IRC log started Sat Feb 19 00:00:01 2000 [msg(TUNES)] permlog 2000.0219 -:- SignOff hcf: #TUNES (Leaving) -:- hcf [nef@me-portland-us1027.javanet.com] has joined #tunes -:- SignOff hcf: #TUNES (Leaving) -:- water [water@tnt-10-44.tscnet.net] has joined #tunes 'lo 01:40am lo hey i stumbled on a new book on gc it covers everything, even a chapter on self stuff :) also distributed and concurrent gc at least every gc strategy i've ever heard of what's the name? and to top it all off, such a wonderful book was gracing the decrepit shelves of b&n :) garbage collection: algorithms for automatic dynamic memory management very uninspired but a title's just a name :) what's the copyright date? 96 hm though it's a reprint ah, i believe that's the only book on the subject :) or one of the few that i've heard of already... (which happens to be a total of 1) heh 01:50am you know there's a book on advanced oo reflection with an entire chapter by jecel and another by the self group right? i stumbled on it when my ship passed through hawaii wow, he has his own chapter? :) but it didn't tell me anything i didn't already know yeah i was so shocked i showed one of the guys i worked with "hey i correspond with this guy on email" of course it didn't phase the guy but oh well i believe we were discussing math&text&logic (i sort of lump them together because of the way i think) anyway, i thought about this and realized i should point out to you the difference between inductive and co-inductive to help explain why these things are an issue with slate unless you've already learned this stuff? (bleh. the net sucks. the url on the book is un-maintained) so do you want to hear my blather or not? :) learned of it vaguely hm ok er, co-induction :) well it's best to start with the induction principle i heard it best put by the statement: "the inductive principle says precisely that the natural numbers are inductive" anything inductive is isomorphic to the natural numbers in a certain way abi: part #tunes grr well anyway, the integers are *not* inductive 02:00am the point of inductive systems is that they rely specifically on a finite set of atoms for the naturals, it's zero the integers though, happen to be co-inductive from a certain viewpoint you could say that the integers consist of all of the inductively defined counting systems, so that even if you had a "least" kernel of the induction, you can always find one less similarly, the rationals and reals are juxtaposed negative infinity? :) bah :) seriously, the integers can't be isomorphically-mapped onto the naturals but anyway the rationals happen to be inductive from a new kind of viewpoint quotienting over the successor relation *and* multiplication the reals, however, happen to be co-inductive over the same ideas it turns out that there's a general framework for looking at inductivity or co-inductivity within a framework in general heh... two times generalized ;) this is where you get those great mathematical high-falutin terms "greatest" and "least" fixed-points both of these deal with how to look at an ambiguous specification equation the "least fixed point" approach says this: take a value. if it works (verifiably with a proof and positively), add it to the solution set. then use the primitive ideas of the equation to reach the other elements of the solution set 02:10am this is done statically, and doesn't encompass all the ideas one might have in real-world situations involving such a spec actually, there are many fixed points to an equation like this, but there is a theorem of Tarski's which says that there can be only one least fixed-point of any equation which corresponds to induction? the least fixed point there's a simple example relating to "dividing by 0" the inductive view says that no solution positively solves it, so the solution set is empty the co-inductive view says that no solution doesn't negatively refute the equation, so all real numbers solve it hm... maybe too many negatives there :) define 'refute' :) hm i think i stated it poorly i.e. a solution can be provably verified to not apply? right co-induction accepts all answers that can't be refuted by formal proof so one is pessimistic... one is optimistic, in a sense :) yep the results for reasoning are tremendous though a lot of things that make noise in inductive definitions turn out to not belong there at all like state :) but, as icct shows, computers (inductive systems) must cope with co-inductive ideas to effectively manage general-purpose information 02:20am hm got any comments? did i make anything clearer than it was before for you? well, you never touched on greatest fixed-point ok i can explain if you feel necessary s/feel/feel it/ well, it relates to co-induction, no? :) of course i'm simply talking of dual ideas just about anything in cs with co- as a prefix is a dual co-inductive concept for its basic inductive concept do you understand why i would base arrow on a group of co-inductive ideas implemented in slate? i'd rather assume not, so that i make no wrong assumption :) heh in that co-inductivity for arrow means an absence of primitives? yes, that's one aspect of it but then how do you define a system co-inductively? but it's also about handling information ah excellent question this is an interesting aspect of the results of co-inductive ideas within inductive systems, you must construct an object before referring to it to ensure that your definition is inductive 02:30am if you deal with anything but the least fixed point, you don't have to stick with this constructive approach you don't even have to force co-induction in order to abandon constructive languages but, taking the co-inductive viewpoint.... you can describe streams, for instance, with a set-theoretic equation Z=AxZ , taking Z as the greatest fixed-point solution => Z = the set of all streams holding type A you can't define streams without co-induction :) grok? grokking... so... inductively, Z=AxZ would be interpreted in reverse? :) (to think that i wrote the whole arrow draft without knowing of co-induction) huh ? what are you asking? i'm saying Z would be empty, inductively you'd get an exception or something yes... 02:40am so if induction were construction, then co-induction is destruction? heh well, inasmuch as you can destroy the universe by separating out things you recognise as individual, yes destruction => destructuring into its components sure or atleast in the way i used it there but not just the trivial case of destructuring things that were inductively-constructed that's too easy :) so where does co-induction fit into slate :) hm reflection? rumour has it reflection is a property of a system that can refer to itself and manipulate its state or rough on your brain heh well i'm not precisely sure yet... it may include rewrite otoh i might take the lll approach which is the most general ah, so rewrite => co-inductive? it can be, yes it intuitively seems so... well terms are just labellings and labellings can be modeled easily as functions (as graphs of arrow :) 02:50am hmm.... so where does that leave the discussion... not sure basically i would just like the primitives to be invoked in a way that they can be arbitrarily moved around later but i need to think on that a bit 03:00am i'd like to think that we could have a "bottom up" defined system covering math etc and replace it later with a "top down" system aka bootstrapping moving the primitives approapriately without changing end-user syntax well, a different kind of bootstrapping than has been done before hmm, sleep ok -:- SignOff eihrul: #TUNES (Leaving) 03:10am -:- Kaufmann [newbie@dial188.infolink.com.br] has joined #tunes Heh. Bow at the sight of my godlike Karma! >>> Kaufmann [newbie@dial188.infolink.com.br] requested PING 950970687 from #tunes what is it now? 05:30am 57 (c'mon, you can say "ooooh") :) ouchie yeah, baby 05:40am Oh man. I've slept for 16 hours. 05:50am -:- SignOff Kaufmann: #TUNES (kathyanne gets kick'd/ as leaves fall on still waters/ she is here no more.) -:- AlonzoTG [Alonzo@216-164-134-109.s363.tnt3.lnhva.md.dialup.rcn.com] has joined #tunes -:- SignOff AlonzoTG: #TUNES (Have Nice Day :)) -:- HooHaa [joe@jesus.wnyip.net] has joined #tunes Hello interesting alternative os people hi hoo Hi,Im glad someone is home fare: have you taken backup of vital areas on bespin lately? Your page looked reaaly interesting :) What will tunes do that plan9 doesn't? the hll features perhaps? umm.. isn't plan9 a unix-like os? ie it is not a reflective system plan9 claims to be a distributed os with no superuser and I believe an approach to dll type issues like yours 07:20am your right I don't think it is tunes is different! I think your ideas sound very good. I was hoping to find more code on your site I suppose it does make a whole lotta sense to design everything in depth before coding though Thats gotta be neccessary for such a new OS designing too much is not good though Have you thoughtof using 'programming design language' as a design stage? 'pdl' is a formal english like pseudo code to check designs Your ideas are so new you probably need a whole new design paradigm Id bet like an extreme top down type design I doubt any pdl's can capture reflection yea,I suppose so I think the ultimate for your hll language would be to write hll on a preexisting platform and then program hll to reproduce itself in the most effiecnt way possible-the oupt would be a binary tunes outp=output do you know lisp? lisp is ((simply) (perfect)) no just c,assembler and visual basic Ill have to check lisp out I remember its one of Eric Raymonds 4 recomended languages it might be a good idea to check out common lisp then to get a feel of a reflective system thanks check the language review pages on tunes.org for a few pointers 07:30am so the idea is to write hll on lisp and then have hll produce tunes? lisp is one of several alternatives, but it might give you an example on a language which is OO from the core and allows reflection and manipulation of itself while running oh,ok,thanks lisp is a programmin language you write new programming languages in, tailoured for your problem.. Ill check it out, see you guys next time -:- SignOff HooHaa: #TUNES ([x]chat) 07:40am -:- rares [rares@nwhn-sh3-port59.snet.net] has joined #tunes hey 08:00am -:- rares [rares@nwhn-sh3-port59.snet.net] has left #tunes [Segmentation] -:- iok [iok@user24.anent.com] has joined #TUNES * iok/#TUNES is away ([10-min] $WiRC$ auto-away) -:- iok [iok@user24.anent.com] has left #TUNES [Suffering] -:- hcf [nef@me-portland-us635.javanet.com] has joined #tunes -:- SignOff hcf: #TUNES (Read error to hcf[me-portland-us635.javanet.com]: EOF from client) -:- hcf [nef@me-portland-us635.javanet.com] has joined #tunes -:- ult [noone@user-38lc684.dialup.mindspring.com] has joined #Tunes any of you have a decent REPL for CL I could use for a named-pipe hack? -:- SignOff ult: #TUNES (Ping timeout for ult[user-38lc684.dialup.mindspring.com]) 09:00am -:- SignOff abi: #TUNES (dying by hcf's request) -:- abi [nef@bespin.dhs.org] has joined #tunes -:- rares [rares@nwhn-sh3-port59.snet.net] has joined #tunes -:- rares [rares@nwhn-sh3-port59.snet.net] has left #tunes [Segmentation] -:- SignOff water: #TUNES (Ping timeout for water[tnt-10-44.tscnet.net]) -:- SignOff abi: #TUNES (dying by hcf's request) -:- abi [nef@bespin.dhs.org] has joined #tunes -:- water [water@tnt-9-215.tscnet.net] has joined #tunes -:- subose420 [subose420@lsanca1-ar2-059-034.biz.dsl.gtei.net] has joined #tunes -:- subose420 [subose420@lsanca1-ar2-059-034.biz.dsl.gtei.net] has left #tunes [] water: hey hcf: barley, even i just sent u mail concerning abi k i think i'm going to get a squeak mlist flood... i didn't get msg's from them for 20 hours after the usual level of traffic thx gakuk hey fare i got a recent book on gc the other day it has at least every algorithm / strategy i've ever heard of fare: have you run any backups on bespin lately? 