IRC log started Sat Jan 29 00:00:01 2000 -:- SignOff fire: #TUNES (Leaving) [msg(TUNES)] permlog 2000.0129 -:- SignOff _ruiner_: #TUNES (destroy what destroys you) 12:10am -:- witten [witten@un.torsion.org] has joined #tunes -:- SignOff washort: #TUNES (Ping timeout for washort[d125.narrowgate.net]) -:- SignOff air: #TUNES (BRiX [http://www.qzx.com/brix] :: sleep) -:- witten [witten@un.torsion.org] has left #tunes [] -:- washort [washort@d101.narrowgate.net] has joined #tunes -:- SignOff washort: #TUNES (Ping timeout for washort[d101.narrowgate.net]) -:- NetSplit: irc.linux.com split from king.openprojects.net [04:13am] -:- BitchX+Deb1an: Press Ctrl-F to see who left Ctrl-E to change to [irc.linux.com] -:- Netjoined: irc.linux.com king.openprojects.net -:- smkl [sami@glubimox.yok.utu.fi] has joined #tunes -:- abi [nef@bespin.dhs.org] has joined #tunes -:- Plundis [plundis@130.238.23.252] has joined #tunes -:- zarq [zarq@9dyn26.delft.casema.net] has joined #tunes -:- Fufie [stig@tunnel-44-12.vpn.uib.no] has joined #tunes -:- ree [js@twisted.goodnet.com] has joined #tunes seen me abi, seen me? I haven't seen 'me', ree :) 05:40am hey fufie =) sup? working on the assembler as always =) you? reading the acm communication I got today ahh umm.. 'communications of acm' they call it interesting? some stuff about toy story 2 otherwise too much serious business stuff I should soon try to port SDS to cmucl though sds? sds is a framework for developing programming tools such as source level documenters, source browsers, source metric collectors, and similar. or at http://sds.yi.org/ ahh neat 05:50am it works with ACL and clisp already heh I know almost none of these acronyms never used lisp either acl is allegro common lisp, clisp is a gpl'ed common lisp and cmucl is CMU's Common Lisp hah enough of them? there are several others as well :-) so I take it you really like lisp lisp is my language of choice yes mostly because lisp is easy to program the languages I need to solve my problems s/lisp/in lisp it/ ahh 06:00am hah.. got into bespin :-) the morale is to keep a few old ssh files around and you don't need to remember your password :-) hehe I don't need to use remote access luckily 06:20am so I just use telnet inside my lan my cvs-tree is at bespin/www.tunes.org ahh so you use cvs over ssh yes er, ssh over cvs huge security risk nowadays transfering open source code =) umm.. I use ssh for all cvs-transactions I think Tril prefers people using ssh instead of telnet with passwords in cleartext or rsh for cvs yeah, I understand 06:30am -:- smoke [smoke@16dyn132.delft.casema.net] has joined #tunes -:- SignOff Fufie: #TUNES (Ping timeout for Fufie[tunnel-44-12.vpn.uib.no]) -:- Fufie [stig@tunnel-44-12.vpn.uib.no] has joined #tunes Could anyone explain me in short at what hardware Tunes is aimed? ie - would it be as scalable to run on anything from cellular phones to planets? only small planets in the first version and what about cellular phones? a c=64 version would be out of the question i suppose? or is tunes as abstract that a c=64 version would be feasible? i'd like to do some investigating for the UI part of tunes, if possible 07:00am smoke, it'd be as scalable as you'd want it it's ported towards the hardware it runs on it won't be (afaik) a virtual machine I mean in a vm like jvm, or others xmoke: concentrate on computers as you see them in three or five years time.. spending time on c64 ui is imho a waste of time expect tunes to need at least 128M RAM haha I am not joking and it doesn't make sense to make a system designed for 10-15 year old computers 128mb ram? 128M is the least amount of memory I would consider working with if it is based on x86 it is already 10-15 years old well, I don't see why it would require 128mb I'd consider working on 1mb depending on what use I had for it 1mb? yeah 07:10am dos works quite well with only 1mb I think people have higher expectations to tunes than they have for a crappy thing like dos heh, dos isn't crappy =) dos sucks shite and tunes doesn't have to require a trillion bytes of ram to run 128M is a good estimate imho for what though? it's built-in compiler, libraries, graphics engine, various drivers, basic apps you could have 1280 100kb applications (memory/code) running at once in that amount of memory ahh well it's a pos if it requires 128mb of ram for the basic operating system of course, imo pos? piece/pile of shit why? there is no reason a system should require that much memory and how much does linux+x+gnome take? they are horrible examples linux+x+gnome are realistic examples but you could run them in 16mb of ram they use c++ code 16mb of ram + 100m swap? no way :) and have