bit? it's both 16 and 32 bit i've got a VESA 1.2 card.. hmm, but I just use S3 drivers problem is, 1.2 is only 16-bit, and doesn't do linear framebuffer, afaik oh yeah well, I know it sucks Fare: any of those urls any good? my lynx.cfg is broken; I can't see anything (or is it my apache proxy being broken?) 09:10am I don't need vesa to use a linear fb, do i?? (If I use port access) -:- SignOff hcf: #TUNES (Leaving) no, but then you need the docs of the card; -:- hcf [nef@me-portland-us839.javanet.com] has joined #tunes and you need rewrite the code for every card. right man, PC video cards are fucke dup or course you metaprogram the drivers (see www.irisa.fr/compose) hmm.. domain-specific languages? yup we could do that by extending forth/lisp/etc.. not really a new language sure... see Olin Shivers' paper listed on VMs.html what is metaprogramming? smoke: 09:20am I have some vague notion... Programming is like when you write a program and it runs... smoke: http://tunes.org/~fare/articles/ll99/index.en.html metaprogramming is when you write a program that generates a program that runs... =\ alonzotg: sounds like metaprogramming is high level programming, is that right? fare: thanks not eggzacktlie. Take a C program... Then take a program written, possibly in C, that inputs the C program and modifys it... That is a metaprogram... Its a difficult concept to learn but usefull. it's only difficult with C hmm a meta programming language is a kind of programming language, still? fare: nice article btw, very readable ! :) there is no such thing as a metaprogramming language. There are programming languages with standard ways to metaprogram, however -:- Niobe [chaldea@ppp-042.m4-1.osh.ican.net] has joined #tunes -:- Niobe [chaldea@ppp-042.m4-1.osh.ican.net] has left #tunes [] -:- Beholder [beholder@ppp-042.m4-1.osh.ican.net] has joined #tunes 09:30am Fare: Olin's paper gave you the idea for you scheme meta-etc? ah. and this all has to do with (the, so it seems, important notion of) reflectivity? tcn: no smoke: Reflection is important to new OS's... it allows the OS to modify itself, by being visisb visible tcn: Olin's paper is good. Olin is good. I like Olin. But it's not him who inspired me. hmm. smoke: my paper (same URL) addresses the issue of Reflection, a bit smoke: Read fare's stuff, much better than I could explain it tcn: the scheme stuff isn't working yet (for meta-etc) beholder: i just did :) fare: the article suggest that meta programming is useful in all areas of software development, as long as the software concerned is somewhat complex. Did i understand that right? anyways, my 2.2.10 kernel finished compiling awhile ago.. i'm gonna go boot it.. -:- SignOff tcn: #TUNES (tcn has no reason) suggests, even as long as the software is large enough in space/time 09:40am fare: are metaprogramming and optimizing mutually exclusive concepts? Smoke: Imagine how languages embbeded in games that allow advanced scripting and AI and such can help the programmer improve the overall quality of a game.. Meta programming is the same. You allow programers to solve problems using words that are more natural to the problem. beholder: that makes sense smoke: not at all smoke: The reason it requires a large project is the time it takes to create the meta-language you wish the programmers to use smoke: people at irisa.fr/compose have shown that metaprogramming could be used for quite amazing optimizations smoke: the FFTW uses metaprogramming, too fare: fftw? fastest fourier transform in the west fast fourier transform w ? grin fare: Really? I thought it was done in ASM? is there a base structure out of which one can build meta languages? smoke: There are many projects under tunes creating that base structure aha! so i finally understand what tunes is all about? ;) Or am I confused with another FFT implementation? 09:50am one that automatically selected the right combination of optimizations according to the problem size... -:- smkl [sami@DLI.rdyn.saunalahti.fi] has joined #tunes smoke: More than just meta-programming, also creating an OS that implments these features along with persistence and reflection smoke: currently, there exists a variety of tools that each does a very specific metaprogramming task, and that do not cooperate with one another you have compilers, compiler-compilers, code analyzers, proof-checkers, partial evaluators, etc, etc, but the scientists who develop them do not cooperate and cooperation makes music out of tunes :) there's no conductor ? :) smoke: Nope. Just a bunch of people who do their own thing but still communicate with others :) hmm i suppose it's impossible to build One General 'thing', given the current state of affairs -:- hcf has changed the topic on channel #tunes to: Free Reflective Computing System || metaprogramming with LSD fare: Why is the FAQ a broken link? 10:00am uh? where? smoke: A good example of reflection, using the game example would be something like the Quake 2 engine (I assume you've played or seen the game), which is a very versatile 