10:30am heh. it seems he's gone again 10:40am -:- water_ [water@tnt-10-41.tscnet.net] has joined #tunes grr -:- SignOff water: #TUNES (Ping timeout for water[tnt-9-215.tscnet.net]) -:- water_ is now known as water -:- SignOff hcf: #TUNES (Ping timeout for hcf[me-portland-us635.javanet.com]) damn must reboot bbiaf -:- water [water@tnt-10-41.tscnet.net] has left #tunes [] -:- water [water@tnt-10-63.tscnet.net] has joined #tunes 10:50am -:- lar1 [larman@1Cust99.tnt31.sfo3.da.uu.net] has joined #tunes If I make a "shortcut" in win32 will that works as a sym link in *nix? not quite shortcuts can't act as directories for paths Hmm, but the shortcuts can link to files, right? yes, but they still don't alias the file Darn i.e. they work for the mouse, as in kdelinks, but that's it How would you orgnize a bunch of files how do you mean? that belong in many catagories? oh i use beos or linux :) beos doesn't even need the symlinks :) Like a boot sector to BlahOS, it belongs under "Boot Sectors" and "BlahOS" * lar1/#tunes is trying to orgnize the stuff from Borg borg? airs repository of documentation on osdev stuff oh Do you know how windows handles symlinks? heh it doesn't Heh i have an ext2mounter for windows and it doesn't handle symlinks either 11:10am So should I orgnize them ala ext2 or should I just pick a method of catagorization and run with it? * water/#tunes shrugs do whatever's most useful Heh... I will go nuts... just ask hcf about how I try to please everybody :). If this were a pile of HLL docs that were useful to you, what would you do? as long as they're hyperlinked, i wouldn't care Hmm, ok grrr odp hasn't added maude yet like i asked twice maude? abi: maude? somebody said maude was a reflective rewriting logic language at http://maude.csl.sri.com/ or http://www.csl.sri.com/~duran/ it's extremely versatile for a logic language -:- ult [noone@user-38lcn5m.dialup.mindspring.com] has joined #Tunes logic language? yeah you know like prolog? prolog? grr Nevermind, I will look it up look it up on www.dmoz.org 11:20am ok oh oh oh oh oh bbl -:- SignOff lar1: #TUNES (Leaving) 11:30am -:- hcf [nef@me-portland-us202.javanet.com] has joined #tunes hmm foo how r things coming? you see most of the devel already rnt u devel'ing locally? i'm working on about 3 different issues today math/logic/text stuff for one also i need to fit in the stuff we discussed about namespaces and related things the other day (web docs) and co-monads so i can rationalize the use of the : keyword any comments? have enuf papers relating to the issues? ;) hm 12:20pm it'd be nice if one person on this planet could relate co-monads to oop without resorting to haskell have u queried researchindex for 'comonad'? oops forgot i'd been greping for lots of other stuff hm nothing applicable showed up -:- bineng [Anders@j141.ryd.student.liu.se] has joined #tunes hi bin hello! r u including and excluding the - b4 'co'? yeah k, i'l see what i can find comonad/oop wise thx abi: abireq? addressing me is now REQUIRED. i will only listen/learn/respond when addressed publicly (abi: ...) or privately (/msg abi ...) oh anyone built cmucl here? water: Now when we're both here, I can as well hurl a few q's at you.. at the site you says slate doesnActivities with side-effects crap ...slate doesnt' support activities with side-effects at the language-level yeah so, I wonder, how do you do such things? not very difficult to do in fact, that statement should be modified (sorry there are a few inconsistencies left in the docs) 12:30pm * ult/#Tunes has built cmucl many times np the behavioral meta-object will allow you to change the mutability of state it's a real bitch. 'change the mutability', does that mean to change whether the objects can be changed? which in turn determines whether "side effects" are actual side-effects or just forcing a new environment with new bindings bin: sort of ok there's a lot of detail to mutability that we're working with wouldn't things be simpler if you assumed mutability (or the contrary) universally? not really the system we're working on allows a greater amount of unification than ordinarily possible and mutability defaults will usually be very intuitive smkl: you're probably better off using Peter's debian packages (if you're on linux) potato ones i don't have debian, and i don't want cmucl .. an another person is asking help yikes... too many monads papers on compiler modules there is a page about cmucl on linux which tells how you convert the debian packages to other systems there are working rpms as well but I can't remember where bineng: need more details? water: nah, not necessary ok instead I can ask how slate will deal with infinite sets such as integers or real number. Are they limited to a certain size? How would they be represented? 12:40pm ah i'm working on that today :) oh really Anything you can say about it? yes i discussed the issues early this morning with eihrul you may as well just troll the logs k i can answer more questions, of course well, ofc it *would* be simpler if you could just sum it up.. er... not really mostly because i haven't answered a lot of the questions -:- redronin [redronin@nrbg-m191-158.pool.mediaways.net] has joined #tunes hcf: hm i think i've found one (been trolling through references at RI) hoi hi water: url? ult: I've heard you have experience in building cmucl? hcf: hold on let me read the thing first hm well it explains oo lang concepts very well in terms of procedural/functional concepts * bineng/#tunes brb not incredibly useful to me still want the url? 12:50pm sure http://citeseer.nj.nec.com/did/149666 hm. haskell++ ok not what i hoped for :P * bineng/#tunes is back redronin: hmm ult and fufie seem to be idle, fufie said that you could use debs if you use linux u'v been thru wadler's poop right? yep 01:00pm geez. some of these haskell papers require a phd in category theory Gakuk hi fare lo water * Fare/#Tunes gotta write a paper about the TUNES / PHENIX architecture, but I've got problems organizing it phenix? smkl: I've tried so much different packages of cmucl that I've given up counting them ;) red: what system are you installing on? but another topic: the TUNES project sounds sounds amazing!! heh yeah when we make it work, maybe Fare:Linux (kernel 2.2.10, glibc 2.1) which linux? (why 2.2.10??) can someone explain what phenix is for me? Fare: Suse 6.2 a mispelt phoenix? red: it's "Fufie" :) sorry I mean Fufie 01:10pm red: get the cmucl packages from debian potato, and use alien or something to get the files and it should work * water/#tunes nudges Fare I need to rebuilt 'cause all binary packages are built withoud --export-dynamic option so callbacks from C->lisp code seems not to work red: if you go deja, you can also find a recent discussion about cmucl+linux on comp.lang.lisp someone ^G Fare Fare: good idea! red: the debian packages has source which is specifically meant for rebuilding.. it isn't too hard brb changing servers -:- SignOff water: #TUNES (Leaving) -:- water [water@tnt-10-63.tscnet.net] has joined #tunes -:- eihrul [lee@usr5-ppp05.lvdi.net] has joined #tunes I've tried the actual unstable debian-binary with th actual debian-src but it breaks on not finding the kernel.core - which I think have to be generated! hi eih lo red: sounds weird.. you can have mine At the time I download the stable debians but if this also fails... redronin: so you're here, too? fare: please explain tunes+phenix no, get the potato packages.. not unstable or stable potato is frozen Fare: what do you mean? Fufmann: no, it's the fscking british who mispell "phoenix". The French correctly spell it, phénix. redronin: multiple irc servers... fuck the spelling, what is it? hcf: what is PHENIX? * water/#tunes waits again through Fare's usual s*** actually, it's spelt ¥¦§¨©ª«¬­ a joint project between labs of the CNET, the LIP6, and the INRIA, to build a flexible execution framework iow lll water: can you elaborate about "s***" ? water: bingo! Fare: yes smkl sent me here to ask ult about my cmucl problems what's "iow"? you could have said "phenix as lll" -:- SignOff ult: #TUNES (Read error to ult[user-38lcn5m.dialup.mindspring.com]: Connection reset by peer) in other words that would have actually been clear are you all developers for this promising TUNES project? we'd like to think so ;) and I'm more or less in charge of defining the first deliverable of the project, an architecture document redronin: I see no developer for TUNES, anywhere and i'm more or less a competitor (not officially endorsed by tunes) 01:20pm iow, tunes' best bet or worst threat :) iow ? especially if i make tunes in a way Fare will not accept which seems likely right now water: I emited no opinion in either way, for I don't know enough. exactly hcf: what do you mean with "iow" abi: iow? hcf: i haven't a clue odd abi: iow is In Other Words axo ;) hmmm... "Final Coalgebras as Greatest Fixed Points in ZF Set Theory" 01:30pm foo? questions? comments? discussion? the tunes-projects sounds similar to the ideas a friend of me an I had for some time yeah, why do ppl endore lisp so much? hm red: a lot of people have had similar ideas, including myself water: icuc, http://www.md.chalmers.se/Cs/Research/Semantics/APPSEM/webpage/ http://vesuvius.cs.uiuc.edu/sipl/scill.html bin: because of the ease of macros and the cons structures, i guess water: I looked through the log but didn't find a clue to the answers of my number questions. bin: what about the stuff on inductive / co-inductive ideas? went past above my head hm so do you understand induction at all? thx hcf induction, guesso. co-*, not at all. you understand the viewpoint that 0 / 0 = all real numbers? I haven't seen that one before, but why not. that's co-inductive but what refrains 0/0 from including complex numbers then? usually, 0/0 is undefined (raises an exception if you will) the current set of algebric operations you're working with I've only heard that 0/0 is undefined but this sounds interesting - can you explain? 01:40pm both inductive and co-inductive are *aspects* of solving equations in abstract algebra water: ok. So how can this help in *any* way in practice? ack well a lot of crap in programming languages is co-inductive but gets crammed into inductive definitions so you wind up with lots of noise i.e. undue complexity is co-induction more narrow than induction, then? for instance, state is a co-inductive idea no, they are on par with each other opposite ends of a spectrum induction is proofing a beginning-state and then proofing the change of one state to another kind of Can "state" really be applied here? its more specific than that sorry but I'm not so good describing such complex things in english as my mother-language is german ah np yeah induction always requires a kernel element to start from isn't 'assertion' more appropriate word than 'element'? no they are equally valid k grr research index has some weird bugs water: like what? does this clarify why i'm working on the math/logic/text system? hcf: some of the url's return their entire url as filename, which screws over the file s/url's/links/ water: how do I use co-induction for solving assertions - can you give an example? hm 01:50pm i think it mostly relates to declarative programming -:- AlonzoTG [Alonzo@207-172-184-240.s240.tnt6.lnhva.md.dialup.rcn.com] has joined #tunes for example some stuff in rewrite logic displays co-inductive semantics water: of pdf, ps, dvi what is ur preference order? (1) .ps.gz, (2) .pdf (3) .dvi k I guess it's to complex to explain in short but do you know of any good papers on co-induction? dvi would be first, but i haven't found a good win32 dviewer s/good/simple/ maybe icct abi: icct? i guess icct is Interaction, Computability, and Church's Thesis at http://www.cs.brown.edu/people/pw/papers/bcj1.pdf i haven't found anything more intuitive :( sounds good i may as well feed you this one, too abi: agt? well, agt is Around Goedel's Theorem at http://www.ltn.lv/~podnieks/ are you studing computer-science? heh. i'm an amateur, yes self-taught respect yes, i have none ;) and neither do i ;) water: icuc, http://www.brics.aau.dk/BRICS/RS/95/20/BRICS-RS-95-20.ps.gz ftp://reports.adm.cs.cmu.edu/usr/anon/1991/CMU-CS-91-190.ps ftp://ftp.cs.unibo.it/pub/asperti/fundamenta.ps.Z yikes dude thx what are those, anyway? hm ok 02:00pm [co]monad poop yeah the third one i haven't seen oh damn i forgot an unresolved slate issue i have to finish the matching between sexp syntax and my set-styl objects-with-slots idea to get macros somehow anyone got ideas? :) uh, magic? hm i do know some pretty powerful wiccans :) hmm i've never used CL so i don't know about macros :/ smkl: symbolic tree manipulators, that's all ok. i guessed they would be something like that there are other kinds of macros, too, but i'm not concerned with those om perhaps you need a set constraint solver how does the macros fit into your semantics? smkl: how's that? fufie: well it's mesage-passing everywhere with messages and objects able to trade places in various situations 02:10pm trading semantic places, not just syntactic but are macros allowed to change that? no, they just move symbols and require the evaluator to support the transformed structure, right? that sounds like very simple macros what kind of syntax do you use? red: a kind of sexp like in lisp red: check out the semantics page on the slate site in CL macros are allowed to do basically whatever they want to do, redefining semantics, syntax, etc hm well that's due to reflective nature of lisp applicative programming, no? yes i.e. hofs? and because lisp takes the programmer seriously CL macros are very non declarative? I would say so well i know i can get hof out of slate trivially I doubt CL macros are what you want but it's not quite as direct as with lisp just yet yes but i can't put my finger on what's appropriate for slate have you checked out Dylan macros ? hm not closely * water/#tunes looks I think that might be closer to slate ok i'll consider them got a url maybe? two sec dylan? yeah some of the links i found were dead www.functionalobjects.com ? http://www.functionalobjects.com/ heh okay sounds like a Big Clue (TM) ja cool. maude is now in ODP 02:20pm hm i just realized that my identification of parentheses with namespace stack operators changes the mapping to sexp significantly afk hm is there a comparison anywhere of dyln macros vs lisp macros? 