a huge amount of abstractive layers neither linux, nor x nor gnome use c++ gnome uses it I thought kde does gnome uses c and kde runs faster than gnome heh e is a damn slow system you don't need to run e have you tried the qnx demo disk but I think linux+x+gnome is a realistic example of what is needed of a modern system you can run that on 16 or so mb of ram and in five years time that memory use will be 128M but they aren't programmed well they are programmed well compared to most other stuff out there I hope for this industry's sake that memory uses go down that's a copout though memory use should go _up_ 128M is a ridiculous amount already 07:20am ohh boy =) have you compared two equal machines where one of them has 64M and one has 256M? no, but 64m and 128m the focus is unfortunately on hardware and processors.. not on memory and 32m and 64m you could get the same functionality that you do today years ago in a much smaller and efficient program my 256M give me a definite speedup for real-use compared to people who use 64M and a pretty large speed-up over 128M for what operating system? linux+x+gnome you're referencing already bloated works of software no I am not look at much better programmed systems like beos, qnx beos is a single-user system qnx is also very different.. but for comparable use it also uses a lot of memory linux can be real time easily you can boot a entire gui, web browser, inet connection, etc off of one floppy disk and have the same functionality as you do with x and a window manager with extremely small memory requirements ree: yes you can.. but then you will need a lot of memory to run the floppy stuff no, you don't you need a small amount of memory for the ramdisk then a few mb more for the applications small meaning at least 16M? 12-16 complete system and then you need to add more because you want a compiler you need the compiler compiler isn't in memory 24/7 in tunes it will be in memory 24/7 hm interesting i see you have all very different opinions on the hardware to use and the compiler needs memory I can hardly see why a memory resident portion of a compiler would take 100mb of ram ree: I am not saying it takes 100M of ram for the compiler ree: but the basic system you mentioned took 16M add the compiler and we're at 24 right away i think that if you design tunes for a machine that will be top-notch in 3 years, people will run tunes from 2003 to 2010 and then it's over then the system needs a good gc and memory to let the gc be good is that what you want? that's a illogical waste of memory to keep the compiler in memory 24/7 smoke: who are you asking? all of you smoke, I'm not with the tunes project I just like discussing ideas ree: ah! that explains a lot :) ree: no it is not.. tunes will need a compiler all the time to improve the source.. just like java needs a JIT s/source/code/ ahh fufie: so you see tunes as a project for the next ten years? so it will be on a vm smoke: I see tunes as basic research.. and with a decent system in 3-5 years how many years has it been so far? ree: it will have a vm yes.. whether that vm is software or hw doesn't matter to me tunes will need a compiler for same kind of reasons that linux needs dynamic linker 07:30am fufie: i always thought tunes was about the OS ree: too many.. smoke: tunes is as I see it about the understanding of the OS.. and to do something about the bad separation between compilers, languages and the OS itself should i see tunes as a research project or as a future os ? smoke: yes research right now it mainly starts other research projects what hardware architectures have been researched? that's part of the review subproject research should not be hampered by people who wants to run stuff on past technology imho this slightly reminds me of stories of new versions of aol research is about finding new technology trying ut new stuff s/ut/out/ you should also take care not to label old technology as useless take for example the horrors i'm going through in order to make constant framerate graphic applications. it was very easy to do this in singletasking systems, but it's nearly impossible on modern multitasking OS'es research should also be about past technologies.. a lot of good things are already thought of and forgotten how to make past technologies work in the future you mean? for example it may also well be that in the future simpler systems are used it could well be that in a year or five the western economy breaks down and we have only old equipment to work with then we should face that grim prospect if it happens a programmer is an optimist after all i am not not at all even 07:40am why do you program then? to get a job? for