IRC log started Wed May 12 00:00:00 1999
[msg(TUNES)] permlog 1999.0512
-:- SignOff Fare: #TUNES (Read error to Fare[quatramaran.ens.fr]: Connection reset by peer)
-:- SignOff Tril: #TUNES (Ping timeout for Tril[sloth.wcug.wwu.edu])
-:- Tril [dem@sloth.wcug.wwu.edu] has joined #TUNES
-:- mode/#Tunes [+o Tril] by ChanServ
-:- SignOff ruiner: #TUNES (Leaving)
!netgod:*! committing add of irc.linux.com
!asimov.openprojects.net!! SERVER Numeric Collision: becquer.openprojects.net != irc.linux.com
!asimov.openprojects.net!! SERVER Numeric Collision: becquer.openprojects.net != irc.linux.com
!asimov.openprojects.net!! SERVER Numeric Collision: becquer.openprojects.net != irc.linux.com
!netgod:*! gratuitious wallop for no reason whatsoever
-:- AlonzoTG [Alonzo@client-151-200-125-76.bellatlantic.net] has joined #tunes
-:- Fare [rideau@quatramaran.ens.fr] has joined #Tunes
-:- SignOff AlonzoTG: #TUNES (Have Nice Day :))
-:- AlonzoTG [Alonzo@client-151-200-127-40.bellatlantic.net] has joined #tunes
-:- SignOff AlonzoTG: #TUNES (Ping timeout for AlonzoTG[client-151-200-127-40.bellatlantic.net])
-:- SignOff Tikiman: #TUNES (Ping timeout for Tikiman[mcrigal2.student.Princeton.EDU])
-:- Tikiman [tikiman@mcrigal2.student.Princeton.EDU] has joined #tunes
* Tikiman/#tunes is away: (Auto-Away after 10 mins) [BX-MsgLog On]
* Tril/#TUNES is back from the dead. Gone 19 hrs 54 min 51 secs
-:- AlonzoTG [Alonzo@client-151-200-127-40.bellatlantic.net] has joined #tunes
<Tril> y0
<AlonzoTG> wuzzup?
<Tril> trying to figure what to add next to tunes.lsp
<AlonzoTG> what does it do?
<Tril> well it has objects and types, you can put objects in types, create new types, see what objects are in a type, and what types an object is in
<Tril> everything basically just gets a number
<Tril> I might add names next
<AlonzoTG> hmm.
* Fare/#Tunes changed the perms in /cvs
<Fare> I hope it'll work in a more reliable way
<Fare> I think the perms for CVSROOT are wrong, however
<Tril> Fare: I don't want anyone to write to my lisp module
<Fare> Tril: dunno how to do that
<Tril> That's why I made it owned by dem
<Fare> Tril: at one point, nobody could read it, and same for retro/keymaps/
<Tril> the problem is to checkout, cvs makes a lockfile in the directory with the cvs file. But that also allows checkins
<Fare> Tril: maybe you should revert my changes, then
<Tril> did you change tunes.lsp?
<Fare> Tril: I still don't understand how file permissions work under CVS
<Fare> Tril: I think I did kind of a chown -R tunes.tunes *
09:00am
<Fare> l
<Tril> you made all the dirs setuid group
<Tril> I don't care as long as itworks
<Tril> what do you think of the code
<Fare> I have problems with the abstract programming model you implement
<Tril> you mean the functions I provide, new, del, copy, so, and st?
<Tril> Is it a security risk to install a guest user that has a shell of "clisp -i tunes.lsp"? Can someone read/write arbitrary files on the filesystem?
<Tril> hmm
<Fare> I'm disturbed by the notions "add an object to a type" and such
* Fare/#Tunes is trying to get fwprc to work with debian's br0ken telnet
<Tril> adding an object to a type is my entire paradigm.
<Fare> at home, my ISP has been down for 6 days
<Tril> broken?
<Fare> at work, I can't pierce the firewall anymore since I installed debian
<Fare> at uni, well, I was there yesterday
<Fare> and at least it works
<Fare> (plus we have a new printer mucho better than the older one)
<Fare> (maybe I'll go work at uni more often)
* Tril/#TUNES knows nothing about piercing firewalls
<Fare> plus my "autoreconfigure everything when switching networks" scripts don't work anymore under debian
<Tril> If debian's telnet is broken, you should report a bug (first see if it was reported) http://www.debian.org/Bugs
<Fare> Tril: everything is explained in the F-P mini-HOWTO
<Fare> (although I have lots to modify in next version: httptunnel, mailtunnel, ssh, etc)
<Tril> what's wrong adding an object to a type?