02:30pm http://www.federated.com/~jim/drm/drm-76.html dylan macros from the ref, in case u need it ok heh. "rewrite rule semantics" i love these clues -:- SignOff hcf: #TUNES (Ping timeout for hcf[me-portland-us202.javanet.com]) -:- hcf [nef@me-portland-us202.javanet.com] has joined #tunes 02:40pm hm the syntax page will probably get canned oh? yeah i'm thinking i should factor out a lot of the literals' crap of course i need to refactor it into syntax and bnf abi: glass is at http://realm.progsoc.uts.edu.au/~subtle/projects/glass/ interesting 02:50pm uhhh html basis for lexical component not a choice i would have made, but oh well :) 03:00pm * water/#tunes watches the tumbleweed roll by * eihrul/#tunes kicks the tumbleweed. hey :) what's up, man? you quiet or just busy? mudding oh so kind of neither and just reading through all the links casually :) looking at glass right now btw the right example last night was x=0/0 not just x=y/0 what's the difference in particular? well x=0/0 => 0*x=0 which means inductively x is undefined but co-inductively, all values of x apply for the normal system of algebra, this means the real numbers ah, co-inductivity was that everything that is not disallowed is allowed, right? yes ..while induction would be that anything not explicitly allowed is ruled out. Or something. water: what about 0 * x = y... just seems that y *has* to equal 0 there :) :P though... if y does not equal 0 then x would be undefined no? inductively yes co-inductively hm well, has does co-induction help the situation? :) s/has/how well it simply doesn't in this situation, afaik but it's an open field doh... but hey, i suppose 0/0 is a consolation atleast :) maybe co-inductive views from an alternative algebraic system would allow infinity in fact.... 03:10pm i think it really would if you included the logic of limits so yes i believe it does help, but you have to shift the perspective a bit how so? water: if I may bug you about this again, how about numbers in slate? Finite in definition or not? using the algebra of limits to include infinitary ideas into the real number system sorry, it's not a very intuitive idea 0 times some infinity can be non-zero? :) but 0*the number of real numbers=y co-inductively yes well, it is intuitive :) water: To take a concrete example: sooner or later my bignum grows too big for your system. Does slate neglect or address such a problem? in fact, i figured this out using a weird formal system when i was in geometry in high school bin: no, of course there are ways to handle that why do people say bignum? it's a friggin integer for crying out loud! "there are ways"? water: do differentiate from the butchered meaning of integer provided by most languages... it's just to annoy you, I'm sure bin: yes by having a representation of the memory system sice bignums are specifically bigger :) s/do/to water: your answers are very elusive well rewrite logic handles gc strategies, ok? that way the allocation of new memory atoms can be intercepted at the language level using bmo's of course :) -:- AlonzoTG [Alonzo@207-172-184-240.s240.tnt6.lnhva.md.dialup.rcn.com] has left #tunes [] water: And the result is? How does it look for the user? as though an exception is raised, pretty much, although it's more coherent and user-friendly for instance, stuff could be moved into a persistent store transparently it's up to the environment s/persistent/secondary/ ? yeah whatever alright 03:20pm like i said, it's up to the environment yes any more questions? :) umm, I think that's enough for one evening, thank you :) not for me it isn't :) s/:)/>:)/ ->:)-><-> heh * eihrul/#tunes ponders if that's a valid befunge program. god damn its tough to get to the meat of this glass paper looks more like false or tml eihrul: i have a sort of personal question to ask any of you done any work with pipes and CL? do you talk to your family about what you're working on with me? me? yes you i was just sort of wondering what they thought of it 03:30pm * eihrul/#tunes shrugs. they know barely anything of what i do i mean you're reading all these papers and books like acd&i don't they wonder what kind of book you'd spend that much money on? * eihrul/#tunes shrugs. they just give me money lol they know i spend hundreds of dollars on books but they don't really bother to know what do they worry about you? the times they have asked, it hasn't really mattered... because they didn't know what any of it was sometimes heh but, it's good that way ok sounds like roughly the same way my family treated me i've heard horror stories of children in my school with math teachers or engineers for parents... except i was more into *totally* impractical stuff like math, logic, and philosophy, as well :) how so? well, they're forced to do as their parents... water: icuc, http://www.cs.cmu.edu/~fp/elf-papers/INDEX.html * hcf/#tunes is away: (afk) really? well, not necessarily forced... but it just cheapens the whole experience when knowledge is forced upon you :) yes gee that sounds like a good argument for not giving technical types the right to raise kids well, noone should necessarily have the right to raise their own children :) true i mean, when do children get a say in the matter? do they get to pick their parents? no... i always thought that was quite unjust... just? i just think it's completely un-intelligent a lot of your educational experience depends on the family in which you come from... of course i.e. not everyone can afford as many books as i can :) well not only that but if you lived in the ghetto, you'd find all sorts of factors against you availability, safety, peace and quiet that too and i would never have been able to learn of taoism or different mathematical ideas without some tolerance 03:40pm or perhaps it's degeneration of society that's ok, though, because it's past time to start building something new -:- SignOff eihrul: #TUNES (Leaving) ? -:- eihrul [lee@usr5-ppp05.lvdi.net] has joined #tunes mmm, socialist revolution... nah, something better like a language revolution or something well, depends if you're solving the problem bottom-up or top-down... that'll upheave *everything* bottom-up the *very* bottom takes a few generations though :) sure, but an effective weapon / technology at that bottom level could be great leverage like <> >:) hm the glass guy brings up good points about various languages doh... forgot to continue reading like how smalltalk's "image" idea is very non-conducive to portability well not so much portability as diversity * eihrul/#tunes up-scrolls. about 40% through the paper though how do you check-point the state of the system without some sort of image/ or atleast what's necessary to construct it... hm not worry about the system so much as the parts? well, the system is a single slate object :) that doesn't mean we can't refactor it properly * eihrul/#tunes nods. 03:50pm darn, i was waiting for a lurker to shout "the entire slate system is just one object?!? witf?!?" :) also, did we ever decide on what the root namespace is? hm nope any thoughts on the slate object being said namespace? hm well it does make an excellent root for system gc i suppose so so for now, that seems appropriate keep in mind, though, that i'm seriously toying with the idea of a circular system of containership -:- SignOff hcf: #TUNES (Ping timeout for hcf[me-portland-us202.javanet.com]) hmm? i'm saying the 'root' could be actually contained by another object, as opposed to being cloned -:- hcf [nef@me-portland-us202.javanet.com] has joined #tunes trust me, if i actually decide on this, it'll be well thought-out -:- SignOff hcf: #TUNES (Leaving) well... even the root object needs a container :) -:- hcf [nef@me-portland-us317.javanet.com] has joined #tunes er the slate object maybe or maybe it should be cloned, with the clone contained hm in order to clone, one must refer to the original object 04:00pm ergo there must exist a namespace containing the root namespace for this purpose i propose that the root namespace shall refer to itself :) any objections? nope was thinking that before when i said even slate object needs a container :) "when in doubt, make a circular reference." :) heh. now we have three hierarchies with loops at the top Top, Root, and Slate? yep may the void help us if we find need of a third fourth oops sorry, i dropped a bit between my cerebrum and cortex :) hm no that analogy doesn't work oh well -:- air [brand@p0wer.qzx.com] has joined #tunes hm let's shift the subject how about concurrency? -:- SignOff air: #TUNES (Read error to air[p0wer.qzx.com]: EOF from client) -:- air [brand@p0wer.qzx.com] has joined #tunes hmm... no further ideas on that as of yet ok grr... i just remember something else i forgot to continue reading :) what's that? the merlin/language 04:10pm oh * eihrul/#tunes ponders if he has an attention-span disorder. no, i doubt people of earlier times ever had to deal with such amounts of information hell 200 years ago, the most complex mathematics was the calculus water: ok, done w/ comos for now, whats next? now, every single day i stretch my attention span to absorb more concepts than they ever dreamt of hcf: no more for today please :) water: they probably had less noise as well... odd phenomenon that more signals have created more noise yeah if you met a guy like AlonzoTG you could just walk away :) no, mommies make the noise -:- SignOff bineng: #TUNES (c'ya <k!14>) the signals are pretty much just a reaction you have to remember that the signals we're making are a desire against human life to an extent hm the glass guy has an immutable keyword for reference typing s/immutable/'immutable'/ 04:20pm ah. i just remembered something notice that the messages sent to an object to create a new one also constitute an object this can be used to implement mi to some extent you can clone the 'changeset' that produced one new object and apply it to another (which already has a parent) so you wind up inheriting things in a sort of indirect manner without screwing up the original intent of the single-inheritance hierarchy one problem is that you lose the metaphor that the multiplpe parents interact in any way other than sequentially you grok this? hey eihrul -:- future [xdef@209-6-184-165.c3-0.wth-ubr1.sbo-wth.ma.cable.rcn.com] has joined #tunes somewhat, an example might help hm (just to make sure i grokked correctly) ok consider that in mi you're inheriting slots from multiple parents, usually (and hopefully) ones that don't conflict so you can encapsulate the changes required as 'add-slot' methods so if i have a 'car' object that inherits from the 'vehicle' hierarchy, it can inherit it's 'red-paint-jobness' as a set of added slots as well as state-constraints 04:30pm if there's some way to keep the process lazy and still get those slots to respond to messages, then we don't even have to actually replicate the slots at the new location so just inserting the slots from the particular extra parent directly? yes and using object-encapsulation to help clarify the process you still have problems with intersections though yes that's true but it does give you explicit control over how to resolve it so does MI... well sure you just define a method that combines or resolves the two yeah but we do the same here without the extra method since we already have to express object creation as a sequence of messages modifying existing objects well, parents could be added incrementally :) eek yeah but my objects are *sets* of slots, not lists even an added resolving method doesn't help conceptually, just implementationally well, what if two extra parents are added via your method... and they have intersecting behavior? they won't one will override the other if method names collide the first one added is muted and the problem i see here is simply that mi is conceptually dirty 04:40pm the only thing it offers is intersections are handled in a definitive order :) so people trying to implement mi using the system i propose will have more explicit control, and they should, because mi causes complexity *the only thing*? harumph :) whereas in mi the order in which the tree of parents is search is not specified :) though... well i guess this means we don't need to implement mi within the language it might be cleaner to create a union of the parents and