IRC log started Sat Apr 29 00:00:00 2000 [msg(TUNES)] permlog 2000.0429 -:- SignOff coreyr: #TUNES (coreyr has no reason) -:- SignOff eihrul: #TUNES (Ping timeout for eihrul[usr5-ppp167.lvdi.net]) -:- smoke [smoke@15dyn140.delft.casema.net] has joined #tunes -:- SignOff hcf: #TUNES (Leaving) -:- hcf [nef@me-portland-us814.javanet.com] has joined #tunes -:- abi [nef@bespin.dhs.org] has joined #tunes -:- SignOff abi: #TUNES (dying by hcf's request) -:- abi [nef@bespin.dhs.org] has joined #tunes -:- SignOff abi: #TUNES (dying by hcf's request) -:- abi [nef@bespin.dhs.org] has joined #tunes -:- SignOff abi: #TUNES (dying by hcf's request) -:- abi [nef@bespin.dhs.org] has joined #tunes -:- SignOff hcf: #TUNES (Ping timeout for hcf[me-portland-us814.javanet.com]) -:- hcf [nef@me-portland-us836.javanet.com] has joined #tunes -:- XeF4 [xef4@cube2.babel.dk] has joined #tunes -:- SignOff hcf: #TUNES (Leaving) -:- core [core@core.suntech.fr] has joined #tunes anybody awake? yes hey xef4 :) how's your work going, still going to make it in time? (yes, i remember you :) core: I think I could make it in a few more days if needed XeF4: are you supposed to code something functional, or just come up with specs? core: using (Rt)Linux for hardware support until I get time to do it right core: I'm supposed to code something functional rtlinux blows bigtime :) XeF4: ah ok.. big task :) using linux drivers is nice tho. core: yes, but it gives me low-level hardware support and a flat memory arena which is enough for a rough prototype well, you could use plain linux i mean :) 07:20am I could.. i just have some griefs against *rt*linux, but well, for your purposes, it works :) if I replace the timer interrupt I have more than a few griefs against Linux yeah, well, i was trying to play it nice :) but it's better documented than the BSDs; other free OSes aren't mature enough to host development under themselves well, for what you're doing, linux is right, that's true. *bsd is a little elitist in approach :) hi core, hi xef4 core: I suspect the BSDs do that to keep incompetent coders from contributing code s/keep/prevent/ hey smoke! smoook! XeF4: it's probably true. the *bsd kernels are technically very superior though. (subjective point of view, and mine only, as a system programmer :) core: what language is clementine written in by the way? i mean, whenever i was bumping on something to implement for clem, that also existed in other OSes, i took a look at linux and *bsd solutions for it, and usually found that linux was using something convoluted and bsd something simple and efficient :) smoke: C, with some assembler sprinkled where relevant :) core: ah. i expected you to be under the influence of Tunes language nitpicking :) smoke: i'm not. i'd rather code in an existing language than a nonexisting one. smoke: i know eihrul fell into this, and fare has been promising a compiler for years (hi, fare). smoke: water knows what he's talking about though, he'll come up with something really neat, i suspect. hm hey bill gates is no longer the richest man in the world larry ellison of oracle beat him by $1.25 billion that's fascinating :) :) it's really cheap to compete for the being the richest 07:30am how about the smartest or the one who does the most for the world :) it's all virtual money though.. assets in their companies &c core: hehe that'd be Joe Bloe from Sussex, england Blow even well, it's virtual money, but there's some hard cash, don't worry. it's annoying to lose $20 billion in stock, but when you sit on $500 million of actual cash, you don't really care. you can live 75 or 80 lives. 07:40am -:- dirt [wrong@niantic0315.mohawk.net] has joined #tunes back.. doing anything nice? core: not at the moment, FBSD disk just died XeF4: 4.0-RELEASE? 3.4 core: I mean.. the physical drive died XeF4: ah.. wee :) XeF4: hope you didn't have anything vital on it nothing too vital I just don't have another spare hdd handy I swapped interface cables and the drive died 08:30am -:- Ghyll [karltk@msx-osl-12-41.ppp.cybercity.no] has joined #tunes -:- NetSplit: heinlein.openprojects.net split from hogan.openprojects.net [08:59am] -:- BitchX+Deb1an: Press Ctrl-F to see who left Ctrl-E to change to [heinlein.openprojects.net] -:- Netjoined: heinlein.openprojects.net hogan.openprojects.net -:- core [core@core.suntech.fr] has joined #tunes -:- ult [ult@user-38lc63b.dialup.mindspring.com] has joined #Tunes -:- SignOff Fare: #TUNES (:Connection reset by pear) -:- Fare [rideaufr@esmeralda.enst.fr] has joined #Tunes Fare? yes? heh ah, life. encore la, toi? mais bien sur. kwa29? tu as bien dormi? oui, jusqu'a 