<Fare> it's a br0ken notion for a type
09:10am
<Tril> it only works for extensional types, yes
<Fare> or maybe you're not really implementing types, but something lower-level
<Fare> and then, I'd like to understand the high-level thing you're trying to implement.
<Tril> well, this is the low level persistent hyperlinking store.
<Tril> The point is to make it easily extendible so anyone can develop a system based on it
<Tril> think of types are predicates for user-specified data, that can't be recalculated. (That's what they are now)
-:- SignOff AlonzoTG: #TUNES (Have Nice Day :))
<Fare> I think that the most important thing to do would be to write a compiler for a simple lisp language something persistent using your store
<Fare> that'd help you understand the constraints on your store better
<Tril> well that's SORT of the direction I'm going..
<Fare> shit. Too many things to write, and I don't write any
<Fare> I've finally clarified the notion of implementation in a way that suits both traditional languages like the lambda-calculus, and transition systems like the pi-calculus
<Tril> in your mind only?
<Fare> mostly
<Fare> it uses category theory, so it's a bit tough
<Fare> I mean, it uses it heavily:
<Fare> a type is a preorder, considered as a category
<Tril> abi, preorder?
<abi> tril: i haven't a clue
<Fare> (with expressions being inferior to expressions in which they reduce)
<Fare> abi: a preorder is a transitive binary relation
<Fare> abi: forget preorder
<abi> Fare: I forgot preorder
<Fare> abi: a preorder is a reflexive transitive binary relation
<Tril> my text says preorder is a particular recursive algorithm for traversing a binary tree.
<Fare> abi: preorder?
<abi> somebody said preorder was a reflexive transitive binary relation or a category such that there is at most one arrow from any object to any other
09:20am
<Tril> maybe you mean partial order
<Tril> err partial order is an antisymmetric transitive binary relation
<Fare> abi: partial order?
<abi> hmmm... partial order is an antisymmetric preorder, or a preorder in which every object is unique up to isomorphism
<Tril> "Traversing Binary Trees. There are several useful ways to list the nodes of a binary tree.  The three most popular methods are called preorder, inorder, and postorder. The _preorder_ listing of a binary tree has the root of the tree as its head, and its tail is the concatenation of the preorder listing of the left and right subtrees of the root, in that order."
<Fare> abi: partial order?
<abi> i think partial order is (as a binary relation) an antisymmetric preorder, or (as a category) a preorder in which every object is unique up to isomorphism
<Fare> Tril: that's prefix order, infix order, and postfix order
<Fare> Tril: not directly related
<Fare> abi: preorder?
<abi> preorder is a reflexive transitive binary relation or a category such that there is at most one arrow from any object to any other
<Fare> abi: forget preorder
<abi> Fare: I forgot preorder
<Tril> Fare: Your preorder, is it antisymmetric also?
<Fare> abi: forget preorder
<abi> Fare, I didn't have anything matching preorder
<Fare> abi: preorder?
<abi> wish i knew, fare
<Fare> abi: preorder?
<abi> preorder is (as a binary relation) a reflexive transitive binary relation, or (as a category) a category such that there is at most one arrow from any object to any other
<Fare> Tril: a preorder is not necessarily antisymmetric. When it is, it's a partial order
<Fare> so.
<Fare> reduction defines a preorder on expressions of any given type
<Fare> so that every "type" in a given "language" is actually a category
<Fare> now, contexts are arrows from type to type
<Fare> (a context is an abstract syntax tree for that language with a "hole")
<Fare> Tril: still there?
<abi> there are lots of ideas in tunes that could serve you.
<Tril> i am here..
<Tril> i'm not following 
<Fare> ok, let's step back.
09:30am
<Tril> im probably stuck at the preorder part. i dont know what that looks like
<Fare> a language gives you expressions, and types (optionally, a one type for every expression)
<Fare> plus it gives you execution contexts.
