IRC log started Sat Apr 15 00:00:01 2000 hm rosette is like scheme [msg(TUNES)] permlog 2000.0415 * water/#tunes fires up the new Brian Transeau album hm a mail-queue abstraction mechanism thx hcf np hm recursively acotr-based.... sounds a bit sloppy -:- nate37 [nate@ppp-206-117-3-240.dialup.pcmagic.net] has joined #tunes s/acotr/actor/ mostly because it seems everything has to by sync'd to be used but maybe there's a way that this is done simply hm they have a separate mbox for the queue, but the "meta" slot refers to a *structural* mo 12:10am -:- SignOff nate37: #TUNES (nate37 has no reason) * water/#tunes moves on to look at coop bleh green is too much like java 12:20am -:- SignOff hcf: #TUNES (Leaving) * water/#tunes returns to working on the slate tutorial * yinzen/#tunes cheers heh :D is anyone familiar with re-factoring and/or patterns? hm i guess not 12:30am -:- SignOff lar1: #TUNES (Leaving) -:- gmol [gmol@24.68.42.158.on.wave.home.com] has joined #tunes water> ? hey gmol sleepy or shall we continue our discussion? kinda sleepy. where were we? * water/#tunes is writing up a slate tutorial no matter tommorow ok? we can talk now, sure Ok..so we have established that the meta obbject is just a defined way of accessing and changing an objects methods, yes? sure but i'd better distinguish between how accessing and changing are controlled an object's mo controls how its slots are accessed 12:40am but the *slots'* mo's control how they are muted, since you have to invoke *their* slots to do so brb k back right show me snippet of changing using a mo to change a method well any state-change is an example myPoint x : 5 : is a slot of x which changes x so that it returns 5 'changes' x? yeah it plays with x's internals our current and best idea is that x simply inherits behavior dynamically from 5 so that : simply sets the parent slot of x to 5 please formalize it, an object (as we duiscussed last time) is a set of slots, names and a map between them, how does a slot change an objects internals? of course meta-behavior plays a strong role behind the scenes k so where is the meta obejct? in the example implicit, basically meta-behavior gets set when the object is created Well the meta object is an object like anyother with slots etc...so show me the object and the names of the slots it contains (in this case this is the likely situation, not what must be so) hm well the mo must have a lookup and apply method for making its objects' slots work ok wait lets go back 12:50am most of the rest is book-keeping basically an object is a set which contains: k slots names a map between them heh you're really stuck on that each object has assumed slots such as meta object and ...? i wouldn't say that the object "contains" the names as such sure it doesn't have to contain names "meta", "parent", "..", "result", .... how many reserved slots are there? hm well there are also a few primitives primitives=? such as clone, etc what is aprimitive? a primitive is a semantic foundation of the object notion that does not break encapsulation such as add-slot, remove-slot a lot of this is on the semantics page fine ok so we agree about the definitions for most part now..right? sure now you didn't answer the last question, how many reserved slots are there? well i'm not sure how many i'm willing to commit to right now i might factor some behavior into auxiliary aobjects but overall i'd say about 10 primitives most likely it's around 7-9 now slots them selves are objects, correct? sure primitives will simply be annotated by compiler references now the meta object slot...why do you need it, when the add-slot, remove-slot primitives are there? (potentially replaced by re-compiled meta-level code as achieved) ah because simply changing the slot-composition of an object is changing the object itself i want to be able to carry the object around unchanged and affect its interaction without changing its fundamental behavior well if I change the meta object of an object, it is no longer