IRC log started Mon Jan 17 00:00:02 2000 i have to define some core operational vocabulary and build the other ideas on top of those rmm, what sort of interfaces? [msg(TUNES)] permlog 2000.0117 well, time to reboot :) java-style eeew i don't see why such are really needed i.e. protocols shared between unrelate objects except for asking whether a certain object conforms to it and that might as well just be a predicate well, that's what the class idea was partly for and it makes multiple inheritance unecessary which relieves the implementor of many headaches :) ok bbiaf -:- water [water@tnt-9-136.tscnet.net] has left #tunes [] -:- water [water@tnt-10-65.tscnet.net] has joined #tunes ok, let's continue without such a predicate if you had a virtual method with, say, an "inner", the method that extends it would have to use the virtual method's source in the new context 12:10am yeah but on the other subject ok an object either responds to a message or it does not if it responds to a certain set of messages its within that class * water/#tunes contemplates undecidable message protocols :) ok, behavioral classification so its only utility would be in that you can ask: can i safely interact with this object in all ways needed by this protocol? the only such danger in not using a predicate would be an object happen-chancily responding to some of the protocol but not all of it yeah, assuming you can automatically grok the protocol so long as you could iterate over the structure of an object and its chain of parents, this could easily be done from within slate i mean, two objects might call methods they have by the same name that have the same number of arguments, but that doesn't mean the protocols are compatible even on that one message-name it also would take a while of course, if it's built up incrementally, it might not be a big run-time load it'd almost be easier just to use a prototype that designates the object conforms to the protocol represented by the prototype that way the predicate need only be applied once well, i think that there are still ways to use predicates properly mostly just by some type analysis or just comparing source code hm... this should stay out of the language core too complicated so far * eihrul/#tunes nods. of course, we might figure out a simple way later on, but for now well, as i suggested, you could view the predicate as a compliance test, if it passes, it wins a link to the prototype that designates its certified :) s/its/it's yeah, i just don't see that approach as being useful basically, cacheing the predicate maybe i don't see how you are thinking of predicates 12:20am as a test (over this oh-so-expressive ascii link :) no kidding, "as a test" bah forget this, it's not worth discussing now there must be many things more productive dictionary.com: "To make a statement or assertion." no, i mean at what level... do you want predicates to read code? well, you never mentioned at what level this is used :) sigh... eh? we'd be better off working out the code model i mean, the actual substance of it the trick seems to be that we're dealing with objects and functions in the same syntax without a functional basis we still of course have symbol bindings and environments and all that, though they might get re-designed i'm thinking that "car" and "cdr" won't cut it anymore * water/#tunes ducks sure, that's why you have first and rest * eihrul/#tunes ducks lol so the question now is how one would produce code for slate? no, i was thinking that you could address the contents of the cons cell with "0" and "1", that way combining them works like making binary strings that correspond to a notion of binary tree address sound cool? eihrul: we're working on that sure, but any name will do they're just a 2 element vector what the given element means is dependent on its use well, i'd like the addressibility thing to be intuitive so foo and bar would work appropriately even yeah but foobarfoofoobar doesn't seem to correspond to 18 12:30am :P i'm just saying the name of the slots is arbritrary, doesn't matter well ok i'll ignore that for now we then have simple function call/object instantiation which don't seem to need to be distinguished, afaict okay... question how mention that such would only take a single argument huh? speak english dude you said function have one input and one output well, for now rmm... s/how/you, i tend to insert the wrong words in the wrong places heh. interesting habit :) perhaps its a minor form of epilepsy? yes, let's