<Fare> in a pure language (like the lambda calculus), contexts are just expressions with holes
<Tril> the preorder just says that every type has an arrow to itself (duh) and arrows commute
<Tril> if so, then i understand
<Tril> hole?
<Tril> are you thinking a hole as an L-value?
<Tril> the hole is the execution context?
<Fare> a hole is something to fill with a valid expression of the language
<Fare> (of the right type, too)
<Tril> you stole that idea from me and I didnt even tell you :)
<Fare> for instance, in C, "if ( [] ) { x ++ ; }" is a valid context from type "expression" to type "statement".
<Fare> (drawn as [] in ASCII, but as a square in LaTeX)
<Fare> no, this preorder is internal to every type: it reflects the notion of reduction among expressions of given type
<Tril> what's the deal with "at most one arrow"?  There should be many different arrows from each type
<Tril> i'm not familiar with the use of "reduction"
<Fare> for instance, in LISP, the term ((lambda (x) (+ x 1) (+ y 2)) can be reduced in (+ (+ y 2) 1)
<Fare> reduction, as in lambda-calculus
09:40am
<Fare> the term (let ((y 3)) (+ y 2)) can be reduced in (let ((y 3)) (+ 3 2)) then in (+ 3 2) then in 5
<Fare> a "reduction" is a "computation step"
<Tril> does lisp allow the syntax \x.y arg? I think not
<Tril> anyway i know what you mean, 
<Tril> so reduction is one step in evaluation
<Fare> well, in the categorical view, the preorder is any number of evaluation steps
<Fare> and you can have reduction as an endofunctor of each type
-:- smkl [sami@MCCXII.rdyn.saunalahti.fi] has joined #tunes
<Fare> i.e. a particular function from the type to the type that preserves the preorder
<Fare> haskell accepts \x -> y arg
<Fare> or rather, (\x -> y) arg, to respect priorities
<Fare> so let's use haskell syntax, which is nearer lambda calculus
<Fare> ok, so every type is a preorder, ok?
<Tril> I think you're using type differently from me. I use type as an equivalence relation between its members. So any one can be replaced by another of the same type in a context that depends on that type. But you're talking about something else.
<Fare> no, it's mostly the same thing
<Fare> every given context is typed:
<Fare> it accepts objects of a type as input, and outputs objects of another type as output
<Fare> in category parlance,
09:50am
<Fare> these types themselves are the objects of a larger category,
<Fare> their objects (the expressions) are elements of types as objects,
-:- AlonzoTG [Alonzo@client-151-200-125-149.bellatlantic.net] has joined #tunes
<Fare> and contexts are arrows from types to types that define functors among these types as categories
<Fare> i.e. they are preorder-preserving functions
<Tril> still don't understand the preorder
<Tril> There is a relation between objects in a type
<Tril> what does it do?
<Tril> what does aPb mean if P is the preorder
<Fare> it means that a can be reduced into b by a number of steps of computation
<Tril> then why are a and b in the same type?
<Fare> because reduction preserves typing
<Fare> (a property known as "subject reduction")
<Fare> if a reduces into b, then b has same type as a.
<Fare> or else, the type system is not sound
<AlonzoTG> AIEEEEEEEE!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
<Tril> are all the objects in a type equivalent?
<Fare> (again, if you don't like types, get a one type for everything)
<Fare> no
<AlonzoTG> MY INTERNET BILL FROM APRIL IS $113!!!!!!!!!!!!!!!
10:00am
<Fare> for instance, in Haskell, there is a type "int" with objects the natural integers
<AlonzoTG> thats a PHAT wad of change!
<Fare> (or is it system integers?)
<Fare> ATG: is that a lot more than your usual?
<Tril> my internet bill is $24.95 and they haven't sent it yet. their loss.
<Fare> my internet (cable) bill is around $50, and the service has been down for 6 days!!!
<Tril> $113 is like what your bill is if your ISP is AOL
<Fare> yeah, even cable is less expensive!!!
<Fare> although cable is obviously less reliable!
<Tril> no, AOL is less reliable
<Fare> then it's even worse than I think
<Fare> maybe at least AOL gives you international local line access?
<Fare> (handy for people who move around countries?)
<Tril> aol doesnt go down for 6 days, but it does hangup on you every 2 minutes
<Fare> anyway
<Tril> I'm exaggerating.. it did that on some computers. AOL is evil because it tries to provide content separate from the internet.
