IRC log started Mon Mar 13 00:00:01 2000 [msg(TUNES)] permlog 2000.0313 -:- lar1 [lar1@1Cust198.tnt1.sfo3.da.uu.net] has left #tunes [] -:- SignOff air: #TUNES (http://www.qzx.com/ :: sleep) -:- SignOff _ruiner_: #TUNES (destroy what destroys you) -:- water [water@tnt-9-165.tscnet.net] has joined #tunes ping 03:00am pong hey foof morning so it is... 3am here * water/#tunes is reading a finely-grained review of macro-processing wrt meta-programming 03:10am url? 03:20am -:- SignOff Fare: #TUNES (Ping timeout for Fare[esmeralda.enst.fr]) -:- Fare [rideaufr@esmeralda.enst.fr] has joined #Tunes -:- SignOff Fare: #TUNES (Ping timeout for Fare[esmeralda.enst.fr]) checkresearchindex for semantic and macros 04:50am url please 05:00am -:- Fare [rideaufr@esmeralda.enst.fr] has joined #Tunes sorry napping one sec Fufie: http://citeseer.nj.nec.com/cs?profile=6118272%2C259947%2C1%2C0.25%2CDownload&rd=http%3A//www.cs.berkeley.edu/%7Emaddox/papers/msrept/xlrept.ps grr 08:10am http://www.cs.berkeley.edu/~maddox/papers/msrept/xlrept.ps thanks you only had to search ri for a bit but np water: wotsit about? not finished with it yet but it goes into macros and meta-programming you definitely wouldn't be interested ;) actually it introduces a language called xl but the intro and page 30 onwards are the interesting parts it seems to be a discussion of his perceived flaws in the lispy macro system it stems from that it seems that he presents a schemish solution, but I'll read the whole thing later at least his naming is schemish he does mention reflective scheme as well as 3-lisp yes this is an open-ended paper 08:20am btw, eihrul's decided to stop working on slate in favor of dolphin :( what is dolphin? it's some component-based os idea that he, air, rares and lar are working on i think it's a useless project, but oh well i think tcn is in on it as well, which means less efffort (than zero?) on retro as long as they find it wothwhile it is worthwhile i suppose so i seem to be the only one trying to work on the hll at all I cannot do much until there is a base to hack code on anyway why not? because I work on development environment and a hypermedia-system which both need a base on to be useful oh that's what you're good at so until there is a base I am better off developing the dev-env and the hypermedia system in CL and then later port when things are ready who decides if the code stays? a good dev-env and a networking hypermedia-system is vital (imho) for a user-friendly and interactive system yeah but "good" is highly subjective naturally which is why i mention the decision process for dealing with code once you get slate defined and sortof running it'll be fun to add slate support to the system as long as you make it a descendant of morphic, i don't care morphic? yeah the self ui 08:30am (also the hacked-together squeak experimental offshoot) there are docs on the web about it we'll see.. I won't interfere on how things look and how they feel i actually don't care about look and feel too much morphic is about putting direct-manipulation metaphors in the center and designing around that sort of like hyperprogramming in napier88 abi UFO is United Functions with Objects at http://www.cs.man.ac.uk/arch/projects/ufo.html afk 08:40am it seems like there's a really fine distinction lately between expressive powers of various languages/systems one the one hand you have macros, and on the other you have more general shape-based ideas, which are less formalized and less complete right now -:- smoke [smoke@15dyn119.delft.casema.net] has joined #tunes hey smoke i suppose i could just settle for a system that is at least as good as lisp's macros 09:40am (common)lisp macros are just great as an initial system from where you can build other systems sure but that doesn't fit into slate it's great if you're starting from lisp the problem with them is that you don't know what environment they are run in, and can't communicate or let persistent data for other metaprogramming activities * water/#tunes nods that's why i'm looking into this nebulous field of shape transformation well, it's not nebulous, just scattered i'm even considering a kind of structural meta-object 09:50am but mostly i'm thinking of being able to formalize shape have you looked at the shape project stuff? they have a rather nice category theory based notion shape transformation? abi shape hmmm... shape is at http://linus.socs.uts.edu.au/~shape/shape.html more like rewriting being able to express lambdas over structure without complicated syntax reflectively, of course, as well i suppose one immediate benefit would be to get graph-rewriting techniques for free 10:00am -:- SignOff smkl: #TUNES (bbl) -:- water_ [water@tnt-9-193.tscnet.net] has joined #tunes -:- SignOff water: #TUNES (Ping timeout for water[tnt-9-165.tscnet.net]) -:- water_ is now known as water -:- AlonzoTG [Alonzo@216-164-133-219.s219.tnt3.lnhva.md.dialup.rcn.com] has