then inherit it :) since obviously there are so many ways to do it within the language union suggests concurrency of addition of inheritance i.e. non-determinism hi future not necessarily... if you can only take the union of two classes er bad word :) oh really? objects if the second operand of the union takes higher precedence so what? dude listen then you apply it successively you can constrain concurrency to resolve it by static or even dynamic preference interfaces :) what about them? can be used to specify preference like i said, there are way too many ways to do mi within the language so drop it you brought it up :) 04:50pm yeah but you didn't have to argue so much wasn't arguing :P sure sounded like mi-advocacy to me it wasn't fanatical, though true ok, subject shift you pick this time * eihrul/#tunes ponders. well i was just reading the glass paper before we got into this :) yeah so was i and i got tired of it * eihrul/#tunes checks the logs. * eihrul/#tunes hmms. what of math and text? it's crzy how even our signal discussions of issues are directed by the random findings of hcf well got questions? well, how are you going to do them? :) i'm definitely not ready to give pat answers yet oh gee that helps :) ok... if you're still researching this will take at least a few days * eihrul/#tunes nods. and interaction about the ideas can often speed up the process well, i need to atleast know what you're looking at so far if you expect me to help in that respect :) unfortunately there's not much out there hm perhaps i should go out for a bit and relax 05:00pm sure damn, i finally finished that damn glass paper nothing incredibly new i find it odd that such people don't reference tunes do they know about tunes? dunno tunes is a gedanken project.. why reference it? gedanken? gednaken as in thoughts gedanken even huh? gedanken is german for thoughts oh a gedanken project is people who talk instead of actually writing code.. well i suppose any project that claims to do away with ascii as a standard is a thought-exercise only :) ok we wouldn't have to talk if fare had coded years ago or anybody yup but then, tunes needs a lot of research that in a lot of cases hasn't been done until recently all right! enough talk ;) bbl a former boss of mine had a PhD in computer science.. she couldn't program hello world even if her life depended on it.. her thesis was just thoughts "it would be nice if.." a classic gedanken project -:- water [water@tnt-10-63.tscnet.net] has left #tunes [] funny why is there such a separation? :) que? the same people who decide what programmers will be using have no idea what programming actually is? :) 05:10pm yup, sad isn't it? and a lot of the gedanken people actually get hired and get well paid jobs and visit customers and design solutions for them.. bereaucracy at it's finest :) s/it's/its and it's the hackers who gets blamed if things don't work... not the gedanken people because they have a nice degree and wear nice clothes :) well... there really are incompetent hackers :) they do exist and what about the hacker with nice clothes and a phd? does he still get blamed? :) yes because he's smarter than his boss ;) reminds me of something... in advanced compiler design and implementation the particular phd, who's designed compilers for hp and sun... made lots of errors in his C code :) er rather, C code examples in the book... 05:20pm it makes you wonder :) heh he made it many times over and even explains the code to do something it doesn't... so it's not just an accident :) though.. a lot of gedanken people write decent thesises.. the rest of us spend the time hacking on half-a-dozen projects and never get around to writing text well, aren't you on your way to getting that phd? :) child of the gedanken in the making? :) I have actually written a lot of code for my thesis :) just watch your back! the gedanken monster may come up one day and thwap you upon the head maybe.. not today though.. I am trying to make a lisp-app do some eq. of select() on a pipe here.. but it seems to ant to do a blocking read and eat cpu s/ant/want/ 05:30pm you made pipe non-blocking? no I want to do a non-blocking read well, does that not require making socket non-blocking? (well, not technically, but) sortof... it's a pipe though, in an effort to make the listener portable pipe/socket, little different nce... they're all file descriptors :) in C, I would basically do a select() call yeah, but you're trying to implement an equivalent, no? :) I might end up writing the reading-function in C though.. 05:40pm there's no O_NONBLOCK ? or O_NDELAY not in ANSI CL :-/ I must go to the implementation defined level -:- AlonzoTG [Alonzo@209-122-212-191.s572.tnt7.lnhva.md.dialup.rcn.com] has joined #tunes or C hrmm maybe I can make a kluge with READ-CHAR-NO-HANG 05:50pm in cmucl you can also use reading from sockets with serve-events you can then hook handler-functions for input and output 06:00pm red: yup.. seems as if that is just what I need -:- rares [rares@nwhn-sh14-port164.snet.net] has joined #tunes 06:10pm I think I will use it for writing an irc-bot in CL i've written an irc-bot in C++ as irc-protocoll designstudy for my "aisoku"-projekt -:- lar1 [larman@sdn-ar-002casfrMP155.dialsprint.net] has joined #tunes but as I took a further look in lisp and scheme I realized the primitive functionality of C++ so I decided to use CL in my project have you written any code with SERVE-EVENT for file-descriptors? not yet, but I have a short example in the user-man. of cmucl - should I paste it in? (it's short) sure or the page number C++ is a nonobject centric object system oh sorry, after looking in the man I see theres no such example, but the needed functions are on page 106 yup that statement is devoid of content which statement? C++ is a nonobject centric object system 06:20pm I think C++'s staticality (is this the right word?) is the worst thing in it how so...? If it's compiled you have few possibilities for changing your objects yes, object definitions cannot change dynamically, because if they could the system would no longer be typesafe well it is, C++ allows you to build objects however it's still very procedural the object to object relationship system is very "contrived" so to speak (class definitions rather) am i lagged? no my brain is in what sense is it contrived? and you need incredible hacks like templates and rtti to gain a little dynamicality back what's the word I'm looking for... gimme a minute i haven't really found any good reasons to use RTTI thanks red that's what I'm trying to say where templates are only syntactically dynamic tho, i stay away from templates because they tend not to be too portable you need hacks to make it get more power out of it what's a better way to achieve the same goals as templates? (i agree it's a bit of an over-extension of the language) future: in C++? in general 06:30pm dynamic type-safety? all objects in a language should have their type by itself (like in lisp) or like in C++ RTTI but in lisp the essential trick is to check an object's type at run-time to ensure safety right? you don't need RTTI in C++ but the actual RTTI implementation isn't useable 'cause of compiler-differences and many other quirks you can use templates and everything without RTTI. I don't like it myself. it goes against the spirit of OO but in the real-world its sometimes necessary lisp-people call this typ-system "dynamic typesystem" am i right in saying that in most higher-order langs, generic language features (as in templates) require run-type type-checking/binding? yes templates lack polymorphism as a result of this with templates you can reach some (source-code-level) dynamicality but no runtime-dynamicality sounds like a significant performance penalty.... tho i don't know the numbers on that * AlonzoTG/#tunes is making progress on his 0$! =) Fare, there is something called template specialization in c++ It's going to be a bitch to bootstrap but it'll be really mean! =) its a 'new' feature and i don't know much about it, but i guess it addresses the problem of polymorphic templates oops, i meant fufie it only shortens the amounts of templatecode to write it doesn't templates lack polymorphism and is therefore not orthogonal with classes and inheritance the behavior of the template can change depending on the type its instantiated with is that not polymorphism? no why not then you also need to take into consideration subclasses late-binding polymorphy is late-binding (runtime-binding) you decide at runtime what operations you call for an given object i dont see the contradiction here to get around it you must save pointers to the objects in the templates instead of objects and you have gained nothing so it's possible with virtual-methods to runtime-link in C++ Objects that are later written and compiled then the host-program itself (like a plugin-system) C++ sucks... there's your whole argument in 2 words :) 06:40pm C++ is a half assed attempt at objects ok so c++ is a bundle of quirks and historical accidents... yup but where's the practical alternative well, in what it tried to accomplish, it did okay lisp but it's what it was trying to accomplish that was the problem... read "worse is better" about the different styles of programm-design and why isn't lisp used widely in industrial apps? because C++ is more wide-spread :) cuz lisp doesn't let you design half way there are several conclusions on this om lisp is the second oldest existing programming-language (after fortran) so this means programming languages don't die... they just fade away :) http://www.kde.org/food/worse_is_better.html lisp is used in industrial apps it was built to compute symbolic problems like symbolic integration and differentiation Fufie, i can only think of emacs it's not used in desktop apps except emacs industrial != academic is autocad or interleaf industrial? autocad is written in lisp? part of it, and lisp is the extension language nasa uses lisp hrmm. i can see lisp as a nice 'glue' or 'interface' language lisp is an any-language :) the right solution is almost never adopted anyway. MAC vs. DOS, BETA vs. VHS, etc.. to say it's a glue language is an understatement... tunes vs. everything else :) concrete can also be used as glue :) but to the point: lisp had the reputation to be slow and only usefull for AI-programming redronin, still has that reputation' also, c/c++ probably had better portability back in the day hopefully... compilers will advance to the state that the level of the language has no bearing on its efficiency :) and we're already getting there :) eihrul, for that to happen the language has to have semantics that are amenable to optimization future: yeah, a C implementation would be considerably simpler to make than a full featured Lisp implementation we're getting there? future: no.... for that to happen, the optimizations have to be more amenable to the semantics HrMM going the other way is counter-productive why? optimizing c++ code is an IMPOSSIBLE problem! (complete optimization) well, C++ is a waste as in the '80 the world realized that the AI-research hasn't brought the roboters and electronic-brains they ever dreamed of, the AI and LISP research was nearly dead because sometimes, low-levelness can be a barrier to optimization :) like, if a language supported set-wise matrix operations, you could take advantage of built-in matrix instructions. as in SetA * SetB .. u can't lways deduce that from a loop in a procedural lang future: loop unrolling, code hoisting, memoization, etc :) redronin, the robots are here :P extending the language is not the solution or rather, extending it with black-boxes for optimization eihrul, yah but where do u see real advances happening in optimization adaptive compilation! :) cmucl has a lightning fast native code compiler!!! you mean run-time optimization? no adaptive compilation explain.. using run-time feedback to recompile the program ever more efficiently hot-spot-technology red: it's rather slow but compiles fast code redronin: yep, and self hot-spot is run-time opt you mean hot-spot 06:50pm future: it's called "adaptive compilation" though :P oh. i thought thats what i said ahh abi: self? rumour has it self is a prototype-based object system at http://www.sun.com/research/self/ or old mirror at http://www.cs.ucsb.edu/oocsb/self/ there are some papers on it there especially on how they use instrumentation to gain type information i will read this stuff looks like its really old yeah OOPSLA '87 Conference Proceedings so just think how much the technology has come along :) did java kind of absorb all this (sort-of) ? well, self is pretty old but some of the papers are recent future: yeah, hot-spot... pretty much the same research effort :) (since java is a sun project) well you need a run-time environment to do adaptive compilation so? why is that a problem? every