<Fare> yeah, AOL is evil, anyway
<Tril> Fare: Program something :)
<Fare> so let's go back to our sheeps.
<Fare> are you ok about the reduction relation and contexts?
<Tril> how does it relate to a type as a precondition for a function argument?
10:10am
<Fare> it relates to types in ML rather than to types in LISP
<Fare> (or types in Coq, too)
<Tril> what about types in coq?
<Fare> that is, by definition, a function will only operate on objects of the right type
<Fare> a type mismatch is unexpressible in the language, and caught at compile time
<Fare> types in coq are even stricter than in ML, so that all computations terminate in finite time
<Tril> only static typechecking?
<Tril> ick
<Fare> well, the notion of type as modeled in category theory only matches static typechecking
<Fare> but it matches it very well
<Fare> (actually, it pre-dates it, I think)
-:- SignOff AlonzoTG: #TUNES (Ping timeout for AlonzoTG[client-151-200-125-149.bellatlantic.net])
<Fare> hum. Still following me?
<Tril> then that notion of type isn't powerful enough for tunes. we need types that are undecidable
<Fare> well, it's a different notion, anyway
<Fare> too bad it's the same, unqualified, word
10:20am
<Fare> maybe use "category" instead for those in Cat theory?
<Tril> english is meant to overload words
<Tril> get used to it
<Fare> ok.
<Fare> So, the contexts themselves constitute the arrows of a category whose objects are these static types
<Fare> and that category characterizes the language
<Tril> maybe Brian will show up here soon, and you can talk to him..
<Tril> sounds a lot like arrows
<Fare> now, let's see the notion of an implementation
<Fare> ideally, an implementation is a "partial functor" from the concrete language (as a category) to the abstract language
<Fare> but since there are only total functions in Cat theory, we instead introduce an intermediate language of observable concrete terms
<Fare> so Abs will be the abstract language, Conc the concrete language in which to implement it, and Obs a sub-language of Conc
<Tril> kk
<Fare> kk?
<Tril> i meant to say 'k' but hit it twice.  K = short for OK
<Fare> k
10:30am
<Fare> well, Obs will be included into Conc by functor such that every static type in Obs is faithfully included into its image in Conc
<Fare> that is, if a reduces to b in Obs iff its image in Conc reduces to the image of b
<Tril> are you going to be able to use this theory to achieve iepos' bootstrap technique?
<Fare> yes and no
<Fare> this theory itself is necessary to it
<Fare> but not sufficient
<Fare> I've met one of the world's best AI specialists in reflective systems
<Tril> and?
<Fare> and he tells me that the indeed logic theory is useful, for it is necessary to express the constraints that reflective systems must fulfill
<Fare> but that the larger part of the system will be made of heuristics and "illogic" parts that do their best to fulfill the constraints (aka solve the problems)
<Fare> if there ain't a logic system to express the constraints, you cannot even express the problems to solve
<Fare> but the logic system is only the language to express the problems, and solving them, in the general case, will require a lot of "illogic" developments
<Fare> to achieve "intelligent" behaviour, or just efficient response to complex problems, you need have introspection:
<Fare> the system must use heuristics to detect when it must be more intelligent, and when it must use brute force instead
<Tril> ok i have to go to class 
* Fare/#Tunes is pleased to see that ircii creates new "windows" inside screen -- yeah!
<Fare> Tril: later!
10:40am
-:- SignOff Tikiman: #TUNES (Ping timeout for Tikiman[mcrigal2.student.Princeton.EDU])
-:- SignOff Fare: #TUNES (Excess Flood)
-:- Fare [rideau@quatramaran.ens.fr] has joined #Tunes
-:- billyboof [hatefull@nrwc-sh9-port3.snet.net] has joined #tunes
<billyboof> whats up everyone
<Fare> lots
<Fare> writing an article on what an implementation is
12:00pm
<billyboof> writing an article on it?  in what?  on the web?
<Fare> first on my computer
<Fare> will put it on the web when readable
<billyboof> ahhh....