use the single object i/o idea for now what about binary functions like '+' or '='? they take cons cells * water/#tunes ducks or even unordered pairs at the worst you can do something like haskell, but why bother? hmm... that introduces message passing into the simple obtaining of arguments, but it removes the same thing from the calling object how does haskell do it, i forget it uses higher order functions oh yeah no, i'm not interested in those per se hm well it doesn't change syntax from the usual lisp stuff 12:40am i mean, function calls look the same from the caller's perspective so long as he doesn't have to cons arguments and that's handled by the mop :) now, though, object-oriented basis for language would allow one to pass arbitrary containers instead of just cons cells heh bleh, i'm not sure how far to go with this design at this level incrementally in all areas is a good strategy heh sure, whatever that means well, if you focus too much on one part of the design well ok so, we have car, cdr, apply which work on all ast's, both objects and functions well, apply would only work on full objects hm (object message arg1 arg2 ....) that's a message-send (apply object arg1 arg2 ...) that fills in object slots and initializes and of course clones a question that arises for me is how to leave a slot unbound, i.e. for partial-eval? maybe a special keyword just for that concept currying? i think currying is a bit too much for him yet ;) or you could apply a slot re-ordering and just apply the first n arguments abi forget currying water: I forgot currying that would leave the rest unbound 12:50am no, currying relies too much on functions but is partial evaluation similar in concept? of course it is i just don't thinking higher-order functions are easy to grasp if you could address lots by name for initialization, you wouldn't need to reorder s/lots/slots wha twould the syntax be then? not sure but, if you view (object message) as a unary message that clones the object and further addressing of the slots by name to initialize them as just sending a message to the slot to assign it and then finally initializing when all slots are filled that need to be would that be all the slots, though? hmm, no, atleast not in the case of methods why not? some variables may just be temporary results whereas others are parameters there's no easy way to do this not without breaking the lisp model hmm afaik, partial evaluation in lisp would just be done with a closure yeah but even partial eval in lisp requires non-trivial syntax 01:00am just (lambda) how so? for arbitrary selections of arguments? how would you partial-eval on the middle argument of a 3-argument function? you make a function that supplies an implicit argument write it out and otherwise just routes all the remaining arguments (lambda (x) (+ x 1)) that's jsut one argument (lambda (x z) (+ x 1 z)) what about (lambda (x y z) (* x y z)) how would you substitute for y only? you'd just make a function to wrap it that binds a specific argument to y bleh, i must be getting tired, i just realized how easy it is (duplicating much of the work of currying) it's still not message-passing i think that's what's bugging me therefore not pure oo well, the other way i described is very similar to currying how is currying oo, though? not quite OO, but still fits within framework of message passing :) ummm no i'm not giving up pure oo for anything how would you represent partial eval then? well, lets look at functions as objects we're basically asking how we can clone by assigning but only partially-assigning so, we apply an eccessor method by name ah of course the syntax would be... (object (y 1)) for isntance 01:10am that's not too terribly different from what i suggested :) or (myFunction (y 1) (x 2) (z 4)) yeah i know that now but you weren't fast enough on giving me the syntax :P the syntax without sugaring wouldn't be pretty for the lambda (x y z) -> x*y*z, the above evaluates to 8 ((object member) y 1) (((object member) y 1) z 2) not functional! its not (object member) returns a clone of the member well, why (object member)? member being the specific message but it doesn't update the object the *object* is what counts this is just viewing a method as an object that clones itself when retrieved from another object's slot, and then evaluates when all components needed to complete the message have been delivered hm not quite why would retrieval itself imply cloning? it would seem that the "delivery" is the only thing