programming language has some run-time environment even C its not, but in reality its hard to deploy i don't see why... ermm, the C runtime env isn't dynamic so? doesn't mean any other can't just as easily be and you have dynamically linked libraries and things like that :) right right in Lisp systems, this isn't a problem :) you could do *alot* more with optimization if you could rely on some *constraints* eh... constraints are hyped up ;) not quite well part of optimization is deducing the hidden constraints, right in some respects but you can assume constraints that are themselves variable and invalidate the results of optimizing for those constraints arr[i] in java always does a bounds check, but with optimization you might be able to deduce that the check is unnecessary if the constraints are changed... the important part it to not lock certain constraints :) s/it/is right.. but the range of this kind of deduction is limited to *local* areas you can optimize parts of a loop or a block of code or whatever, but things like global optimization are totally out of reach read the self papers or what-not :) not really.... ehh maybe cmucls compiler has strong optimizations like: tail-recursion, local-calls, constraint propagation, automated type-check generation,environmental analysis, control-analysis,stack-analysis, copy propagation, lifetime analysis, Pipeline reorganistaion,constant folding unused expression elimination, unreachable code deletion, source to source transformation inline-expansion..... redronin, most c compilers do most of that but not at runtime ok so what's the point of our discussion? :) just wondering... oh. i forgot oh yah, how to get super-optimized code * eihrul/#tunes shrugs. re-optimizing the code :) if the language is equivalent to the spec, and can be formally modelled, then you can optimize it to inifinity so that it can be dynamic and optimized at the same time i know no c-compiler that supports tail-recurion GNU C does tail recursion :) it also has real closures :) 07:00pm but be Stallman's Lisp fix :) i've tried it it doesnt work for my simple test :) s/but/must prolly using it wrong then :P tail-recursive faculty after all this hot-spot stuff, java will end up being a little slower than c++. which is really good.. but it's not quite the holy grail of optimization * eihrul/#tunes shrugs. you're missing something though... you can also do argument-caching if ever you tried to make C++ code as polymorphic as Java code it'd be only as slow as the optimized Java so in suitably complex code... (hotspot in hardware !!! crusoe) where polymorphism is needed it would make little difference... and for the last fucking time crusoe does not do hot-spot in hardware! it does it in *SOFTWARE* :) fine fine i'm sick of people who think it does :) but it's firmer than most software it merely makes it easier for the software to do its job it doesn't do the job itself * eihrul/#tunes shrugs. not really a lot of hype over nothing, if you ask me :) ok, so how bout real hotspot in real hardware wouldn't be worth it hrmm. maybe not. that would be one complex piece of hardware would make debugging harder :) yes... and as compiler optimizations get better you'd have to put out new hardware :) that's not good... performance might get a bit unpredictable if the hardware is doing lots of magic stuff behind the scenes sparc's don't even have multiplication but eventually (50 years from now) everything migrates to hardware anyway umm ?!? er division :) and you want compilation in hardware :) that'd really fuck with the pipeline... :) hehe okok. lets be RISCy Heh though it's still faster to do special case multiplication than hardware multiplication :) you mean as a switch ? or table lookup? combinations of shifts/adds/subs oh which end up being faster than the single mul instruction hrmm didn't know that all 3 operations take only 1 cycle multiply takes 11 cycles on x86 for example is that tru for *any* arbitrarty set of operands? so you have a lot of room there well sure.. doing shifting for powers of 2 is faster than mult even for non powers of two... -:- ult [noone@user-37kbas5.dialup.mindspring.com] has joined #Tunes for instance... * 7 can be done by multiplying by 8, and subtracting true which is 2 cycles :) 07:10pm -:- SignOff lar1: #TUNES (Leaving) ever tried fast-exponentiation? for numbers of arbitrary length like needed in diffie-hellman encryption that kind of automatic-optimization is not too feasible if you can't express multi-precision numbers in the language itself.. that's basically the problem i think.. all the huge gains in optimization would come from uncovering the high-level strategies used in the low-level code why does so many "modern" languages have no inbuilt arbitrary-number or complex-number system well... same reason there's no square root operator :P the idea is to in-build NOTHING and make everything expressable in the language itself... the language will have no syntax! it will float on a cloud of thoughts! solves thr problem nicely ;P but arbitrary numbers are NO _extension why should 32-bit number operations be 'intrinsic' and 128-bit numbers aren't ? if you don't build them as absolutely low-level code they are ass-slow redronin, same idea with matrix operations modern x86 chips have special instructions for matrix operations. they are incredibly fast redronin: hah... but... who cares? the idea is to make them fast! :) but you can't use them without descending into asm! to make any arbitrary idea specifiable in the language fast! om not just some special purpose idea the language provides! eihrul, aye there's the crux the problem is also the inconsistency you gain when you split up number-systems 07:20pm there's on rational-implementation here and on complex implementation here ... and another arbiturary number implementation there... yep hrmm * future/#tunes reads self wow. it was four years between self 4.0 and 4.1 most "modern" languages are using the "von Neumann" model of computers * eihrul/#tunes ponders quantum languages... quantum langs? just a buzzword :) languages should be abstracted from the underlying hardware as of late in everything.... i was joking :) redronin, have you seen dataflow languages? they don't have any von-neumann-ish restrictions future: no - explain :) 07:30pm i'm not really qualified to give a full explanation :P but, in essence you have statements which are arranged in 'blocks' (everything in a block can be executed *simultaneously*) and the blocks are hooked together in a 'pipeline' like arrangement so data flows from one block to another and so forth but, you don't specify how to move data around and where to put it.. it just flows seamlessly along the pipeline and all operations are happening *at the same time* sounds interesting i guess ppl are using it on parallel computing platforms you can do a search and see some interesting results om I think there is no language that fits all programming problems agreed what does om mean? it's my mantra. =) hehe (both om; (ohm) and mantra are in the dictionary) Like C/C++ which are good at programming conservative OSes on conservative hardware my prototype will be written in ASSEMBLY. =) prototype of what? the prototype of my OS. =) so then the key is a general-purpose language which can be adapted for new tasks. either through OO methods, or something much more high-level what OS? users.erols.com/alangrimes/ a meta-language I am working on the specifications right now... ooh u flamed slashdotters :P I don't want to treat you but lisp is considered as _A_ language of this type redronin, i guess so all OO-paradigms and others came long after the first lisps and lisp was able to include all this things so AlonzoTG, have u used Be? 07:40pm nopers. I looked at the kernel a little bit... Anyway... I use a P100... =\ mach bummer nah... I have a 19 inch monitor... Found a nice piece of bellybutton lint yesterday It's convergance is only off by a single pixel! =P haha yeah, I was just being silly... I'll clean up the page if I can get a fully functional prototype togeather... how far along is the project? I have been working on it for many years but am nearing completion of a register level software specification. that's as far as it has gotten. have you looked at the Flux OSKit? The biggest problem that I am facing at the moment is trying to figure out how to debug a partially complete OS. Yeah I am on the oskit mailinglist. all these ideas are great. simplicity is key :P =) absolutly. The os must run on humans as well as computers! =] alonzotg: what're you making it with? The prototype will be written with NASM Debugging is my biggest problem. when your'e able to access the serial port then hook a observing computer at it * Fare/#Tunes is back briefly gakuk =\ 07:50pm The serial port won't be accessable untill long after the kernel has gone into multithreaded mode.... =\ yes the prototype will be multithreaded/multitasking in realmode... =P -:- SignOff hcf: #TUNES (Ping timeout for hcf[me-portland-us317.javanet.com]) you should make a low-level debugging interface which uses e.g. the serial port for controlling and tracing the beast -:- hcf [nef@me-portland-us317.javanet.com] has joined #tunes http://tph.tuwien.ac.at/~oemer/qc/ <-- procedural language for quantum computing http://www.research.digital.com/SRC/personal/najork/cube.html <-- three dimensional visual programming language why not four-dimensional? or odor-based? don't ask me ;) 08:00pm olfactive programming language, anyone? -:- SignOff hcf: #TUNES (Ping timeout for hcf[me-portland-us317.javanet.com]) or maybe a touch-based programming language, for the blind? * Fare/#Tunes proposes a fractal-dimensional dimension 3.1415942424242 language. yes! 3.not quite pi? -:- hcf [nef@me-portland-us1017.javanet.com] has joined #tunes LOL ult Why base languages on geometry and not literature? Next language will be in verse, and will only allow quotes of great classics of american (no, persian) literature. (and comments of these quotes) ult: exactly. why not base languages on thinking so you have never have to write any codeline ;-) why have languages? Just _demand_ results from government! yes because you have to get from thought to electrical signals in the system why problems when solutions are so much better!!! and by the way thinking is slow action is faster hehe no action w/out thought ;P Amiga is an action based design rares: after the demands, it becomes the government's business, so I don't care anymore. think once act many Amiga _is_? most PC's think many act badly isn't any sentence beginning with "Amiga _is_" somehow preposterous? yes Amiga is very intuititively designed yes I've also played alot action-games on my old amiga... I once met an _intelligent_ PC. It saw it ran OS/2, so it committed suicide. there's no calculating X+Y positions in video memory the thing actually follows the video gun and fires after a delay Amiga is VERY unix-like designed calculating the delay is light years faster than calculating position OS/2 the is the Unix version what NT 4.0 is Windows NT OS/2 was the full name of a MS-IBM cooperative project then MS Microshafted IBM royally oh? i didn't know it started out as a collaboration wasn't NT designed by the same guy who did VMS ? 