12:10pm
-:- iStormy [stormy@rain.futuresouth.com] has joined #Tunes
<iStormy> brethren
-:- SignOff iStormy: #TUNES (iStormy has no reason)
12:20pm
-:- SignOff billyboof: #TUNES (hatefull@antisocial.com)
-:- ruiner [nate@ppp263.wi.centuryinter.net] has joined #tunes
-:- carlito [sabanmr@137.28.109.84] has joined #tunes
<carlito> what's up ruiner?
-:- NetSplit: clarke.openprojects.net split from koontz.openprojects.net [01:13pm]
-:- BitchX+Deb1an: Press Ctrl-F to see who left Ctrl-E to change to [clarke.openprojects.net]
-:- tcn [tcn@cci-209150250111.clarityconnect.net] has joined #tunes
-:- Netjoined: clarke.openprojects.net koontz.openprojects.net
-:- abi [nef@bespin.dhs.org] has joined #Tunes
<tcn> hey abi
<tcn> u suck
-:- SignOff abi: #TUNES (Ping timeout for abi[bespin.dhs.org])
-:- abi [nef@bespin.dhs.org] has joined #Tunes
-:- NetSplit: clarke.openprojects.net split from sterling.openprojects.net [01:19pm]
-:- BitchX+Deb1an: Press Ctrl-F to see who left Ctrl-E to change to [clarke.openprojects.net]
01:20pm
-:- Netjoined: clarke.openprojects.net sterling.openprojects.net
-:- abi [nef@bespin.dhs.org] has joined #Tunes
>>> tcn [tcn@cci-209150250111.clarityconnect.net] requested PING 926541407 319929 from #tunes
-:- SignOff tcn: #TUNES (tcn has no reason)
-:- SignOff carlito: #TUNES (Ping timeout for carlito[137.28.109.84])
-:- carlito [sabanmr@137.28.109.84] has joined #tunes
-:- Tril [dem@sloth.wcug.wwu.edu] has joined #TUNES
-:- NetSplit: varley.openprojects.net split from koontz.openprojects.net [02:03pm]
-:- BitchX+Deb1an: Press Ctrl-F to see who left Ctrl-E to change to [varley.openprojects.net]
-:- SignOff Tril: #TUNES (Ping timeout for Tril[sloth.wcug.wwu.edu])
-lilo(lilo@varley.openprojects.net)- [GlobalNotice] all
-lilo(lilo@varley.openprojects.net)- [GlobalNotice] oops, sorry, I'll try that again
-lilo(lilo@varley.openprojects.net)- [GlobalNotice] Hi all.  A power outage was experienced at Albert's resulting in varley being taken down.  Please note that it's a good idea to find a server near you, rather than clustering on any given server.  For a geographical list of servers,
-:- Tril [dem@sloth.wcug.wwu.edu] has joined #TUNES
-lilo(lilo@varley.openprojects.net)- [GlobalNotice] ....check http://opirc.nu/.  At some point in the near future, whatever alias you use to connect to us will have all the geographical names.  Thanks.
-lilo(lilo@varley.openprojects.net)- [GlobalNotice] I'll be repeating the preceding message in a few minutes.  Thanks.
-lilo(lilo@varley.openprojects.net)- [GlobalNotice] Hi all.  Varley recently died due to a power problem and is back.  Please try to spread out to a server near you; check http://opirc.nu/ for a set of server names.  Sometime soon all of the aliases (irc.debian.org, irc.redhat.com, etc.) will point
-lilo(lilo@varley.openprojects.net)- [GlobalNotice] to the whole list.  Thanks.
-lilo(lilo@varley.openprojects.net)- [GlobalNotice] Again, if you can use a server near you we'll have less commotion when any single server dies.  Thanks! :)
-:- tcn [tcn@cci-209150250096.clarityconnect.net] has joined #tunes
-lilo(lilo@varley.openprojects.net)- [GlobalNotice] Services will be coming back in a moment.  Plan to reenter your nickname password if necessary.  Thanks! :)
-:- SignOff tcn: #TUNES (tcn has no reason)
-:- Netjoined: varley.openprojects.net koontz.openprojects.net
-:- ruiner [nate@ppp263.wi.centuryinter.net] has joined #tunes
<ruiner> where can I get core war?
-:- SignOff Tril: #TUNES (Ping timeout for Tril[sloth.wcug.wwu.edu])
02:30pm
-lilo(lilo@varley.openprojects.net)- [GlobalNotice] Hi all.  Services should be returning momentarily.  If your security setting requires it, expect to have to re-identify yourself to nickserv.  Thanks for your patience!