3D engine. Imagine how versatile it would be if, while it was running, it could rewite it's own routines to be faster, or more efficent fare: Umm... under the root project, http://www.tunes.org/faq.html it's at Tunes-FAQ.html what pointing page? beholder: it'd be possible if the engine were coded as a pluggable library Ok, someone should update the link under: http://www.tunes.org/tunes.html beholder: but then again, the data structures might have to change too. smoke: sure; the cost of rewriting data structures is to be taken into account smoke: But as a pluggable library it's very static... It would be able to rewrite itself to ANY piece of hardware it's running on based on a set of rules, or observations of it's runtime. That way even hardware that hasn't been created yet could be optimized. smoke: And optimized in ways that simple abstraction (DirectX, or OpenGL) couldn't possibly do smoke: 10:10am beholder: is it possible theoretically ? smoke: Yes. Reflection is the ability for programs to "see" themselves, and to modify themselves without re-compiling. Assuming it wasn't written "perfectly" the first time, and know what it needs to to to make it faster, then yes. A program generated by a program generator, is at least as buggy as the program generator. =\ beholder: but then the program could not be compiled at all? ATG: exactly ATG: this means that program generators benefit _a lot_ from the bazaar model of debugging ATG: and that using metaprograms in an open way results in very safe programs whereas metaprograms written in a closed way will be failures okay.. I'll ask for some reeding material on that later... i.e. you increase sensibility of safety wrt bug-finding parameter So, if i want to write a metalanguage for demos (~realtime multimedia), where do I start ? :) smoke: you start with actual demos; smoke: The program is compiled. It recompiles itself. 10:20am smoke: you define a domain-specific language for them that you embed within, say, Scheme -:- _QZ [brand@p0wer.qzx.com] has joined #tunes you write a simple compiler (or a set of macros) for your domain-specific language, and then you pipe the result into an existing optimizing compiler (like Stalin) alternatively, you use partial evaluation techniques (a la Tempo) hmm -:- Beholder_ [beholder@ppp-042.m4-1.osh.ican.net] has joined #tunes is stalin free ? yup -:- SignOff Beholder: #TUNES (Read error to Beholder[ppp-042.m4-1.osh.ican.net]: No route to host) damn... got kicked and the other's main focus is picture analysis a/other/author/ the name doesn't sound all too well -:- Beholder_ is now known as beholder that's the reason why you should boycott Stalin compiler. send the author an email smkl: well perhaps it's a joke of very bad taste? -:- binEng [Anders@dialup47-2-60.swipnet.se] has joined #tunes abi: stalin is an optimizing compiler for scheme at ftp://ftp.nj.nec.com/pub/qobi/stalin.tar.Z stalin is the master of execution hey bin stalin is a very aggressive compiler full house today hello all (apart from abi, that is) abi, are you here? i don't know, tril abi: confused as usual? bineng: wish i knew :) 10:30am abi: Could you explain your existence? i don't know, beholder stop discombobulating me hehehe abi are you here yes tril: are /you/ here? muahaha * beholder/#tunes wonders if any of us is truly here.... abi: watch it I have your admin password tril: i'm quite aware of that abi: Write Tunes... Now! beholder: you first somebody put too much time in that bot :) * binEng/#tunes wonders who tells abi what to say abi: But I don't wanna wite Tunes! Damnit, you're the infobot... that makes you smarter right?! smoke: not really, someone here is a ventriloquist tril: phew Tril: icuc, http://www.dur.ac.uk/~dcs3py/pages/work/Documents/index.html * AlonzoTG/#tunes just ran a sourcery program on my soundcard driver... It's shit. =P 10:40am * beholder/#tunes will brb -:- iepos [iepos@d35.k1-1.tecinfo.com] has joined #TUNES hi iepos iepos! oh, hi * hcf/#tunes is away. 11:00am -:- SignOff Fare: #TUNES (the system is going down! Grr!) * beholder/#tunes is back hi hey Tril... how's your type-system thing going now ... ? -:- Fare [rideaufr@esmeralda.enst.fr] has joined #Tunes hello again, Fare... oh pretty good. I have a big plan for what I am going to do next heh heh.... so what is the big plan? Big plans? I've located the source of a lot of my confusion. A lot of things make sense if I clearly distinguish the typesystem (the static system) from the persistent store (the dynamic system), and examine how they interact. The persistent store contains cells, and other structures that have cells in them. Cells contain references to objects over in the typesystem. Cells can be read or written (so they're variables). But object IDs calculated by the typesystem stay the same, and are only changed by modifying metaobjects that are in the persistent store. -:- SignOff iepos: #TUNES (Ping timeout for iepos[d35.k1-1.tecinfo.com]) -:- iepos [iepos@d3.t1-6.tecinfo.com] has joined #TUNES aack... power outage. heh iepos: external