joined #tunes hi at 11:00am om * water/#tunes is reading the tunes docs again 11:10am water: tell me about how to improve them well i can definitely tell you some stuff you've heard before but as for specific items it would be good to identify what features of what languages satisfy hll desired features and how they do it even just one example per feature would help like quotienting for example i can't think of a language that does what that describes btw, i'm most likely going to post a Slate-as-TunesHLL that item-by-item addresses your writing 11:20am not that it's complete by any means of course if i can get macro-level extensibility, then i can push features out to the libraries and oh yes there is a relatively straightforward translation of the macro feature of lisp into slate, but it crosses namespaces right now, so it's not a usable notion yet oh yeah there's a problem i have with the membership relation between objects are you listening? well i'll just ramble then :) oops anyway, i'd like to know if there are any more requirements for the membership relation other than that it exists how that "cross namespaces" ? membership relation? well the macros operate on code that generates object yup is that a problem? the object is the functional expression that takes the default object and modifies it into the full one yes because each of my objects is a namespace this expression,btw, can consist of things like "add-slot" and "clone" 11:30am not sure what where the problem is. which is how you get self-like objects as opposed to purely functional ones hm btw, maybe a pure interface rather than "add-slot" can be useful. wtf is a "pure interface"? I like to keep encapsulation and statefulness orthogonal why? pure, like in "pure objects". as in bullshit because they are orthogonal features that have simple algebraic specifications yes but what are they? is there a paper on this? encapsulation is about defining (partial or total) grammars for inductive types. so? statefulness is about defining interfaces to an implicit persistent store this is way too abstract theorems get much simpler if you decouple these fuck theorems "C++ is like a swiss army knife, with all blades burnt into open position" i already can compare my slate model with cardelli's object calculus don't do the same with your system yeah i've read that get a clue dude i'm solely conceerned about the how the language works describe what you mean in some real terms in terms of things i already know and use tunes terms suck, because no one uses them except you, and only when you're talking about tunes of course, you've stopped listening once again completely rude -:- SignOff AlonzoTG: #TUNES (Ping timeout for AlonzoTG[216-164-133-219.s219.tnt3.lnhva.md.dialup.rcn.com]) hum I want to be able to declare stateless objects you could at least let people know when you have to go away ok as in labelled tuples slate can give you that 11:40am and have no feature create an implicit state least there exist a corresponding stateless feature what? "no feature"? check your grammar please are you in a bad mood? no i don't grok what you just meant I mean that there shouldn't be any builtin feature that depends on stateful objects least there is a corresponding pure feature or maybe there can be purifying and statifying metaprograms/combinators ok slate provides that as well i don't know about the last clause there since i don't really know what you mean hm must all conversations with you have a built-in 5-minute lag? like, you can transform a 'a*'b -> 'a*'c into a 'a ref*'b -> 'c hm abstract reference or low-level pointer? hm the first one it can do already low-level stuff requires implementation support 11:50am unfortunately, i don't have the algebraic meta-system defined yet, so the specific example of multiplication i can't support yet fun f -> fun ar -> fun bv -> let av0 = !ar in let (av1,cv) = f (av0,bv) in ar := av1 ; cv ;; i just have this problem that i don't have grouping operators, but instead namespace stack operators namespace stack operators??? no it doesn't work like that in slate, not directly yes in order to bind arguments to functions you have to send messages which requires using the object's inner namespace to get the selectors so you're having a problem due to dynamic scoping? no it's not dynamic scoping the problem is that lisp macros all work in the same namespace (supposedly), and mine don't so the mapping isn't direct really, it's not that much of a design problem, it's a matter of getting lisp users to accept this but it does bring up something i wanted to discuss can't you select the namespace in which you define stuff? what sort of namespace meta-structure does tunes endorse if any? "lisp macros all work in the same namespace"? 