08:10pm Dude I just bought An Amiga 1000 and I plan using it extensively for real work it runs linux and contrary to pupular belief it's still in development just not by Commodore who develops it futre NT diverged toward VMS after the break up OS/2 diverged toward Unix -:- witten [witten@un.torsion.org] has joined #tunes I forget but there's a new design out soon called the BOxer MS have bought the whole VMS - developer team from DEC and built there Windoofs NT out of it (this is why WIndows NT was first available for ALPHAS) SVGA capable and ready to kick ass heh umm no ill believe it when i see it ;P Microsoft dragged their feet before they went Alpha and then they dumped alpha it's in any Anti-MS publication you can find at BarnesandNoble nope they dumped alpha only recently right.. Lightwave 5.5 ran on Alpha so did Photoshop 5.0 not well thoug h lots of bugs, crashiness not well though... pfffft this is MS we're talking about even by MS standards I donno if we need another conservative OS That's why Unix owns the Alpha market :) though the buyout of DEC and the end of VAX boxes had a bit to do with Unix's gain -:- ToHeB [iii@usr53.atil.net] has joined #tunes om I need objectpersistence, network-transparence, high-level-languages. Muli-Interfaces, no distinction between control and programming languages.... no files, no directories um you have to have files at some point but a cleaner system is [possible why do you need files???? how do you differentiat between content otherwise redronin: welcome to the Tunes project heh redronin: did you read the Glossary on orthogonal persistence? files are a system thought out when the first persistent memories came out. On computerplatforms with little ram an poor processing capability sorry I'm going the way of Amiga make things fast and accessible to human beings at its core Fare: I have overflied it - but I will read it later :) I think going toward databases is a way to cut corners but it fails in the case easy to build systems from scratch rares:donno what you mean 08:20pm im hungry processing is slow and stupid... it's good old fashioned unsuccessful artificial intelligence tunes -- make me a sandwich ow om my os is soo elegantly beautiful! =) future: thats it ;) look at the stiuff they do now as opposed to thewaste they produced 20 years ahgo -:- SignOff ToHeB: #TUNES (Read error to ToHeB[usr53.atil.net]: Connection reset by peer) rares, who? AI farts ahh 80's AI load of reules rules should come out of a fucking model the model is reality future: what flavor? it's like if hour = 1 printf("1"); when printf("%d", hour); is so much more erlgant (your sandwich) Fare, oh i'd take a turkey club if you have it do you get whjat I mean yah rares I think dbs have similar limitations they define structures that can't be changed there's much more freedom with files rares, what you seem to be describing is just the idea of compression, or pattern-recognition rares: don't understand me wrong - I've also a Amiga history, from A500 over A1200 up to A4000, programming all from Agnus till Paula - but time changes and so the Amigas Design is so outdated like the whole bunch of other drek from Windoofs till UNIX Amigas Design at the same speed as a PC vaporises the PC just ask Downix downix? what do you have with your dbs's HAM is the only thing that is superflous HAM is shitty at scrolling -:- mibin [mibin@an1-655.tiscalinet.it] has joined #tunes Downix usually visits tunes but he's MIA lately I know it was a great hack but still a hack with HAM you have bad color-erros while scrolling And yes you can get An Amiga at the same speed as a PC try moving your Netscape window in linux at 150Mhz would you say that SGI's have incorporated alot the amiga ideas into a modern system? it tears and the graphics redraw alowly yes, you've only to plugin one of those PPC hacks into your machine rares, that's a problem with X and the toolkite layers... Put linux on a comparable chip in an Amiga and it flies like you wouldn't believe NOpe hrmm * AlonzoTG/#tunes leaps into bed... -:- AlonzoTG [Alonzo@209-122-212-191.s572.tnt7.lnhva.md.dialup.rcn.com] has left #tunes [] Same OS on an Amiga same toolkit layers i have a P400 256MB and moving windows in Be is seamless, while in X it is choppy faster about 20x It's smooooooth on an Amiga at 150Mhz with 8MB ram haha and 2mb video Hardware is irrelevant - it changes day by day - why don't think more on better software!!!! 08:30pm cuz software is severely handicapped by fucked up hardware designs from 20 years ago (not Amiga, but ISM) just look at the 1024 cylinder crap with BIOSes even now it's fucked up ISM =IBm 1024 cyllinders should be enough for anybody how bout having only four primary disk partitions? redronin: well said or the shitty 4Gig problems in AMiga OS ;) witten the translation crap slows your performance down Is HAM for your SANDWICH ? haha if it's arranged as it really exists on disk it's considerably fasster i wasn't serious um 4Gig at that time was great I mean come on the 1st 32-bit OS was Amiga OS in 1985 1024 cylinders is like 8 gigs future only in LBA mode LBA mode slows the system oh rares: u sure there wasnt a 32bit nix back then? and I hasvethe source for Amiga OS 3.x quite fixable to go 64Gig if I tried Amiga OS is very unixish it's just that it is better than Unix why was amigaos abandoned? because the hardware is very well designed and intelligent unix is not about hardware because Amiga needed a marketing scheme and because AOS is tied to the Amiga hardware AmigaOS was Now there is Amix AmigaUNix amUn1X in fact there was a secret dual boot setup few people noticed because commodore flopped before they could announce it every amiga released in the late years had 2 Oses on it I've spent 1 Month in fiddling out the different AGA-GFx registers of my ol' A1200 to code some trendy Fx (the new AGA CS was undocumented) then after I had found out which regsiters does a magazine came out which contents a complete description. From this day on I never ever hacked the hardware - Hardware IS _irrelevant_ rares? What OSes? AmigaOS and what else? ult Amix I ' hated the AMigaOs for not having real VMem and Memory-protection Amix? WTF? Amiga Unix How do u get into it? I don't know... Downix will be on tomorrow to talk about it if you want to ask him -:- witten [witten@un.torsion.org] has left #tunes [] he just showed up a second ago to make sure his irc reg didn't exp 08:40pm A OS should not be bound so tight to a particular HW like the AmigaOS was [23:32:56] Downix [down@d-gnaps-74.ici.net] has quit irc : [[BX] Silly faggot! mIRC is for kids!] redronin all OSes are bound to their hardware at the compiled stage second AmigaOS being in ASM is perfectly okay because the ASM was high-lkvel the hardware design was so good rares: U don't understand me - the design of a OS should not bound so thight on a Part. HW wait u guys need to wake up and admit that windows is the superior OS cuz its still around :) I know exactly what you are talkin ng about original Linux had the same faults gottit umm bottom feeding lawyers have been around for thousands of years they are NOT a good OS now for Amiga tying to hgardware was a perfectly good thing the Hardware today gets more and more dependent on the fucking CPU I want Amiga designs my favorite os is the EniacOS A next generation OS should be able to run processes on different machines in a network transparently some guy lost his WinMOdem connection because his 3-D game overpowered the CPU redronin, amoebaOS ? redronin Amiga can do that that's the point future: for example you can have 6 diff types of processors in the machine Amiga adapts easily future: but it should not be restricted to run distributes redronin, http://www.cs.vu.nl/pub/amoeba/amoeba.html -:- ult_ [noone@user-37kbamb.dialup.mindspring.com] has joined #Tunes the 680x0 becomes a control chip only if you add other CPUs -:- SignOff ult: #TUNES (Read error to ult[user-37kbas5.dialup.mindspring.com]: Connection reset by peer) Amiga adapts to nothing beside of 68k, DENISE, AGNUS, PAULA... franly I'm kicking myself for ever going PC -:- ult_ is now known as ult you've never seen a PPC powered Amiga? -:- SignOff Fare: #TUNES (Ping timeout for Fare[esmeralda.enst.fr]) 08:50pm rares: I've worked with Amiga, NetBSD, Linux, HPUX, SunOS, Winshit.... on so many different Hardware platforms - there is really no difference at all. I don't know about that rares: I've also seen nice O2s at the US-Army Helicopter simulation center - so every AmigaPPC looks like an egg against this and this machines existed before phase5 came up with the idea of a PPC-Amiga -:- Fare [rideaufr@esmeralda.enst.fr] has joined #Tunes what does O2 do hardware wise because Amiga is a radically diff design from the market ruling design Amiga is so fluid compared to PCs I'm having trouble imagining a better design beyond tons of RAM and Space Hum. rares: support for security, multiple usres, orthogonal persistence, gc, etc What I try to say is, when you look good enough you will ever find a better HW for a particular problem. rares: ever tried a lisp machine? redronin there's 2nd generation better and then there's whole new world better the latter is very difficult to do Fare never had the pleasure O2 is hardware or OS? rares: ever tried playing quake or forsaken or another 3d-thing on a actual 3d-accelerator on a stupid cheap f*#ing PC hardware - you will learn how fluid gaming can be ;-) that';s not what I mean by fluid it's something else O2 as hardware is also irrelevant 'cause IF you LOOK you will find EVER a BETTER HW for your particular purpose 09:00pm redronin I already see Intel scrambling because they're not getting results from going smaller hardware is _IRRELEVANT_ I mean seriously revolutionary design not a refinement take a toaster if you wish to make a toast!!!! you don't get it there hasn't been any innovation so hardware beingbetter is actually a lie i'm not talking about speed you know what the thing is rares: I'am no INTEL, or SGI or AMIGA or enthusiast - I want to run MY software on whatever PLatform I find - nothing else * Fare/#Tunes ^5's redronin okay I'll bite I'm a designer at heart but I learned programming first ands that's my curse I also want general purpose stuff that's why I like files they are easier to edit than special purpose DBs but tell mme what is orthogonal persistence? rares: Iam studing computer-science at the university of applied science in Nürnberg and are working at a software house as programmer - most time I cannot choose what platform I have to use: but so I've found out that what people want is my thoughts burned into bits 'n bytes on whatever is the hype of the time redronin I feel your pain in that case I've been forced to use Visual Studio when Unix tools would have taken 10x lessd time so teach me what is orthogonal persistence? * rares/#tunes listens :) orthogonal persistence is gedankenware as Fare might say 09:10pm okay in English ... heh, i have no clue. check out the tunes site :P Fare hey... you still here? no, I'm almost asleep rares: Yes sometimes it's pain to use something special (like using ORACLE Dev2000 Reports btw) but what counts is your thoughts not the tool you use: I've coded CommonLisp, Scheme, 68k, MIPS x86 ASM, C/C++, SQL,PLSQL,VisualBasic, different Unix shells, Java but there also is: what counts is the thought not the syntax slashdot leeps telling me it downloaded 100% of 42 rares: check the glossary about orth. persist. papers/Glossary/#orthogonal_persistence Is Gedankenware really a common expression in english? (asking as interested german ;-) redronin the thing i I've studied hierarchial systems as an obsessive hobby and I even take ideas out eastern mysticism I've narrowed it down to these rules: You have an active element you have a passive element and you have a carrier element all of those can be further sub divided into three pieces redronin, nope. i think i just made it up. but you can see gedanken at: http://www.it.com.au/jargon/gedanken.html the same functions so I think in terms of files because you can store everything from ideas to code to ASM to shopping lists in them the hierarchy for files is completely reducible rares: change "files" with "objects" and I would take it with orthogonal persistence, you shouldn't have to worry about storing objects at all I'dont wont to save my shopping list IN a FILE. I only want to make sure that I have it when I turn my computer on and wnat to look at it the problem files fit the bill for something both software object (it's contents) and hardware object (something accessible and readable) why disinct software and hardware in an OS in the upper levels 09:20pm I don't care about the upper lebve;ls I totally agree in the upper levels all things are objects (sounds funny) isn't orthogonal persistence just the same thing as JOURNALING? see the problem is I get more done when I think like the computer than when I think like a human actually objectdesign made a c++ compiler where all the objects persisted in a TRANSPARENT way that's what an object-database does Humans are so isolated and oblivious and inconsistent sometimes doesn't journaling just mean storing info about filesystem accesses or something? future: no; journaling is just one way to achieve _safe_ persistence; be it orthogonal or not. objects are no "human-way" objects are "world-way" and computers are only objects *G* I see everything in an elegant model where as human terms are very superficial and limited that's my tale from the hard side :) safe against hardware/power failures well journaling is just the idea that bits on the FS are always in sync and modifications are implemented as new 'records' in the fs journal. kind-of like an implicit versioning system _orthogonal_ persistence doesn't meddle well with _files_, you know. yes yes they are opposite to one another object databases implement this object databases suck also, checkout http://www.freshmeat.net/appindex/1999/08/26/935676498.html they are codeless data plus, they use languages that SUCK what if the entire data model was an object db? that wouldnt be so bad. and would probably give you all the orthogonal persistence ColdStore is a gigabyte-scale persistent object store which mmap()s objects into a file (for persistence), (from the link above) I would think of a orthogonal-persistence optimized direkt-disk-access system similiar to virtual-mem future I'll buy fully object systems when computers become light based difraction/recfraction