modem w/ a UPS for everything else? Sunny day here... damn hot... wish we had rain and some outages ;) tril, afraid not... but the power came right back on oh nevermind i kee pforgetting that ping timeouts are delayed from the actual outage thought you showed up pretty quick :) * beholder/#tunes wonders how it could reach 36 deg C around here... 11:20am could it? It is... does abi do temperature conversions? It feels like Texas... and I'm in Ontario... Which is very unusual I don't think so, you don't know Celcius system? Tril: she would u add the routine 20C is room temp., 100C is boiling point of water 0C is freezing, that's about all i know :) 36 is body temp It's 96.7 deg Outside then ok Farenhite Very hot for around here yes, i imagine I watch the show "Due South", they make Canada seem like it's one big glacier Hehehe :) They're up in the Yukon territory in that show I think -:- binEng_ [Anders@dialup42-1-5.swipnet.se] has joined #tunes right. the ones that are in Canada. Most of them are in Chicago, but occasionally the mountie goes home for an episode We have the same temperature variance as Michigan, if you've ever been there -:- SignOff binEng: #TUNES (Ping timeout for binEng[dialup47-2-60.swipnet.se]) -:- binEng_ is now known as binEng ... even New York is similar 11:30am * Tril/#TUNES reads ESR's new paper "The Magic Cauldron" Tril: Where is that paper? Uh ... will anyone please explain why the Sith Lord and Trade Federation risk everything to capture a teeny periphery planet? Can we have a clue why Naboo was important -- any hint at all? Hello? oops http://www.tuxedo.org/~esr/writings/magic-cauldron/ previous contents of clipboard :) (David Brin on the phantom menace) Heheh I was wondering that myself who's David Brin? i dont know, but he wrote some good articles on salon.com that were linked on /. afaik david brin iz a scifi author hmm... "We present eight models for ... funding ...; two for-profit, seven for-profit." ? i mean non-profit on the first i guess one of the models must be both :-) 11:40am who's David Brin? who's David Brin? -:- Tril has changed the topic on channel #tunes to: Free Reflective Computing System || metaprogramming with LSD http://www.clas.ufl.edu/users/seeker1/noetics/lilly-metaprogram.html ||http://www.tuxedo.org/~esr/writings/magic-cauldron/ -:- hcf has changed the topic on channel #tunes to: Free Reflective Computing System || metaprogramming with LSD http://www.clas.ufl.edu/users/seeker1/noetics/lilly-metaprogram.html || esr's The Magic Cauldron -:- SignOff AlonzoTG: #TUNES (Have Nice Day :)) -:- hcf has changed the topic on channel #tunes to: Free Reflective Computing System || metaprogramming with LSD || esr's The Magic Cauldron can irc servers be crashed by buffer overrun in the size of channel topic? (go try in some other channel if you do) /topic iz likely governed by the same code as the rest of irc cmds, thus overrun proof'd, by nowadays at least -:- SignOff beholder: #TUNES (Ping timeout for beholder[ppp-042.m4-1.osh.ican.net]) no, i mean it probably has a strict limit on length thats what i mean also, limit'd length like the rest like ~450 or so 11:50am -:- SignOff binEng: #TUNES ( ) -:- Beholder [beholder@ppp-001.m4-1.osh.ican.net] has joined #tunes Tril: have u written any papers? not really every time i think i have an idea for a paper, it ends up being a paragraph heh i have lots of disconnected paragraphs 12:00pm so that's why i'm focusing on programming or experimental design -:- AlonzoTG [Alonzo@209-122-216-26.s26.tnt3.lnh.md.dialup.rcn.com] has joined #tunes any zope news? i subscribed to the zope mailing list for a few days, got flooded >200 msgs didnt read any of them, unsubscribed. i'm going to wait until i learn more about zope to join the list I found another way to do the database that doesnt require SQL, i.e. uses zope internal database, but i think the SQL way is better because we could access the same SQL db from other apps rather than relying on zope. which is the same thing as saying , "without relying on Python" Tril: Why not just use that simple script binEng made i haven't seen it yet! something is running on tpp.dhs.org? i would assume perl has zope mod Yep, it's just a script that allows you to update pages without requireing shell access hcf, maybe . it would be called a Bobo database module, though I don't think the zope object system is accessible in perl (Aqueduct, I think it's called) although it's very easy to access if you know python Tril: you can see it like this: http://tpp.dhs.org and click on the "edit" link at the bottom Tril: I think it would work well for Tunes members maybe for regular site maintenance if it was changed to use CVS but that's not our main problem i want references to database entries to be embedded in the html 12:10pm it's pretrty neat that you were able to come up with that, though * Fare/#Tunes doesn't think HTML is such a good source language, anyway my typesystem design is superior to a relational database, anyway Tril: hav you an updated document on it? not yet