that nees cloning s/nees/needs that's what i meant well, it didn't come out that way :) well, lazy eval would obviously help out the implementation however, the clone needs to happen before any assignments are made if we're going to toy with ideas like this not *before*, *while* how do you mean while? assignments are cloning actions themselves oh, the immutable objects rear their ugly heads :) yep that's half the point of cloning in my idea of the system so in that sense (object message) never need return a clone right, lazily i mean, you could program every single reference to something to clone it, but that's pretty useless 01:20am hmm, i must sleep alright -:- SignOff eihrul: #TUNES ([x]chat) -:- SignOff hcf: #TUNES (Leaving) 01:30am -:- SignOff water: #TUNES (Leaving) dum di dum 02:00am -:- binEng [Anders@j141.ryd.student.liu.se] has joined #tunes ah, up with the sun. almost.. hi Fufie hi bineng You're up early too, eh 02:10am the phone awoke me.. I was awaken too. Otherwise I'd slept for a few hours more, I guess.. read your glossary mail ah seems as if Prism introduces a dozen new meta-words.. a glossary might be helpful I'm working with that now. I have been doing that for the past two days Turns out that the glossary just grows. Currently I have some 50-60 terms ouch maybe going with a standard vocabulary saves SDS a lot of trouble then Could be... Seems that the terminology comes from cybernetics and semiotics, fields I have previously little experience with. But there are good sites. 02:20am -:- TwAt [Cunt@193.63.36.106] has joined #tunes Muzik :) -:- TwAt has changed the topic on channel #tunes to: TUNES, Free Reflective Computing System || slate lol heheh big hormon production this morning? :) -:- SignOff TwAt: #TUNES (Excess Flood) excess flood? Looks like you were right Fufie, lol -:- TwAt [Cunt@193.63.36.106] has joined #tunes what were we saying ahh yes hello hehe -:- TwAt has changed the topic on channel #tunes to: BinG, Free Reflective Computing System || slate lol -:- TwAt has changed the topic on channel #tunes to: BinG ok ok Junglist Massive check this out, Free Reflective Computing System || slate lol could you stop that, TwAt? lol nice that sounded quite good -:- TwAt has changed the topic on channel #tunes to: BinG ok ok Junglist Massive check this out, Free Reflective Computing System || slate there you go not ruddddddeeeeeeee any more i suppose -:- binEng has changed the topic on channel #tunes to: TUNES, Free Reflective Computing System || slate now let it be what!!!!!! you heard put the junglist massive back -:- TwAt has changed the topic on channel #tunes to: Thus room is full of gays!!! come on in for a good pumping!!! lol lol -:- TwAt [Cunt@193.63.36.106] has left #tunes [] -:- binEng has changed the topic on channel #tunes to: TUNES, Free Reflective Computing System || slate idiot -:- TwAt [Cunt@193.63.36.106] has joined #tunes hmm doh twat: relax some it's early morning erm im joking we don't appreciate your jokes 02:30am 11:40 is early???? morning lol yes erm no I was up until 5:00 this morning, so yes it is pretty early. thats your fault lol your getting no sympathy from me you should go to bed you see already been there then you sleep and then you can get up twat: stop sounding like a mum and then it aint early at 11:40 well 10:40 where i am how nice for you oh yes very in fact oooooooooooo orgasmic 02:40am -:- TwAt has changed the topic on channel #tunes to: WANKERS ARE HERE!!!!!!!!!!! COME AND JOIN THE JIZZY FUN LOL well i think your a load of arse's and that you should go and get a life!!!!!! BYE -:- SignOff TwAt: #TUNES (Dont be a twat!!!!) don't... move -:- PUTTY [Cunt@193.63.36.106] has joined #tunes -:- PUTTY [Cunt@193.63.36.106] has left #tunes [] * binEng/#tunes shakes his head * Fufie/#tunes giggles -:- binEng has changed the topic on channel #tunes to: TUNES, Free Reflective Computing System || slate 02:50am Fufie: This script I made, is it related to SDS or S14 in some way? 03:00am which script? That one I talked about on the mlist, to generate a html file of the glossary not a lot I guess oh well it's more like a utility and both SDS and S14 are systems if you look at the sds web-pages.. all html-pages there are generated by python-scripts but the utility reflects a system, a code for a glossary (to use the terminology) I might have a look at the script and see if I can translate it to a SDS context.. that might be useful how'd you mean? well.. SDS has powerful tools for XML, and glossaries and linking