the same object no? "it" is a relative term i intend on supporting all kinds of "it" varying degrees of meta-behavior can be included with an object's definition as desirable i must write a post on this to the tunes mlist, ... "identity and meta-behavior in slate" You say: by changing the slots of an object, you change the object....what's the difference if I change the slots through add-slot etc. or through a meta object, the slots are still changed no!!! 01:00am in the meta-object you cannot affect structure of the original object meta-objects define how the object's behavior gets invoked and applied not *what* the behavior consists of so show me a snippet of changing an objects behaviur application via a meta object this is why i say meta-behavior, not meta-structure ok a simple example is to replace the application of a message expression to an object with the same application to some kind of clone of the object where cloning itself can be modified itself by another meta-object -:- yinzen [foobar@cx158568-a.okcnw1.ok.home.com] has left #tunes [] since i don't have meta-objects completely concretely defined right now, i don't have a relevant piece of code that is anything but pseudo code suck it up ;) this is not due to conceptual problems, only due to design considerations of safety vs expressiveness What's a message expression? * water/#tunes sighs when you evaluate everything but the top-level expression in a message, you pass that to the object well i'm sorry but that sentence doesn't at all illustrate what you said a meta-object was before me asking you to give me asnippet gmol: well if you want a tutor, pay me $50 per hour please you haven't defined what a message is relative to slots an object etc... because i don't get payed to make you feel comfy about my ideas RTFM!!! a message is an object. the inputs are slots of the object input-behavior is specified by the meta-object of the slot Look buddy, I'm sure your a nice guy in real life, but getting defensive when asked to give a concrete example is a good sign your ideas are fluff...take care -:- SignOff gmol: #TUNES (xchat exiting..) meta-objects in simple applications will either stop message-applications or will apply messages to computed results of the original objects (like clones) jack-ass it's 1AM and he wants me to make a dissertation * water/#tunes sighs 01:10am university-weaned academics are so impatient who does he think he is that he has some kind of privelege to listen to me only when i fit his academic model of how research is done? how many times must i have to deal with these people who won't add to my ideas? how often must i coddle little students who get to work in well-funded labs while i spend my off-hours doing this stuff? * water/#tunes has become a bit depressed is anyone here paying attention? gmol: if you condescend to read these logs, please note that academic reference papers on the subject of meta-objects as i discuss them are at http://www.tunes.org/~water/slate-references.html QED 01:20am -:- smoke [smoke@16dyn233.delft.casema.net] has joined #tunes -:- water [water@tnt-9-165.tscnet.net] has left #tunes [] has anyone read the OpenML announcement? 02:20am -:- XeF4 [xef4@88.ppp1-23.worldonline.dk] has joined #tunes -:- nate37 [nate@ppp-206-117-3-240.dialup.pcmagic.net] has joined #tunes -:- SignOff nate37: #TUNES (Hey! Where'd my controlling terminal go?) -:- SignOff XeF4: #TUNES (Leaving) -:- water [water@tnt-9-165.tscnet.net] has joined #tunes -:- nate37 [nate@ppp-206-117-3-240.dialup.pcmagic.net] has joined #tunes -:- SignOff nate37: #TUNES (Hey! Where'd my controlling terminal go?) -:- Fare [rideaufr@esmeralda.enst.fr] has joined #Tunes -:- nate37 [nate@ppp-207-151-70-69.dialup.pcmagic.net] has joined #tunes -:- SignOff nate37: #TUNES (nate37 has no reason) Gakuk! 