computers then I'll drop my files ideas cuz then a fucking ball will be a fucking ball :) in practical terms, what we want is to be able to shut off a machine, then turn it on and preserve the state yes? there's something called 'checkpoint' which lets you save the state of the whole machine. i think it can be done on NT and some UNIXes. kinda like turning off the light then back on yah.. its an inredibly awesome idea but what happens if you shut off a network connection? when it is 'revived' it will get all screwed up I totally agree with you guys if you can build hardware than can use objects at its core until then files are my thing that didn't come out riught a file extension is like an object property. but that's all you can do with files. and trom the handling side there are no files where I have to store my transistent objects to but I only create e.g. a "shopping list"-object fill it out and go away. 09:30pm right.. and then you can easily search for an item in your shopping list or compare it with previous shopping lists or automatically remove items if they become too expensive the thing is I see the computer as inherently a simulation of reality until better more physical tech exists (and fully accessible to human beings of course) so I prefer to learn the way the machine works cuz then I can build a better design look at corba for example: you create distributed objects only distinct through their object-reference. When you will manage to find them again you will have to use a directory-service like Nameservice or Tradingservice to register the referece to it. With a Namingservice you bind a Name with the Reference and with the Trading service you give the objects some properties (e.g. age, type, location name and so on) why don't computers fucking have TRAM ? for _efficient_ safe persistence so when you will use an object you only query a namingservice for its name or a trading service for all objects with a particular set of properties (battery-backed buffers for disk) is TRAM real? without tram, how would you know when it was safe to take a *snapshot* of the current state? hrmm. if the program was deterministic i guess it wouldn't matter.. future: it's easy to build from current technology, but no one does it and no one will, as long as current proprietary OSes dominate the market see that's just it all you need is a shape(set of arguments) a set of behaviours(general instruction of what to do with the damn arguments) and a style(to differentiate stupid designs from good desiogns for identical objects) to makean object unique. That way no name serbvice is required. You could conceivably reduce the thing to one object where the shape style and behaviours are the params Fare: make it so! lol eihrul what we can further think of is abstracting processing to processing-objects and any kind of memory as storage-objects. This storage and processing objects can be distributed and have properties like the speed and range or capacity or persistent/transistent.... rares, sometimes you have different implementations of the same object 09:40pm I hope Downix' disk-equipped systems have TRAM kukag what's the difference between a processing and storage object? gota go there shouldn't be one, afaik :P is downix real? nothing a process has to be stored sooner or later yes Downix is real his computer died is this a new os? that's why he's been missing Downix is a hiuman bewing silly oh I hope Downix' disk-equipped systems have TRAM a Unix addict like myself -:- SignOff thomas: #TUNES (Ping timeout for thomas[193.217.63.152]) im still hungry process-objects are abstractions of a "processing-unit" (like a CPU). A programm can choose which of the available process-objects it should use for computing through through searching the needed processing-properties * future/#tunes bbl -- store24 redronin that's what I justr said see that's just it all you need is a shape(set of arguments) a set of behaviours(general instruction of what to do with the damn arguments) and a style(to differentiate stupid designs from good desiogns for identical objects) to makean object unique. That way no name serbvice is required. You could conceivably reduce the thing to one object where the shape style and behaviours are the params redronin: gah, i don't think the separation of the two is useful redronin: having just objects is less complex :P eihrul: sorry the existence of storage and processing objects INHERITED from a abstract or (possibly) concrete base-class seems valid to me they are both "object" of course it think it's no more than raster versus vector ways of thinking processing objects are vector drawings storage is raster redronin: you don't need the separation for system to function though :) eihrul: eihrul: where do you see a separation?????????ßß union object { 09:50pm processing stuff; storage thing; } an object system doesn't make any sense to me if there is only one (a, 1, uno) Object Object-class i mean * eihrul/#tunes shrugs.... classes suck anyway :) redronin what I mean by that ios that in the world of a computer you could have a melting pot of behaviours called theObject then when you have a problem to solve the system creates your specialized object for you from an alaysis of your problem and proceeds to execute it the key is to be able to quickly match problem pieces to available objects er behaviours Of course, of course. But what I tried to explain was an OS intern mechanism that makes it available to distribute processing (should I better say computing?) and storage of Objects of any kind so you mean a Beowulf collective Object gah there'sa only so far i'll go before I'd rather be playing my guitar in the sunset with a woman to keep me company so the OS can decide what kind of processing unit or storage unit is needed for a particular purpose like persistent storing of another object redronin justr remember to keep the thinking to a minimum it slows every system to a crawl I donno but do you know what your machine can compute (and is computing!!!) every second!!!! thinking is just as bad for computers handling millions of web queries as it is for humans observing the road at 70 MPH on the highways It's the task of every OS to e.g schedule processes if you try to think about your surroundings on the highway you will crash guaranteed well in some instances that's all theOS saghould be doing gah Ican't type 10:00pm the OS would not "think" so much 'cause you describe in your programm what you want from it (like the creation of a object with the "persistent" property) and the OS KNOWS what to do!!! It drives me nuts that we have the tech to make a very good robust system but the markets are so oblivious to the science we get stuck w/ sghit redronin exctly the problem is everything about PC hardware forces OSes to think the purpose of a OS is that it thinks! the problem is that the Amiga design makes most of the thinking unnecessary that's why it's fast think about how compilers reduce the number of instructions required to do something when optimised same idea Think of how much more a Un*x system has to think 'cause it has so much more OS-features like dynamic-priority scheduling of processes, memory-protection and much much mre that's whyexactly oops exactly even the key is to be able to quickly match problem pieces to available objects Amiga Os is not fast 'cause of it's superintelligent design but 'cause of the lack of functionality ^^^ how do you *define* a problem? redronin Amiga OS doesn't have as much device work to do because of the superior design of the hardware I argue it would still blow unix away if it had menmory management have spoken upon device work? ?? confused rephrase? And I can say to you that it will look like an egg as a super-high-load FTP-server against FreeBSD sorry I meant do i have spoken upon device work? 10:10pm rares, you said match problems to objects... how does a person define a problem so that it can be matched.. the shape first future the shape of the object is the parameters without using declarative languages (and here are we again at the AI-Stuff) the shape of the problem isd the parameters wouldn't lots of objects match a given set of parameters? rares describe me the parameters of an object that computes a faculty! here's n example: wait let me think ... :) -:- water [water@tnt-10-119.tscnet.net] has joined #tunes oh boy hey water you're missing out on the fun hm same crowd what fun is that? -:- future is now known as present (defun faculty (n res) (if (zerop n) res (faculty (1- n) (* n res))) some guy named paast complained i had his nick heh rares: heres your Object -:- SignOff rares: #TUNES (Ping timeout for rares[nwhn-sh14-port164.snet.net]) coward ;-) what was he referring to, anyway? -:- present is now known as future okay, quit shifting time around like that, please ;) hehe. the inexorable flow of time cannot be stopped by man i have enough problems just living in a universe with general relativity and quantum state 10:20pm * water/#tunes is no mere man he mentioned that the OS could - given an "problem"-object can decide what to do LOL yeah sure and I asked he should give me an object describing a faculty problem :) heh yeah that makes sense -:- rares [rares@nwhn-sh14-port164.snet.net] has joined #tunes shit reeeeeee hehe hey rares where's your problem ( your problem-object i mean ;-) okay let's say you have a circle you have a radius thats no problem for me *g* you have the circumference * water/#tunes is expecting some lisp-level stuff here water shhh :) rares: (continue) the thing is it's an anonym,ous object not even the name circle to it just circ=2pir (how is it distringuished from a sphere or cylinder) future you shhh tooo :) future: very good question yeah anyway your problem object has the following a circumference memeber empty and a radius = some int if the objects are only distinct through their properties than you have that problem future mentions no I don't surface area is not circumference different object by the way in this system how do you locate it without unique-identifier (actually you can enforce a principle of monotonicity to prevent cylinder/sphere) okay here's the story let's say you had a sphere maybe it doesn't matter if it's a circle or a sphere. just so long as the problem the thing specifies gets solved. now shhhhhhhhhhhhhh EXACTLY thank you future 10:30pm spherecontains circle within it see? no naming just knowledge just bullshit, too aND THE BEST PART if you include both values in the problem you can get a true or false from the system water enlighten me... :) sigh i can accept specifying an algebraic problem, or a constraints problem. but how do real-world problems map to this? name one i assume you've looked at lisp's constraint solving issues (?) oh.. prolog.. this probably works for such a prototype-OO example but will be much more difficult when describing algorithms which havent mauch to do with OO at all (like the faculty of a number) future first then redronin gimme examples water: no we haven't talked of lisp at all, we talked upon topics what a "next-generation" Os should look like c'mon where's the examples :) redronin: :P rares: what example do you want??? you said something like the faculty of a number ok lets explain do you know the lambda calculus of Church? I haven't studied it yet but I think I can understand it (I learned nothing for 3 yrs thanks to winblows) ok you wanted an example for a "problem-object" for the faculty yeah even winblows has functional programming languages on it -:- mibin [mibin@an1-655.tiscalinet.it] has left #tunes [] you have one input parameter "n" water no windows discussion group I eve went to ever led to this kind pof stuff I;''m talking about the general windows culture oh trash cs culture ok, here's your problem: find a record in a database according to some search parameters and display it on the screen -:- lar1 [larman@2Cust47.tnt26.sfo3.da.uu.net] has joined #tunes then it's even easier in lisp you can say : (defun faculty (n res) (if (zerop n) res (faculty (1- n) (* n res))) 10:40pm this is about as far removed from the idealized metaphors of functions/constraints/algebra as i can think of factoiring listen to read red, even c'mon go on water said listen so I'm listening ah I should go further? if there's anything else yes nite -:- ult [noone@user-37kbamb.dialup.mindspring.com] has left #Tunes [] anything else? the function-definition above isn't anything else then a object of type function that has 1 param and solves the faculty so what follows is, that if you want to parametrize your "problem-object" you need to pack in all information for computation of your problem too therefore it's nothing else then a algorithm-object or a function-object or so. -:- SignOff water: #TUNES (Ping timeout for water[tnt-10-119.tscnet.net]) -:- water [water@tnt-10-209.tscnet.net] has joined #tunes