therapy of course :) most programmers I know like building stuff.. they want to see things work :) i'm not that much of a God myself but this is getting a little personal :) i'd like to know more about Tunes :) well.. tunes will need more memory than the crippled ms-dos did :) don't be afraid to use memory :) fufie; i am cure me :) :p 07:50am i am not afraid to use memory but of course there should be algorithms that modify other memory hogging algorithms I think the problem is that programmers are getting lazy =) like the latest "technological breakthroughs" in language design which are those? ree: you may well replace lazy with stupid object oriented, abstract this and that is abstraction bad? :-) is object orientedness bad? abstraction for the sake of abstraction is is anything for the sake of the same thing bad by definition? yes abstraction is a very important tool to allow you to express the solution to problems in a more intuitive and well-defined way in most cases it adds uneeded representation pah and can easily break a unified methodology the problem is more that people work at a too low level in their languages and are unable to abstract abstraction should be a form of a opinion oo was an important step up from asm it shouldn't limit you you can do oo in asm one can build cathedrals with toothpicks as well in the same amount of instructions rather lines using a meta approach a meta approach? you mean abstraction? :-) not really 08:00am sorry, can't really type like this right now damn terminal under windows trying out the latest qnx edemodisk -e how about installing linux+x+gnome? :-) fufie: so do i get things right if tunes is the user interface between hardware and user? (in other words `everything software') ? bsd+x+kde is nicer imo =) tunes would be the environment the user is in yes though I think X really isn't that grat of a desktop X is a windowing system not a desktop I know I am calling it a desktop using a generic term I've programmed in Xlib before have tunes people thought about using robots instead of the common desktop pc and mainframe ? was Xlib a good abstraction level? it's a necessary level I don't think they have a ogood api interface so no smoke: I want to hack lego robots in CL.. but those people only put in 32k memory.. I want 32M so I could do image processing on the fly :) I don't think abstraction in needed ree: so everyone should write directly in molecules? ree: move molecules around in the processor you're taking what I say out of context yes it was meant for a programming level of a system because you stop at some point in your abstraction thinking you have to start somewhere we need abstraction at all levels until we get the level we find suitable talking this way could fill a book fufie it doesn't make sense to write apps in Xlib alone or coincidently a tunes.org site not even Xt alone is good enough that's because the interface is poor to beigin with even gtk is low level there is nothing wrong with low levelness if it gets te job done the cost of that job is the important part fufie: what about making a very good clisp compiler that can put image processing in 32k? depends on your motie (motive) ree: make good and intuitive apps which are powerful and configurable not the best keyboards here to be lacking backspace functionality on * smoke/#tunes wrote an application in Xlib alone smoke: you need the incoming images as well.. fufie, no reason you can't make them using lowl evel semantics a very simple application though :) 08:10am in the same amount of time as a heavily abstracted system fufie: you could parse those in groups of hundred pixels at a time and store intermediate results in a nifty way of storing intermediate results ree: you can do it.. but I prefer doing (format t "hello world~%") instead of writing a hundred lines of asm fufie, you could do that in one line of asm it'd take the same amount of lines of some HLL had they not ha d the library interface ahh.. but libraries are an important abstraction level but in asm you'd have full control over what that function did but they don't need to be implemented in a static pile of garbage they can be a collection of macros function definitions why on earth are you in #tunes ree? they don't have to take up a single instruction until you use them you dislike abstraction oo hll well-defined semantics libraries I dislike the way they are implemented currently you can have well defined semantics (the fact I' m pushing foR) in a low level implementation but, I'll just be quiet about it which hll do you know intimately anyway? c? rumour has it c is a tool like another. you don't want to use a screwdriver for hammering