-:- ChanServ has changed the topic on channel #Tunes to: http://www.tunes.org - Free Reflective Computing System
-:- SignOff smkl: #TUNES (sleeping)
-:- dem_ [dem@sloth.wcug.wwu.edu] has joined #TUNES
-:- dem_ is now known as Tril
-:- timestamp [dancer2@207.16.36.12] has joined #Tunes
<Tril> hi
* timestamp/#Tunes is just a timestamp maintainer client
* timestamp/#Tunes is transitional
<Tril> huh?
<timestamp> long story 8)
-:- AlonzoTG [Alonzo@client-151-200-120-225.bellatlantic.net] has joined #tunes
* timestamp/#Tunes is run by lilo
<Tril> oh hi lilo 
* TUNES is run by Tril
03:20pm
-:- iStormy [stormy@rain.futuresouth.com] has joined #Tunes
-:- SignOff AlonzoTG: #TUNES (Ping timeout for AlonzoTG[client-151-200-120-225.bellatlantic.net])
<iStormy> hi
03:30pm
<Tril> 
<ruiner> hmmmmm
03:40pm
<iStormy> anyone know what point size a 13-pixel high font is?
<iStormy> 8x13
<Tril> uhh
<ruiner> nope, sorry
03:50pm
<Tril> as i understand they are fundamentally unrelated. point size is for size on paper fixed in inches or something and has nothing to do with size in pixels
<Tril> i'm just guessing i know nothing about fonts
04:00pm
-:- SignOff ruiner: #TUNES (Leaving)
-:- SignOff iStormy: #TUNES (iStormy has no reason)
-:- _QZ [brand@p0wer.qzx.com] has joined #tunes
-:- iStormy [stormy@rain.futuresouth.com] has joined #Tunes
-:- AlonzoTG [Alonzo@client-151-200-126-16.bellatlantic.net] has joined #tunes
* AlonzoTG/#tunes slaps iStormy around a bit with a large trout
-:- ruiner [blah@ppp306.wi.centuryinter.net] has joined #tunes
-:- SignOff ruiner: #TUNES (Ping timeout for ruiner[ppp306.wi.centuryinter.net])
* Tril/#TUNES  is away: (afk) [BX-MsgLog Off]
-:- Fare [rideau@quatramaran.ens.fr] has joined #Tunes
* Fare/#Tunes is back
<Fare> Tril: around?
<Fare> Tril: I simplified the notion of a language, and as a result, the notion of type I get is arbitrarily expressive
<AlonzoTG> I'm going to mail out my draft specs to the prototype tonite!!!!
<Fare> (although still static types)
<Fare> so a language is the data <E,C,R>, of a category E of terms, whose arrows are reductions,
<Fare> of a category C of types whose arrows are contexts, and of a realization functor R:C->E->2 that maps observable types to actual "types" of E in a way compatible with subject reduction
<Fare> a morphism PHI of languages L and L' is the data of usual morphisms for the first two components E and C, i.e. functors from E(L) to E(L') and from C(L) to C(L')
<Fare> as well as for the third component,
-:- SignOff Tril: #TUNES (Ping timeout for Tril[sloth.wcug.wwu.edu])
<Fare> of a natural transformation from R to R'\circ (R(L) \otimes R(L'))
<Fare> and there you are!
<Fare> an implementation of an abstract language A with a concrete languauge C is the data of an intermediate language O, subset of C (as given by an embedding language morphism j:O->C),
<Fare> and of an interpretation function from O to A (as given by a language morphism phi:O->A).
<Fare> once you have that, everything works fine.
05:40pm
<Fare> hum.
<Fare> anyone reading?
<Fare> anyway, that's in the log, so it ain't lost.
<AlonzoTG> fare:
<AlonzoTG> Could C be implemented in a an FSM in hardware?
<Fare> sure
<Fare> uh, which "C" are you talking about?
<AlonzoTG> So that an OS runnig O  could load program written in A...
05:50pm
<Fare> oh, yeah, sure.
<AlonzoTG> Or just implement A in hardware...
<Fare> Usually, you don't choose A and C, only O
<AlonzoTG> Describe A and I'll take a shot at it... :)
<AlonzoTG> Anyway...