05:40am -:- eihrul [lee@usr5-ppp178.lvdi.net] has joined #tunes -:- smoke_ [smoke@16dyn233.delft.casema.net] has joined #tunes foo bar baz * water/#tunes 's head feels a bit fuzzy 07:00am -:- SignOff zarq: #TUNES (Ping timeout for zarq[10dyn95.delft.casema.net]) -:- zarq [zarq@10dyn107.delft.casema.net] has joined #tunes re reoi 07:10am -:- ult [ult@user-37kbapp.dialup.mindspring.com] has joined #Tunes -:- SignOff smoke: #TUNES (Ping timeout for smoke[16dyn233.delft.casema.net]) -:- SignOff smoke_: #TUNES (Ping timeout for smoke_[16dyn233.delft.casema.net]) -:- smoke [smoke@16dyn233.delft.casema.net] has joined #tunes * thomas/#tunes is back ( took a while :) -:- SignOff water: #TUNES (Ping timeout for water[tnt-9-165.tscnet.net]) -:- XeF4 [xef4@25.ppp1-4.worldonline.dk] has joined #tunes -:- water [water@tnt-10-226.tscnet.net] has joined #tunes * water/#tunes is now fully awake hi all hi 08:40am -:- SignOff thomas: #TUNES ([x]chat) oh c'mon... does anyone have something to talk about? 08:50am * water/#tunes gets back to writing the tutorial -:- SignOff ult: #TUNES (Leaving) eih: there's a paper on performance comparison between lazy and strict evaluation at http://research.microsoft.com/Users/simonpj/Papers/pjlester-1.11.tar.Z microsoft uses tar ?! ms research is not quite microsoft 09:30am isn't he the author of o'haskell? 09:50am -:- SignOff XeF4: #TUNES (Leaving) dunno it's not mentioned on his home page oops wrong home page * water/#tunes considers some ideas concerning scheme vs slate what do you think of the explicit lambda keyword vs fp using slate objects? 10:00am -:- rocco [rocco@rocco.res.WPI.NET] has joined #Tunes hi rocco hello - just going to lurk for a bit well there's not much going on What does it mean that the Slate Tutorial is "in progress" i'm still typing it there are a lot of things to explain to the average programmer that i'm trying to make sure i get right 10:10am i unfortunately screwed up explaining meta-behavior to gmol last night, but then i've explained it to many other programmers who grok the idea now meta-behavior, in a nutshell, is ... hm maybe i'll get it right this time... in slate you have behavioral meta-objects which can intervene in how behavior is handled behavior being message-sends to objects, and objects' protocols as a whole for instance a meta-object can dynamically forward a message-send to a *clone* of the object, rather than the original object does that make sense? does this all happen with the objects' blessing, or regardless of it? well the object has a slot "meta" which denotes its meta-object but meta, being a slot, is subject to meta-behavior :). so the interaction is variable its supposed to be a simple concept that is powerful because it gets applied everywhere variable, but predictable, right? anyway, because "meta" is a slot, an object can dynamically alter its meta-behavior absolutely, the system of meta-objects is explicitly programmed and customizable of course as with any programming system, you can screw things up in a hurry 10:20am anyway, the idea is that a library of meta-objects can provide things like encapsulation protection, concurrency management, inheritance semantics, etc as another example, i'm working on implementing an equational manipulation system that is extremely generic, based on the use of meta-objects water: well, this thing is compiling properly... that's good that's good!? heh ok i assume it doesn't work, then? definitely, having to compile a research paper is lame :) oh the research paper i thought you had a new eval.lisp ;) er, i meant "isn't" er rather, this research paper on functional languages isn't compiling properly :) * water/#tunes whacks eihrul with a big wet fish well i think i'm 3.5 for 5 for explaining meta-behavior recently 10:30am the conversation with maneesh was a total failure -:- lar1 [larman@adsl-63-203-74-81.dsl.snfc21.pacbell.net] has joined #tunes re Hey pjlester is a book about compiling haskell yeah i guess i was misled :/ pj is a haskell designer and author of ghc k 10:40am eih: hey if gmol shows up sometime while i'm not here, please explain meta-behavior to him for me 10:50am hey eih -:- SignOff lar1: #TUNES (Leaving) 