something :) or the standard way of doing things and it can be fast too, but we don't want C as our solution, do we? ;) it's interesting to see the ideas that float by here every noe and then c is a crappy language but c isn't a hll which hll do you know well? that's using a very specific term for a hll I've done rexx, java, shell, php have you used java much? not really just a few programs here and there network apps that communicate on a socket basis with a browser so basically you don't know any general purpose hll by your definition have you tried looking at some modern hll, like scheme, common lisp, ml or haskell? I could pee in the wind and call it rain as well no but I will I think you should do so before you criticise recent development I'm criticizing current development -:- washort [washort@d142.narrowgate.net] has joined #tunes you criticise current language development based on knowledge of rexx, shell and php? not language current widespread implementations what do you dislike about the php implementation then? 08:20am nothing, it fills its purpose but it is just yet another language php is about as modern as pascal as languages go it's c and shell script taped together yes but I still don't get what about hll you criticise their lack of a unified approach obviously you think php is good unified approach? no I never said that well, assembly for example assembly isn't unified assembly is a million dialects every instruction operates on a specific entity it uses the same qualities throughout so does java java has a ton of different properties for each way of doing something almost limitless assembly gives you a common base to work off of so does java yet at the same time stays at the lowest level possible java can be seen as a common base several systems compile to java its base is abstract from the lowest level it is made up of smaller parts that helps java to be cross-platform to avoid the asm-squabbling it is a higher level than assembly assembly is a higher level than microcodes that's because java's main idea is to be portably you can't program in microcode you already tried that conversation before but asm is an abstraction above the microcode hm is this discussion leading anywhere but a mere fight? :) smoke, not really sorry for interrupting :/ not a fight either smoke: I am quite relaxed :) but I know my opinion has no place in tunes glad to hear that :) -- i've run away from windows<>linux wars to fall into language wars :) the asm religion really died in the 70s sometime i love both asm and lisp both sides have equally good points the asm people couldn't understand why people needed functions it all depends on the opinion of the programmer they had jumps it was just three more lines of code.. or a macro 08:30am (hm. i -love- asm and i -like- lisp :) just like c people in the 80s couldn't understand why people needed classes they had structs and they had function pointers no reason you can't combine these technologies into asm and today oo people can't understand why functions are objects and why it is cool to change the language and the compiler and they already have been for awhile Fufie: that's true. =) The Python ppl, in a limited way, seem to understand that functions aren't special and tunes is the next level again where you are able to work with the OS at a higher level well, it was good talking with you ree: heh, have fun :) have to get back to work, unfortunately we haven't implemented ai programs which code for us would it be a wise thing to make some connections between the user and the hardware human? like, there could be a module of tunes named `ree' or `asm hacker'. the user wouldn't notice of course ree: no offense of course :) i like asm hehe humans are the least dependable thing in a computer system, in my experience (on the otehr hand most of that has been tech support) smoke, I'm really interested in implementing many of the ideas tunes has just not using a extremely high level of abstraction -:- SignOff smkl: #TUNES (Ping timeout for smkl[glubimox.yok.utu.fi]) washort: most of the code is human generated anyway even most hardware is smoke: ah. that's different, i suppose wash: yes, Guido knows that, but I think they should get proper closures in python.. ESR mentioned that he wanted to add that -:- smkl [sami@glubimox.yok.utu.fi] has joined #tunes smoke: i just get this picture of a function being called that calls something in module 'asm_hacker' that mails ree and demands to know what to do next washort: i'd like to depend on a Great Asm Hacker more than on a compiler in several occasions washort: completely right! that's what i thought of Fufie: yeah. i wish they had that too