12:00pm heh I'd like first-class namespaces yeah well in lisp all functions are essentially in the same namespace, no? not really yeah well in slate they definitely are CL-like one-level global namespace stuff is braindead, as far as distributed systems go that's what i was referring to CL has packages (should've been hierarchical) which are namespaces right but i've heard very bad things about packages in lisp not that i've tried to use them and within those namespaces it has a function-namespace which is where functions and DEFMACROs live.. lisp macros > DEFMACRO that's not the issue the issue is that right now i have made a temporary design decision to make the namespace/object relation hierarchical fare: shouldn't one really have a module concept instead of packages for a distributed system? and i find that i have to break this hierarchy in order to get a usable system sigh fare: CL packages are just namespaces (readtables) and little else while a distributed system needs a much more powerful mechanism in slate, modules are passed around as namespaces, which are only objects i haven't added any special features to slate modules yet umm.. does packages in slate serve the same function as in CL? ie in CL they are a tool for the reader-stage they're stricktly namespaces right now not for encapsulation or anything such well all objects in slate are namespaces then it is not the same concept at all including what's commonly referred to as methods most likely not packages in CL are just tables of symbols that the reader discovers yeah i call that a hack why is that a hack? because lisp already has lexical scoping that should be enough * Fufie/#tunes smiles well enlighten me, then read up on symbols in CL, they're not quite what you obviously think i doubt i can't do what you consider to be so amazing within slate in a more obvious way nothing in CL is obvious 12:10pm it's designed to make its gurus arrogant maybe CL is not what one thinks when one starts with CL, but is that necessarily bad? even people who mess with smalltalk metaclasses have humility most people start out to think that CL packages are like packages in java or modules.. they're not who cares? i don't want to hack the reader people who care for good language design and for creating useful languages usually care lol yes and that's why end users love lisp so much end users? oh that's right, end users are scum ;) like the programmers? heh no like end users your grandmother your dentist oh.. CL is not made for my grandmother or my dentist exactly it's made for serious programmers yeah people with hard-ons :) just like a scalpel is not made for my grandmother to slice her bread with yes but your grandmother must use that scalpel and its 500-page manual if she wants to do anything important with bread why must she do that? because otherwise she has to trust a professional I would prefer that she used a bread knife to slice her bread with so why do you criticize languages like BASIC? (rhetorical) BASIC serves no purpose at all.. I don't recommend it for any use oh really? so it's not a butterknife? it's a cheap plastic knife tell me if a computer's bread, then what's the butterknife? smalltalk? smalltalk might be the butterknife or python even those don't have simple ways of approaching them or even scheme even butterknives vary must one type in lambda to use a computer powerfully? when I was young one connected the word Amdahl with powerful computers ;) uh sure and what planet are you from? oh yeah you're from geekland * Fufie/#tunes smiles you love arcanity for its own sake do you always try to insult people? is that your raison d'etre? well i find that no one brings up any decent thoughts without it here so that gives you the right to insult? no one has rights 12:20pm this is irc and there is not OP so? heh any argument with you over more than five lines end in insults.. ok i've obviously violated some mild social custom a few minutes ago you tried insulting Fare sorry and now me yeah well you guys are stagnant wrt tunes maybe.. no one else has a clue where tunes is going because of fare and you definitely don't help sorry to say it but you grok quite a lot and add nothing maybe not, but whenever you ask questions I usually try to patiently answer them even if you dislike the answers i know but you have to get shaken up before you'll see any other point of view i go out of my way to prove i'm wrong every dday but i won't accept what you say just because you have common lisp terms to back them up you don't know me in RL, maybe I have another net persona maybe you ignore what's out there but I try to answer patiently maybe i do as well :) but tunes is deadly sserious to me i've lost a lot of sleep over the last six years trying to get my ideas working i take it personally, whether you like it or not taking things personally is not my style it's not some detached thing where i can ignore it and assume someone else will make it there you go for you, this is all business serious business mind you... :) for me, this is my very valuable time that i had no choice in how to spend it the sooner i get this programming crap done, the sooner i can get back to the life i really care about business: have you tried PLT Scheme lately? yes why? because: 1. scheme is probaly suited to build scheme in for the first prototype/version 2. the plt team has gone a long way in creating a bootable system which is squeaky like s/build scheme/build slate/ most likely 3. they have many excellent tools to build upon and they're a bunch of great chaps who know theory "great chaps" is opinion yes, it's opinion but I might be right :) well then you're suggesting i mail them or something? 