11:20am yeah? damn it now i can't think of what i was going to say well, what's the status of a new eval.lisp, if any? haven't started yet any forseeable problems? none right now finally converted this thing from dvi to ps -:- _ruiner_ [blah@ppp072.wi.centurytel.net] has joined #tunes hm ok, i'm writing up some stuff in the tutorial on meta-objects, though its a little rough right now 11:30am -:- hcf [nef@me-portland-us1046.javanet.com] has joined #tunes hey hcf hey 11:50am water: r u gonna announce (to tunes@) ur tut when done? definitely i need to figure out a good way to introduce mo's though they're tied into everything, so it's hard to introduce it in a way that is clear perhaps with a simple example? 12:00pm well rocco the only uses of this kind of meta-object have been in tiny experimental systems water: perhaps by introducing existing object systems and how *static* meta-behavior affects them... hm yeah that would work it seems, since i can take concepts from other languages and map them into slate objects + mo's while they don't quite have behavioral meta objects in the slate sense, they do have meta-behaviors even though they are unextensible by the programmer yeah i hadn't considered that angle thx * _ruiner_/#tunes yawns hmm, any way to view postscript in a palm? :) <_ruiner_> heh palmtops.... i wish there were you can ps2pdf and use mobibook to convert to html well, indirectly through conversion? (palm => pilot) there's also adobe's pdf2html online pilot's can handle html? i'm pretty sure they can if they can't, there *must* be a free utility to convert html to palm docs -:- SignOff _ruiner_: #TUNES (Leaving) we have such nice lurkers here in #tunes 12:10pm eih: icuc, http://users.erols.com/arenakm/palm/RichReader.html 12:20pm Does anyone know what the status is regarding using Wiki for TUNES Web Development? dead well, it hasn't been used in a year afaik Dead from non-use, or dead because of some other reason? non-use hcf: costs money :) Is there any technical/practical arguments against using it? <*musing about bringing it up to date*> well tril's working on a db for review/gloss the wiki is supposed to be for discussion of issues and calling of meetings eihrul: i saw mention (on deja) of a freeware version just talk to tril about it Okay :) 12:40pm now that i think of it, would rather waste money on a vectorization book and gc book than on a pda :) 12:50pm what odd conversations we have here which vectorization book are you talking about? dunno, but i'm sure i could find one might as well put the kni stuff to use :) kni? the p3 instructions oh hm well i'm going to go out and think for a bit with some books might buy some books myself :) spend wisely although a gc book seems kinda out-moded if you have re-writing ;) of course that's not true *yet* well, i have $500 01:00pm i can afford to buy slightly out-moded books :) hm i suppose but there's nothing quite like reading code and research papers etc on the go ;) * eihrul/#tunes hmms. well, how many lines do you get on your screen? :) well my screen is 320 by 240... usually 30 lines on a screen -:- nate37 [nate@ppp-207-151-70-118.dialup.pcmagic.net] has joined #tunes * eihrul/#tunes ponders. much better than palm (12 lines) yeah well perhaps the visor has that resolution 320x240 well cya later. its much nicer reading outdoors -:- water [water@tnt-10-226.tscnet.net] has left #tunes [] 01:10pm -:- SignOff nate37: #TUNES (nate37 has no reason) -:- smoke_ [smoke@16dyn136.delft.casema.net] has joined #tunes -:- SignOff smoke: #TUNES (Ping timeout for smoke[16dyn233.delft.casema.net]) -:- smoke_ is now known as smoke -:- SignOff rocco: #TUNES (xchat exiting..) -:- SignOff hcf: #TUNES (Leaving) -:- hcf [nef@me-portland-us636.javanet.com] has joined #tunes -:- gargamel6 [placr@212.83.165.170] has joined #tunes -:- SignOff smoke: #TUNES (z) -:- SignOff gargamel6: #TUNES (Leaving) -:- me2 [me2@1Cust247.tnt3.houma.la.da.uu.net] has joined #tunes -:- Downix [down@d-gnaps-137.ici.net] has joined #tunes -:- SignOff me2: #TUNES ([x]chat) -:- _ruiner_ [blah@ppp311.wi.centurytel.net] has joined #tunes -:- SignOff _ruiner_: #TUNES (Leaving) -:- SignOff Downix: #TUNES (Ping timeout for Downix[d-gnaps-137.ici.net]) -:- SignOff hcf: #TUNES (Leaving) -:- _ruiner_ [DIY@ppp102.wi.centurytel.net] has joined #tunes -:- lar1 [larman@adsl-63-203-74-81.dsl.snfc21.pacbell.net] has joined #tunes -:- mikes [user5991@ABD80F5E.ipt.aol.com] has joined #tunes -:- mikes [user5991@ABD80F5E.ipt.aol.com] has left #tunes [] -:- SignOff smkl: #TUNES (Ping timeout for smkl[glubimox.yok.utu.fi]) -:- smkl [sami@glubimox.yok.utu.fi] has joined #tunes -:- Vegeta [LOL@spc-isp-tor-58-20-655.sprint.ca] has joined #tunes HET ALL COME TO ROOM DRAGONBALLZ -:- SignOff Vegeta: #TUNES (Leaving) * eihrul/#tunes sighs. <_ruiner_> lol <_ruiner_> poor eihrul well, that is definitely noise we can all scoff at :) <_ruiner_> I dunno, its a cool show.... <_ruiner_> <_ruiner_> good thing water wasn't here well, yeah... but it is a little offtopic... <_ruiner_> he'd probably hunt him down and kill him or atleast, more than usual 08:30pm -:- SignOff _ruiner_: #TUNES (destroy what destroys you) -:- SignOff lar1: #TUNES (BRB 2hr) -:- hcf [nef@me-portland-us321.javanet.com] has joined #tunes -:- XeF4 [xef4@126.ppp1-17.worldonline.dk] has joined #tunes -:- SignOff abi: #TUNES (dying by hcf's request) -:- abi [nef@bespin.dhs.org] has joined #tunes -:- SignOff XeF4: #TUNES (Leaving) -:- XeF4 [xef4@supernova.sevenheavens.com] has joined #tunes -:- water [water@tnt-10-122.tscnet.net] has joined #tunes re water water: when Tunes speaks of allowing infrastructural changes, how fundamental are the changes we're referring to? as fundamental as we can manage (i.e. the greatest change which would not break everything) right or which would not require an undecidable process of conversion so reflection to the point of low-level changes automagically modifying high-level code are no joke? mm well it's a desirable goal how this kind of thing should be accomplished is not well-defined. the best unified approach i have seen at it is the use of combinators and/or rewrite systems hm.... is how to implement a rewrite system without losing maintainability of high-level components well-defined? 11:50pm yes, because there already is a demonstrably-reflective rewrite system it's not the greatest thing since sliced bread, but it's very interesting abi maude? i guess maude is a reflective rewriting logic language at http://maude.csl.sri.com/ or http://www.csl.sri.com/~duran/ I've been toying with a system to allow infrastructural changes without reflection um that *is* reflection :) its an open implementation compiler then? hm.. I suppose ;) but only the pipes between modules are reflective maude has a system for making algebras for module systems this could be the glue language I'm looking for.. yeah its fairly robust, but don't count on it for number-crunching yet abi rewriting for gc well, rewriting for gc is "Abstract Models of Memory Management" at http://www.cs.rice.edu/CS/PLT/Publications/fpca95-mfh.ps.gz or see cora that's paper on using rewrite systems to specify a full range of memory-management strategies (implementation starting on Monday) this system works as a collection of code fragments which form complex systems by passing layered messages between modules modules act on those messages, filling in, deleting or modifying layer information not sure what you mean by "layered" messages each message accompanies a structure allowing a module to look at any known level of abstraction for that message if not already obvious, this is just a slight generalization of a network stack design I was working on one structure per message? yes the word that comes to mind is "overhead" very little overhead, in practice because in most cases neither the structure nor the message body are copied well this is interesting and all, but i have to work on slate [msg(TUNES)] newlog 2000.0416 IRC log ended Sun Apr 16 00:00:01 2000