8H30 :) s/bien // je suis allé chercher mes billets pour les USA, j'ai été voir un pote, et là j'écris du code. samedi classique. -:- SignOff Ghyll: #TUNES (Ping timeout for Ghyll[msx-osl-12-41.ppp.cybercity.no]) brb -:- Ghyll [karltk@msx-osl-15-46.ppp.cybercity.no] has joined #tunes 09:20am -:- water [water@tnt-10-249.tscnet.net] has joined #tunes hey all hey water :) core! how go things with you? * water/#tunes reads the downix interview water: very good i guess :) clementine (implemented in lowly C, but working :) is good enough to use (unmodified) ircII-EPIC on, now.. and personal life is doing well as well. how are _you_? having fun in the navy still? :) well, i'm still working up slate, and eihrul's working on a fully-dynamic compiler and there's Someone in my life now :) his weekly project? :) i'm definitely looking forward to making bindings for clementine :) (for Slate that is) there is? congrats :-) navy person? ;) right, well he's dropped everything else to work on this compielr afaik, and he seems to be getting along fine navy? hell, no! college girl just kidding ;) nice! smart? :) oh yes, and although she's not math-sci-prog oriented, she understands what i'm doing, from tunes to slate to arrow to <> she's probably new-age. (and it doesn't hurt that she spends just about every night at my place now ;) wow.. well, either you're good at explaining it to laypersons, or she's smart :) well irc sucks for explaining things core: I vote the latter >:-} what does she study in college? (yeah, the night thing is nice, huh? :) no, she's not new-age :P water: yeah, it does. well, it works for people who do not usually socialize i guess; but for the normal rest of the world it's a bit of a limited medium :) poli-sci, international studies she's most interested in russia, maybe working there on business eventually and as a translator 09:30am she speaks russian ? yep (obviously) russia is a very interesting country. i was there last month * Ghyll/#tunes knows two russians.. (that's all I had to add to this part of the conversation :) they have enormous potential. the day they put back an actual infrastructure, they will dust everyone yeah, maybe i should be pitching arrow to *them* :) core: yup. problem is that somebody screwed up royally when advising the russians how to move to capitalism... they can live in closed circuit. they have all natural resources, natural gas, petrol, fruits, whatnot, and very smart people to handle them :) Ghyll: no, actually, the problem is they did abruptly. they moved from one infrastructure (communism) to none at all. right now they're putting the pieces back together. core: if they don't become to greedy and dispose of waste in siberia water: i think she picked a country that will definitely evolve politically and economically in the near future :) yeah XeF4: no, that's what americans and western europeans already do :) or make other moves for short-term gain water: is she American (your S.O.)? :) yeah, she's from alaska water: ah ok. nice. it's not THAT common to see US-native people who have genuine interest in the rest of the world :) * Ghyll/#tunes chuckles quietly yes, too often true if she goes to Moscow, tag along. it's an insane city i loved it :) hm and no, i don't have a "CCCP" thing tagged all over my passport ;) well anyway, i have a preliminary slate tutorial up now, as well as some functional common lisp code by eihrul on the website heh at ~water ? water: just curious.. where in Alaska? yeah, although the tutorial is only accessible from *news near fairbanks she's been to siberia twice already siberia is quite barren.. i haven't been there tho. has she been to Moscow or St Petersburg? not that i know water: check http://www.tunes.org/~water/slate-tutorial.html of course i've only known her for a wekk now a week, even damn it! water: it's a bit short :)) ok i'll re-upload and not pre-process it heh water: nice. i've passed the two year milestone with jenny two months ago :) ok it's fixed cool water: you should do something about the notation. how so? water: when naming variables in the paragraphs, you should use varname in a year you can pinpoint when you slowed down on slate.. last week :) 09:40am oh yeah * core/#tunes goes read the tutorial water: the sentence "result doesn't describe the result of 'x'" is a bit confusing well don't the italics help? yes, but not having the variables in fixed width is a bit irregular. yeah i suppose so this is just nitpicking, but since this is a tutorial, you should maximise on established and tried conventions. keep in mind i tend to avoid web-coding wherever possible * water/#tunes