12:30pm otherwise i don't know why it would matter if they're "great chaps" as such, the plt team is probably much closer to a tunes-like system and closer to slate than this channel is close to be all hacking, programming, design, etc is usually teamwork.. cooperating with nice chaps is much better than cooperating with noone at all sure or people who like to go off and build random oses subscribe to their mailing list, read docs, play around and see if you can get slate working within plt scheme and you have a great environment to build upon alright this is friendly advice well thx it might not be the Real Thing, but it's a great environment to experiment with and get feedback from I guess which of their implementations would be best? I usually use MxScheme, but if you have decent lines, get the whole thing s/Mx/Mz/ so drscheme? yes i think i still have an old download damn it not any more and if you allow me to give some more friendly advice: tone down the insults, because it is difficult to convey ones interest and vigour over the Net and easy to misinterpret n/m found it yeah well you and fare are too certain of yourselves so what's the best way to study code in a lisp? in smalltalk i can just use class browsers acl had a nice class browser as well ACL on Win and LispWorks have "friendly" environments i take it you mean over-friendly ACL on Unix has a Emacs-interface which is nice if you know your Emacs but clunky for everyone else (I use Emacs) lord knows it's not their job to teach you how to use the environments ;) i don't i hate key-bindings 12:40pm ACL on Win or Lispworks (for any platform) or MCL for macs then thx i already knew that for Scheme I think DrScheme is the most advanced environment you'd think they'd at least put a code browser in the menu sysytem -:- jham [user5180@209.4.195.169] has joined #tunes they have all sorts of newbie things in the menus but no code browser hm hello jham whats up? we're discussing scheme and tunes cool * water/#tunes goes through the tutorial Im just checking this room out. ok well have you looked at the website? not yet. whats the url www.tunes.org hold on Ill check it out. yeah it just seems odd that they don't put a code browser in front hm? oh "full graphical scheme" 12:50pm * water/#tunes is afk (eating) jham: got any questions/comments? 01:20pm not at this point, this is something new on me. ok i suppose you're not a programmer, then? nope, only a novice. ok shameless plug: ever used smalltalk? take a look at www.squeak.org no I do mostly network admin ok Ill take a look at the site though 01:30pm * Fufie/#tunes yawns.. the caffeine doesn't work anymore :( heh 01:40pm god damn javascript i love it when sites really on it s/really/rely/ doh fufie: is there a work-around for this? 01:50pm well i'll bbl then -:- water [water@tnt-9-193.tscnet.net] has left #tunes [] 02:00pm -:- SignOff jham: #TUNES (Leaving) -:- SignOff smoke: #TUNES (z) -:- eihrul [lee@usr5-ppp130.lvdi.net] has joined #tunes -:- water [water@tnt-10-4.tscnet.net] has joined #tunes hey eih how's life? monotonous school or coding or both? school hm well i got my substructural logic book finally. it seems very good i got to discuss slate quite a bit with fare and fufie this morning 03:00pm mostly explaining how it aims to implement the hll ideas btw, can you sum up clisp packages for me? i'm going into the clhs but it isn't always as clear as mud :) still don't know too much about packages myself fufie tried to understand why i had the namespace system in slate set up the way i did, but in the end he just said "oh that's different from lisp packages" and told me to rtfm i hate it when he does that he at least could have given me some clue as to why i should rtfm or what i should focus on symbols and the reader what about them?? is there something to learn? gimme a url within the clhs at least it's huge clhs has a chapter on both the reader, symbols and packages where? i'm in the packages chapter but it's a forest 10 is symbols and 11 is packages no kidding clhs is meant as a reference, not a tutorial well got a url? i gave you one earlier for that paper http://localhost/~stig/doc/lang/cl/HyperSpec/Body/sec_2-2.html localhost doesn't help much :-/ heh