<Fare> A would be the Tunes HLL, C would be the hardware, and O our implementation
<AlonzoTG> okay...
<AlonzoTG> I'll get back to finishing the draft of my specs...
<Fare> to start with, take A=Scheme, extended with persistence and multithreading
<AlonzoTG> and I'll post it to Unios@tunes.org...
<Fare> why not to specs@tunes.org (if it still exists)
<AlonzoTG> hmm...
<AlonzoTG> I don't think I am or want to be on that one...
<Fare> btw, our GNU/Linux emulator shall be called PIG when it exists -- PIG IS GNU
<Fare> well, send me a Cc: then, for I'm not subscribed to unios@tunes.org; I only read it on the web.
<AlonzoTG> hmmz...
<AlonzoTG> maybe..
<AlonzoTG> It'll be on my website at atc.peon.net/~cyborg/OS/doc/ I think...
<Fare> ok
<Fare> gotta go now
<Fare> bye!
-:- SignOff Fare: #TUNES (Connection reset by pear)
06:00pm
-:- carlito [sabanmr@137.28.129.23] has joined #tunes
-:- carlito [sabanmr@137.28.129.23] has left #tunes []
<iStormy> i'm reading about smalltalk...it's neat (only know c/pascal/basic-style languages)
06:40pm
* iStormy/#Tunes installs squeak 2 to putter with.
<iStormy> wow, it's big.  4.8mb
<_QZ> hmm
<AlonzoTG> ITS UP!!!
06:50pm
<iStormy> "A general whisper ran among the country people, that Sir Roger was up."
<_QZ> "implementable" is that a word?
<iStormy> i've use it, but it's not in my dictionary
<iStormy> which is ancient
<_QZ> s/ultamately/ultimately/
<_QZ> s/controll/control/
<iStormy> are you in the right window?
<_QZ> caps should only be used to start a sentence and not to make the sentenece wave
<_QZ> i am correcting some problems with atg's specs
<_QZ> http://atc.peon.net/~cyborg/OS/docs/specs.txt
<iStormy> ah
<iStormy> i knew no one had said those words on this channel
<_QZ> :)
-:- AlonzoTG has changed the topic on channel #Tunes to: atc.peon.net/~cyborg/OS/docs/specs.txt
<_QZ> please put the tunes url back in the topic
<iStormy> that isn't an url
<iStormy> since we are correcting.  (c:
<_QZ> u can add to the topic but do not annihilate it
<AlonzoTG> hey, it works from me...
07:00pm
<iStormy> *** The topic on #Tunes is "http://www.tunes.org - Free Reflective Computing System".
-:- AlonzoTG has changed the topic on channel #Tunes to: www.tunes.org |   atc.peon.net/~cyborg/OS/docs/specs.txt
<iStormy> that's what it was
-:- _QZ has changed the topic on channel #Tunes to: http://www.tunes.org - Free Reflective Computing System | atc.peon.net/~cyborg/OS/docs/specs.txt
-:- _QZ has changed the topic on channel #Tunes to: http://www.tunes.org - Free Reflective Computing System | AlonzoTG's OS specs atc.peon.net/~cyborg/OS/docs/specs.txt
-:- iStormy has changed the topic on channel #Tunes to: http://www.tunes.org - Free Reflective Computing System | AlonzoTG's OS specs http://atc.peon.net/~cyborg/OS/docs/specs.txt
<_QZ> heh
<iStormy> lets be consistent
<_QZ> AlonzoTG: if i remember correctly u store all data inside the program right?
-:- tcn [tcn@cci-209150250089.clarityconnect.net] has joined #tunes
<_QZ> hola tcn
<tcn> hey
<abi> hi, tcn
<tcn> what's up here?
<_QZ> just discussing atg's os specs
<_QZ> well sorta, gotta fix all teh errors in it first :)
<_QZ> "The User is ultamately in controll not the OS. :) Just yesterday I was going to the bathroom when this great idea hit me."
<tcn> ehhe
<tcn> atg: lemme see it
<_QZ> topic
<_QZ> everyone beep atg until he gets back
<tcn> heh
07:10pm
<tcn> my system is a heirarchy of graphs
<_QZ> heh
<tcn> in other words it's a graph
<_QZ> alonzo?