is done very well in XML.. ic 03:10am 89 nodes in the glossary now I should split it up into one page per node Fufie: Do you know of any definition of or discussion around "VM"? no definition no or maybe.. two sec 03:20am umm no too bad 03:30am -:- SignOff binEng: #TUNES ( ) -:- dalvarez [ircusr@212.68.72.98] has joined #tunes hey all hi dal someone knows whether SPARCv9 supports a multiplication instruction directly in the RISC inst set or a .mul macro like in the v8 class ? and how many cycles every of both ways takes to exec on different processors I'm thinking about multiplication optimisation (when to use shift and add or the mul) 04:20am afk 04:30am -:- binEng [Anders@j141.ryd.student.liu.se] has joined #tunes hello 05:50am -:- Ghyll [karltk@mp-217-241-36.daxnet.no] has joined #tunes -:- AlonzoTG [Alonzo@216-164-134-100.s354.tnt3.lnhva.md.dialup.rcn.com] has joined #tunes -:- SignOff Ghyll: #TUNES (bar) umm.... must be night where most ppl are it's around half past four here a. m. I guess greetings 4:24 p. m. in get ger * binEng/#tunes was admiring INTERCAL binEng sure knows if the SPARCv9 has a mul instruction directly as part of the inst set do you, binEng? nope (what question I answered is up to you to find out) the v8 has a .mul macro (which I think is resolved into shifts and adds at compile time equally as .div) -:- Downix [down@d-gnaps-112.ici.net] has joined #tunes anyone here know of a good disk partitioning utility? partition magic I just was thinking how that could be optimized (i.e. when to use an eventually existent mul or directly shifts and adds) 07:30am -:- SignOff dalvarez: #TUNES (bbl) -:- SignOff Downix: #TUNES (Ping timeout for Downix[d-gnaps-112.ici.net]) flip flop flap is brilliant! still looking at that intercal stuff? 08:20am a little. om I had a look just before, actually 08:30am -:- Ghyll [karltk@mp-217-229-3.daxnet.no] has joined #tunes * Fufie/#tunes stretches * binEng/#tunes is away -:- thomas [thomas@193.217.63.152] has joined #tunes hei thomas -:- Downix [down@d-gnaps-222.ici.net] has joined #tunes hoy all 09:30am hi 09:40am -:- SignOff Downix: #TUNES ([BX] Time wasted: all of it) -:- dalvarez [ircusr@212.68.72.98] has joined #tunes greetings... -:- hcf [nef@me-portland-us746.javanet.com] has joined #tunes 10:00am -:- SignOff Ghyll: #TUNES (bar) * binEng/#tunes is back -:- Downix [down@d-gnaps-284.ici.net] has joined #tunes * AlonzoTG/#tunes left clicks on Downix abi: joy page? i haven't a clue, thomas abi: joy? i think joy is at http://www.latrobe.edu.au/www/philosophy/phimvt/j00syn.html or a forth like functional programming language nice.. people who want to discuss bob dylan are posting to comp.lang.dylan :-) heh :) I can stand this debian-emacs, what command would you use to switch to another buffer? select-buffer, right? and not select-frame or whatever I think there is next-buffer Fufie: a named one out of 20 there was this select-buffer command in me switch-to-buffer ahhhh... better can't remember which key I used for that.. I'll let you know the next time I just click it Hmm, no replies to my prism mails yet :( -:- SignOff Downix: #TUNES (BitchX: the right choice for a healthy pet) has eaten list-bindings, too (or am I just confused and there never was one) WHAT THE HELL ?!? 10:10am set-variable : yes; list-variables: no; abi: emacs? hmmm... emacs is a victim of it's initial success or sounds like a great idea in theory, but the implementation sucks thanks, abi dalvarez: sure thing will tunes have a *util* emacs ? somebody must have spend damn much time on this abi thing... it is actually a *she* abi: right? bugger all, i dunno, bineng gee as she is interrupt driven she won't know until you tell her :) right, abi? abi: you're female. foo, abi? thanks, ab_i, works thanks, abi sure thing dalvarez 10:20am would be mad if _she_ would protest herself... 10:30am bbl -:- SignOff dalvarez: #TUNES (Leaving) 10:40am -:- eihrul [lee@usr5-ppp58.lvdi.net] has joined #tunes hello eihrul anything new on this slate thingie? afaik, just some discussion last night (can't say what water has done in the wee hours of the night though) alright 11:00am abi: Green is at http://www.dc.ufscar.br/~jose/green/green.htm abi: no, Green is at http://www.dc.ufscar.br/~jose/green/green.htm okay, Fare. -:- Downix [down@d-gnaps-264.ici.net] has joined #tunes -:- SignOff binEng: #TUNES ( ) so what's up? omlie omlie omlie omma. 11:40am -:- dalvarez [ircusr@212.68.72.98] has joined #tunes hmm green appears to be a weird mix of pascal and java