reeee bleh -:- SignOff rares: #TUNES (Ping timeout for rares[nwhn-sh14-port164.snet.net]) understand what I want to say? you're saying the problem can only be defined by its solution :P (uniquely defined) If I want to solve the Problem I've to manage that the system has all information to solve it when there is a faculty function in the system my Problem describes as e.g (faculty 5) if not I've to give the system the solution in form of a lambda -----BEGIN PGP MESSAGE----- Version: PGPfreeware 6.5.2 for non-commercial use qANQR1DBwU4DV13AakCZXa0QCACaax7Kl5afIbL/NdCbyyTlM8nQrYcXAFoVOwOk PAPNS2rU7F87qsXkOKYEAH+Q5cDmcihrOv1uhoquu2NbAB/icysptmQx6U0+kBrJ 7PLvooannGUKc9iaSI1uXoITl6wb/2mtc7gg3gdjriobLlEv+lR3fNBoPBH+1Bpd qMdMogOiLe76j4SvOYJLWg+XG9oE7xFynTv1iNYbcLwjosF66z7zPve657JKT0ge FVATfzhAuftfaDcsfqiKxmyHuBITrMSjPE+wIfhNDvZY09rcj+9z77FHK4Y+WyGa OcewxR4mbHVnKaMMeubVt7+8iwgwagEhsNaEXtKoDR88aDLRB/95UKPLq9iksECT 10:50pm MxYO0TUoUc5Sa sorry didn't know that did that... -:- rares [rares@nwhn-sh14-port164.snet.net] has joined #tunes what the hell wtf was that? you need to encrypt to irc? I kerep falling off the edge of irc No, I was just testing a cool encrypt current window rares: should I repeat the conclusion? function and I guess it decided to paste too well talk to abi then I didn't know that it would dump the ouput to the channel, sorry np red repeat it all sorry heh u can check the irc log ok quiet please I repeat all from after your logoff understand what I want to say? [07:53:16] you're saying the problem can only be defined by its solution :P [07:53:34] (uniquely defined) [07:54:07] If I want to solve the Problem I've to manage that the system has all information to solve it [07:54:45] when there is a faculty function in the system my Problem describes as e.g (faculty 5) [07:55:13] if not I've to give the system the solution in form of a lambda understand what I want to say? [07:53:16] you're saying the problem can only be defined by its solution :P [07:53:34] (uniquely defined) [07:54:07] If I want to solve the Problem I've to manage that the system has all information to solve it [07:54:45] when there is a faculty function in the system my Problem describes as e.g (faculty 5) [07:55:13] if not I've to give the system the solution in form of a lambda oh sorry (2times) wtf? 1 flood deserves another?! why don't you just tell him to look at the friggin logs?!? this is what is meant by orthogonal persistence are there any other logs? he has no logs if he was offline abi: logs? hcf: i haven't a clue grr abi: tunes logs? bugger all, i dunno, hcf abi: #tunes logs? i haven't a clue, hcf fucking a www.tunes.org/files/irc axo -:- SignOff rares: #TUNES (Ping timeout for rares[nwhn-sh14-port164.snet.net]) so water do you have any interesting urls for us to read about language design depends on what you need to learn oh i need to learn everything -:- rares [rares@nwhn-sh14-port164.snet.net] has joined #tunes http://www.cs.cmu.edu/afs/cs.cmu.edu/user/mleone/web/language-research.html i think il get 'haskell: the art of functional programming' rares: crappy connection ahh i have that bookmarked redronin my ISP didn't drop me it was irc that dropped me use a better server Use a different server, perhaps A very good book is also SICP (structured interpretation of computer programs) redronin, a book on compiler design? yes -:- SignOff rares: #TUNES (Changing server...) -:- rares [rares@nwhn-sh14-port164.snet.net] has joined #tunes here's my idea when you have a long list of parameters and you end up with not enough values a system with no object names can go through easily to solve problems till it gets rid of all unkknowns if you have names it can't deal future:in SICP is among others the implementation of a scheme-interpreter an compiler -> yes 11:00pm think for a minute how the human mind works when you solve a complex problem rares, redronin mentioned that by defining a 'problem' you've already in essence pointed your finger at the solution rares: ha this was good or maybe i misinterpreted future:exactly redronin that's what I'm saying future:you're right solutions do not exist independent of problems :P it is an absurdity future redronin I think we agree then mompl listen * rares/#tunes listens -:- SignOff water: #TUNES (Ping timeout for water[tnt-10-209.tscnet.net]) -:- water_ [water@tnt-9-1.tscnet.net] has joined #tunes given the problem a = 5+1 you can do following in lambda calculus theorem of church damn win32 rares: you are wrong about named objects -:- water_ is now known as water Problem is: (+ 5 1) okay... go on do a Beta-Abstraction (+ 5 1) => ((lambda x . + x 1) 5) umm spell that out I'm not familiar w/ the terms but I do get it so you have abstracted from the "problem" (+ x 1) is more abstract than (+ 5 1) right (+ x y) more abstract with another step you can abstract to (+ x y) exactly and the next step is (op x y) only thast's too far but the information of what to system should do was also available in (+ 5 1) rares: why is that too far? :) heh the systems knows "+" as primitive function and it knows the constants 5 and 1 in real-life, you don't have the symbol '+' to neatly describe the operation... the question is how do you express '+' hen you don't have it "don't go too far, you'll fall over the edge of the abstraction world" eihrul i'm being cautious that's all my brain gets queasy about rthat point redronin so far you've said what I've been saying 11:10pm sigh rares: (op x y) is not to far, its simply (lambda x . (lambda y . expr)) red like i said my brain raises a red flag that's all as a caution np I've been know to go to the point of (op xy) -> (something) and get a headache w/o realizing it :) eihrul: i wonder if rares could handle slate ;) don't see why not... water: what would you say to this topic? higher-order fp mixed with pure oo? hm do you have experiences with declarative languages? rares should read sicp moi? consider this problem: 'i am thirsty'. how do you solve that? if the system knows that 'i have a cow', then the system can deduce that it needs to GetMilkFromCow() and so forth.. * future/#tunes says moo future that's a nested problem silly "nested"? i guess this is what prolog does with backtracking thirsy points to drink needed milk points to drink is rares, but real-world problems are typically combinations of other problems but the system also have to know: thirsty->drink , drink needs water,milk, cow has milk good god water what's the problem you look scared i don't suppose anyone here has looked at cyc then cyc? cyc.. is that some prolog connectionist thingy? huh nevermind abi: cyc? i think cyc is at http://www.cyc.com and for "non-real-wrold" problems read sicp Cyc performs best-first search over proof-space using a set of proprietary heuristics, and uses microtheories to optimize inferencing by restricting search domains. umm.. backtracking and constraints-solving ? water: how do you come to the conclusion that no on of us here has never looked at cyc??? * eihrul/#tunes ponders... 1 million processors... because the cyc fellows discovered something important in trying to compile their database of necessary info for solving "real world" problems * eihrul/#tunes boggles. 11:20pm they discovered something similar to what the 5gl japanese researchers did what? the hard-ai barrier?! you can't solve any real-world problems with these kinds of questions with what kind of questions? the kinds of questions you fellows have been trying to illustrate -:- SignOff rares: #TUNES (Ping timeout for rares[nwhn-sh14-port164.snet.net]) rares has some serious irc problems to illustrate ...? forget it. i'm wrong so they came up with a better way to phrase problems? no they didn't they just discovered how little they knew about real-world problems they didn't phrase problems they phrased wisdom, or better "common sense" just shush i'm sick of this crap in the channel hah water is pissed off When is he not? everyone trying to re-discovered what's been written in well-known books for decades oops, I didn't say that it's so frustrating water, but like all good philosophical quandaries there are never any good answers thanks for the wisdom future ;P don't you guys read?!? talking to people on irc has to be the worst way to learn anything I don't understand most of those ten millon page books not that I have access to many lar1: dude you're 14 water: I'm a little upset with your conclusion that I wouldn't read? although I haven't really perused them you have a lot to learn start with the stuff just within your reach first lar: it's pretty much a waste of your time to be in this channel water: Things like forth, lisp, etc? sure, among other things like how to actually write big programs -:- rares [rares@nwhn-sh2-port204.snet.net] has joined #tunes water I'm almost sure I know what you mean 11:30pm water: rather naive thing to say... lar: you know, programs larger than 10K lines ehm.. am I included in your opinion of "cannot have read enough when he says what he says here" or what? water: What kind of book would cover that? lar1, theres no substitue for experience :P yeah just start hacking away -:- SignOff rares: #TUNES (Ping timeout for rares[nwhn-sh2-port204.snet.net]) oops lar1: be open minded, learn from listening, learn from observing, learn from reading books and not at least learn at school what a waste of time water: what do you mean? * water/#tunes points to the topic Ah.. you mean TUNES is a waste of time? grrrr no, but #tunes ceertainly can be at times -:- rares [rares@wtrb-sh1-port54.snet.net] has joined #tunes just joking! -:- lar1 has changed the topic on channel #tunes to: what a waste of time hey... * water/#tunes whacks lar1 with a Big Clue Stick -:- water has changed the topic on channel #tunes to: TUNES, Free Reflective Computing System @ http://www.tunes.org || slate @ http://www.tunes.org/~water/slate-home.html let me try again There wasn't a topic set before, right? yes there was rares:whar At least my client didn't show one... -:- hcf has changed the topic on channel #tunes to: slate @ http://www.tunes.org/~water/slate-home.html -:- Downix [down@d-gnaps-3.ici.net] has joined #tunes lar1: then you have a problem with your client - get the source and search the problem ;) screw the rest, just talk slate eihrul: how much can you do with eval.lisp without a major rewrite? well, it's < 100 lines of code it's not like a major rewrite is different from an addendum :) ok so even major would be minor -:- SignOff rares: #TUNES (Ping timeout for rares[wtrb-sh1-port54.snet.net]) so how much of slate could you support with it? 11:40pm * eihrul/#tunes ponders. well, with that current code, i don't want to bother :) why not? i first want to rewrite parts of it to support proper meta-objects oh one of the other signifigant differences would be provisions for lazy evaluation hm although it's just a message-send namespace stack, etc mmhmm so, for all that, just as easy to rewrite it :) so when can i expect the new version? ;) well, to support the : operator is tricky... because i can't use lisp lists anymore :) why not? because ':' is not a symbol... it qualifies the keyword package... so use another one though, and how is a sexp represented as an object as well? :) actually... hmmm that might not yet be relevent to the implementation an object with a 'car' and 'cdr' slots, silly well, before you were mentioned (...) as a single object... not a structure of objects s/you were/you structure of objects? -:- rares [rares@wtrb-sh4-port139.snet.net] has joined #tunes a list okay let me try again ://// water: re: cyc is that about 3 body problem being unsolvable by a Turing manchine? hm but ( and ) are also namespace operators rares: n/m in fact, that's they're primary point s/they're/their/ so that's not what they discovered? i think the sexp view of an object is only auxiliary water: they do two things :) well the sexp view of an object is ui-related both namespace operators and message send operators 11:50pm i don't like that * eihrul/#tunes nods. a clean separation of the two could be achieved at the worst :) i don't see why we would need separate message-send grouping it's all based on namespaces well, it allows you to specify when a namespace is relevent... huh? rmm, or rather not to have to mess with irrelevent namespaces :) look at the example on the semantics page the parentheses only need to work namespaces hm even in smalltalk/self, viewing objects as text structures doesn't provide any use other than ui * eihrul/#tunes nods. though, how to provide text :) :P why using especially text - you could also use other things to reference the objects damn i'm finding it hard to concentrate e.g graphic-objects yes we definitely want independence of interface -:- SignOff rares: #TUNES (Ping timeout for rares[wtrb-sh4-port139.snet.net]) [msg(TUNES)] newlog 2000.0220 IRC log ended Sun Feb 20 00:00:02 2000