nods I understand that. I don't like web-coding either, but I like to have things 110% proper (especially when i'm not the guy doing the grunt-work :) there are a whole lot of language aspects i have yet to explain heh sure yes. I see the tutorial is far from finished. But what's done until now is readable. Best language tutorial I've read to date has been the python tutorial. hell, i'm still trying to figure out the Right Way for doing all sorts of things in Slate (particularly CL features) hm ok i'll take a good long look at it It might be that python is just such an easy language, but I found the tutorial very good and clean as well. slate has some really odd concepts though yes, but that should keep the reader interested and on his toes. Python is BASIC for UNIX =P at least it did for me so far. ult: yup. and so much superior to basic. * Ghyll/#tunes 's language conventions are obviously in flux.. hm well since i just got back from work, i'd like to go shower and such is anyone leaving soon? no. ok, in that case, bbi15min I should get out of my pyjamas and take a shower as well.. :) 09:50am -:- SignOff core: #TUNES (Ping timeout for core[core.suntech.fr]) -:- akawaka [akawaka@136.201.103.36] has joined #tunes -:- akawaka is now known as aka|shop hm? core's gone? hi aka rats. I just got visitors. ok, well i have to go make a phone call * Fare/#Tunes is back hey fare yeah water i'm still working out stuff on slate, particularly to see how CL stuff translates best into slate code -:- SignOff ult: #TUNES (Ping timeout for ult[user-38lc63b.dialup.mindspring.com]) how much of slate is there, yet? what metalanguage are you using? C? well, eihrul wrote a basic evaluator in CL is it hackable? 10:20am yeah it's on his tunes account still very rudimentary, though he's more interested in writing a full open compiler for it and we're sticking with dynamic compilation the final product is supposed to be meta-object oriented, that is, wrapping a core code-gen with dynamically-compiled front end to handle various semantics -:- SignOff XeF4: #TUNES (BitchX-75p1 -- just do it.) 10:30am hm muh i'm trying to figure out where to go with the slate docs, as there is much to add, but much refitting needed to do so * Fare/#Tunes scp's slate home -:- eihrul [lee@usr5-ppp101.lvdi.net] has joined #tunes eih: speaking about you and i'm still working out how to do macros slate-style in slate as opposed to lisp-style, which is possible but now desirable s/now/not/ hey eih what's slate-style for macros? not sure exactly yet Fare: nothing good, i hope? probably manipulations of object trees with heavy use of mo's hum, how is lisp style not that already? oh you mean, replace cons cells with structures? 10:50am well lisp's macros work on objects that are unlike slate objects yeah where s/structures/slate objects/ i'm also looking into slate-style monads, which relates hum, as long as there is a quasiquotation technique, it should be ok i.e. using objects as environments yeah quoting slate objects seems to elude me no one needs source-as-lists when one has quotation and quasiquotation. (AND pattern-matching based on quasi-quotation) although i *could* use the functional representation as the quoted object right * water/#tunes has been soaking up "On Lisp" lately I didn't read that one. Does it talk about pattern-matching? yep should get it, then * Fare/#Tunes wonders what e-book vendor to pick in europe (not amazon, obviously) eih: reading your slate compiler comments from thursday hm gradually Kyle will recognize the clue stick after enough whackings ;) what's Kyle? kyle lyhanovski or whatever his last name is what's he doing? he's the "DBOS=Tunes" guy water: well, they're more rants than comments :) Paul Graham? yeah i noticed after a minute or two Fare: haven't you been reading the mlist? reading through it kyle was posting quite a bit for a week or two Lahnakoski yeah there you go 11:00am -:- aka|shop is now known as aka|icecream -:- SignOff dirt: #TUNES (L8er) hm he says he wants to talk to me, but i've never gotten a decent conversation out of him (dirt, i mean) -:- aka|icecream is now known as aka|usbitch lahnakoski means bream stream huh? "bream"? -:- aka|usbitch is now known as akawaka a fish hm -:- akawaka [akawaka@136.201.103.36] has left #tunes [] * eihrul/#tunes wonders what akawaka was here for. he was lurking i sort of convinced him he could do it without actually being here (gave him the log url) so how goes the compiler stuff? well, trying to integrate the control-flow graph more closely into the data-flow graph 11:10am control-flow graph is too much of a pain for scheduling -:- SignOff Ghyll: #TUNES (Read error to Ghyll[msx-osl-15-46.ppp.cybercity.no]: EOF from client) hm and it also makes motion of instructions within the data-flow graph a bit complex but control-flow always has to be represented in some manner or another because of side-effects :/ hm not if you keep the basis purely-functional of course, i reify environments as enclosing namespaces, so there you go :/ well, at some point you introduce side-effects (register level) yeah hm ever read up on data-flow architectures? monads? -:- iStormy [stormy@rain.futuresouth.com] has joined #tunes yes, fare, i'm trying to suggest it to him hi storm Fare: although he's working on back-end stuff right now water: well, my intermediate language is pretty much a data-flow graph with the control-flow graph only so that dependencies can be maintained http://www.cs.cornell.edu/talc/icfp99-contest/solution.htm water: but, if you have any more information on such :) smkl: hmm water: ? smkl: i heard about that but didn't look into it well anyway 11:20am * water/#tunes looks for good data-flow url's -:- core [core@core.suntech.fr] has joined #tunes re core re.. sorry, was away :) hm the squeak mlist went down no new toys to play with ;) 11:30am ping pong 11:40am hm damn it, even the language list isn't up to the task well, eih, the only thing i can suggest is to allow hof's in the il and use monads otoh i do recall a vm project that was designed for fp langs 11:50am C-- ? heh no, it was something else that was better in design then I haven't heard about it! eih: actually, i think some papers on it can be found at RI gimme heh water: monads are just another way of specifying a control flow graph true, but you can get them out of regular functions, no? darn it, my searches aren't finding it -:- SignOff smoke: #TUNES (Ping timeout for smoke[15dyn140.delft.casema.net]) * water/#tunes tries cora x machine? g machine? something weird like that? :) i think it was the g machine or something like that g machine was in pj lester yeah hm bleh, you're probably better off looking yourself :P g machine? that's an old old thing not a bad thing invented for graph reduction yeah well i believe i'm thinking of something else, but i don't even remember the name now at any rate, you still need side-effects, as slate allows one to always reify the enclosing namespace/environment as an object 12:00pm yep if i didn't need side-effects, there'd be no problems :) heh of course then again, if i didn't need side-effects, i wouldn't be able to do much useful work but you have to admit that monads are the closest things to describing how side-effects are possible in slate -:- smoke [smoke@16dyn74.delft.casema.net] has joined #tunes 12:10pm Fare: as long as we're here, do you have any tunes issues to discuss? -:- NetSplit: forward.openprojects.net split from sterling.openprojects.net [12:11pm] -:- BitchX+Deb1an: Press Ctrl-F to see who left Ctrl-E to change to [forward.openprojects.net] -:- Netjoined: forward.openprojects.net sterling.openprojects.net -:- thomas [thomas@193.217.63.152] has joined #tunes -:- smkl [sami@glubimox.yok.utu.fi] has joined #tunes -:- corey [coreyr@net255ip95.parklink.com] has joined #tunes -:- abi [nef@bespin.dhs.org] has joined #tunes -:- eihrul [lee@usr5-ppp101.lvdi.net] has joined #tunes -:- iStormy [stormy@rain.futuresouth.com] has joined #tunes doo de doo hey Fare doh heh would you care to discuss a few Tunes issues? yes? no? all of the above? perhaps a glossary entry or 7? ;) 12:20pm * water/#tunes wonders why he even tries. well, it's pretty simple in slate to handle mutator meta-behavior wrt cloning enclosing namespaces you pretty much just rewrite the mutator to have a ".. clone " prefix hm which could be encapsulated in a single operator, methinks this of course also works in the parent direction i'm still working out how/when/where clone semantics will get over-ridden 12:30pm hey! is someone interested in talking about #tunes or what? 12:40pm * water/#tunes reads some papers to soak up ideas hm an extension of java to handle inner classes 12:50pm still lame :) anyone here? hmm, didn't java 1.1 include inner classes? hell if i know (heck i just deleted the papers i dl'd on it) they didn't even have any insights that the beta community hadn't already produced java inner classes do not build the closure automatically i think hm i can't really picture that what do you mean? 01:00pm well, you have like "{ int i = 0; new class { int ret_i() { return i} } }" ... that won't work uhhh sure wait a sec that *doesn't* work? i don't remember the syntax heh even more lame than i thought yeah ... you need to have the i as instance variable in that class otoh smalltalk isn't much better in that respect beta and slate, however.... >:) * water/#tunes looks at a discussion of mult-inheritance vs single-inheritance of course i'm also looking at ocaml (for any French reflection-oriented coders who might be listening ;) too bad ocaml isn't very reflective :/ that's ok, i think i can use its features in slate perhaps even generalize them a bit 01:10pm you can have inner classes in ocaml though, using "let module" * water/#tunes shudders sorry, that syntax really bugs me hm maude can actually do much the same thing and maude IS reflective :) hm also mixins mixins are pretty simple in slate, so there's no loss of power from, say, agora, to slate 01:20pm * Fare/#Tunes is back hey, got Tunes issues or terms to discuss, Fare? i for one would like to hear ideas about quotienting, if you have any where do I begin coding from quotienting? let's see. yes, I do good because the tunes docs don't say enough about it quotienting is about defining rewrites rules that the system _may_ apply, at its discretion ok what does quotienting offer beyond, say, Maude? abstraction power. i don't follow a quotient is more abstract than the quotiented object define quotient, then quotienting allows to communicate concepts that cannot be formed without * water/#tunes whacks Fare yes, i know this -:- NetSplit: devlin.openprojects.net split from irc.linux.com [01:24pm] -:- BitchX+Deb1an: Press Ctrl-F to see who left Ctrl-E to change to [devlin.openprojects.net] but what do you formally mean by it? equivalence classes? -:- Netjoined: devlin.openprojects.net irc.linux.com -:- smkl [sami@glubimox.yok.utu.fi] has joined #tunes -:- thomas [thomas@193.217.63.152] has joined #tunes -:- eihrul [lee@usr5-ppp101.lvdi.net] has joined #tunes given a set E and an equivalence relation R, the quotient set E/R is such that there is a canonical injection j from E to E/R, and two elements have same images iff they are equivalent ok equivalence classes, then how does this work in a programmatic way that fits into a real programming language? equivalence classes are one usual way to construct quotient structures but the important point is the structure grrrr equivalence classes are not constructions they fit into metaprogramming, not mere programming if all you want to do is program, then you needn't quotients * water/#tunes sighs but if you need to apply metaprograms, the fact that you defined something as a quotient quit speaking phd-language allows the metaprogram to apply transformations that would otherwise be forbidden i don't want a dissertation, i want you to explain how it would work for instance, if you define you program as "anything that implements such abstract program with such attached effects", you allow the system to do optimizations but if you define your program as "execute precisely these instructions in given order", then you have no optimization well how do you implement a system that can recognize such things, in an effective way? not in one day you're so god damned abstract, no wonder you have no phd yet first, you build the framework, then you slowly develop metaprograms, metametaprograms, etc bullshit just like the Tristan Cazenave's Go player metaprograms cannot help to encapsulate quotienting i think you have some concept fusion going on they help _use_ it 01:30pm they require that you have some support from it in the encoding framemork. well you have convinced no one on the tunes list of this, so i don't see how you understand it again bullshit "some support" = "magic wand" but since quotienting is about allowing additional rules to be executed, it requires a metaprogram to control a quotiented program for isntance, if you build rational numbers as a quotient, ok now an example ! :) then it requires a metaprogram to manage what representent you use why? you may normalize everytime, but it might not be good is this a ui-idea or a logical requirement? and if you don't normalize often enough, it's bad *AHEM* why? it requires expertize to normalize at the right moment * water/#tunes sighs water: ever written algorithms that manipulate rational numbers? yes well, then you know that normalizing always induces a constant slowdown factor and i know what you mean, but at the same time your argument tells me nothing so? worse even: many structures have no normal form yes i know this, but what ideas do you have *of your own*? so you can't just "let the system decide". If you want to fully manipulate the objects, you have to control the metaprogram yes i know... i'm dealing with this in slate