<abi> rumour has it alonzo is hooked on phonics worked for me and it can work for alonzo too
<_QZ> stupid abi
<tcn> haha
<_QZ> how do u make abi say stuff
<_QZ> abi: hehe
<_QZ> ok so abi doesnt do that anymore but how do u make abi respond to that
<_QZ> like 'hey' and 'hello'
<tcn> <reply>
<tcn> abi: die is <reply> ok i'm dead
<tcn> abi: die!
<abi> ok i'm dead
<_QZ> alonzo
<abi> hooked on phonics worked for me and it can work for alonzo too!!
<_QZ> :)
<_QZ> abi: i love u
<tcn> abi: i love u
<abi> i love you too
<tcn> haha
<_QZ> hehe
<_QZ> abi: i dont love u
<tcn> abi: i love u
<abi> i love u 2
<tcn> haah
<tcn> that makes more sense
<_QZ> did u ut that on 'i love u'?
<_QZ> put
<tcn> heh.. atg beeped me
<tcn> yeah
<_QZ> AlonzoTG: if i remember correctly u store all data inside the program right?
<_QZ> answer me damnit
<AlonzoTG> I don't know...
<AlonzoTG> I havn't specified that yet!
<_QZ> u said u store all data inside the program
<_QZ> u said there was a filesystem inside the program
<tcn> heh.. atg wants to be a software engineer or something.. "I don't write it I just design it"
<AlonzoTG> I'm not sure yet...
07:20pm
<tcn> "it's up to the technicians to write it, I don't have to know how to do that"
<tcn> :)
<AlonzoTG> I'll explain after I'm finished distributing it to everybody on the internet individually...
<AlonzoTG> brb.
<_QZ> get yer ass back here
<tcn> I always thought OSes should be designed to run a real computers
<_QZ> heh
<tcn> s/a/on/
<tcn> hmm
<tcn> He actually has some of the same ideas I do
<_QZ> he stole them
<tcn> I never mentioned them
<_QZ> OH MY GOD HE STOLE MY MULTITHREADING IDEA, THAT BASTARD
<tcn> haha
<iStormy> squeak2 source download up to nearly 11 meg
<_QZ> u need squeak to compile squeak2
<tcn> I was thinking, link the objects together as a graph, but use a minimum spanning tree to make system management easier..
<_QZ> huh
<_QZ> a graph?
<tcn> a tree that contains some loops
<_QZ> what purpose would this serve?
<tcn> so you can have circular references
<_QZ> for what purpose
<tcn> AI, for one thing
<tcn> in a graph system, any object can reference any other object, not just objects below it
<_QZ> uhh
07:30pm
<_QZ> i dont have any graph and my objects can reference any damn object they want
<tcn> oh I'm currently using a tree.. that's all I see a need for now
<AlonzoTG> om
<tcn> ITS allowed programs to make up their own "hardware" interrupts
<_QZ> hehe
<_QZ> tcn: are u talking about his html example?
<tcn> sort of
<tcn> i'm not sure if that's what he has in mind
<AlonzoTG> fuck my sendmail server is not working...
* iStormy/#Tunes wonders how many days it will take a 96MB p133 to compile squeak2.
<tcn> what's a Currior?
<AlonzoTG> I was trying to figure it out myself...
<_QZ> if i embed image A into document B and document C i can relate the two docs thru image A
<_QZ> is that what yer talking about?
<AlonzoTG> I had written it into the source...
<tcn> s/Currior/Courier/ ?
<AlonzoTG> yes
<tcn> I think a currior is a machine that makes curry :)
<AlonzoTG> something to use as a way to seperate two modules...
<_QZ> haha
<tcn> kinda like a bread machine
<iStormy> a currior is an indian delivery boy
<_QZ> My hope is that it will allow radically new types of devices and
<_QZ> interfaces to be developed. That couldn't otherwise be supported by a
<_QZ> traditional system.
<_QZ> he wants his computer to make bread :)
<AlonzoTG> cybernetix.
<AlonzoTG> :)
<iStormy> wow, the squeak compile is done...not much C to it.
07:40pm
<tcn> nope, just a VM
<_QZ> yer magic number is 69? is that the position u and yer father use?
<iStormy> wowsers...neat
<iStormy> slow, but noat
<iStormy> neat
<iStormy> 25mb ram
<tcn> haha
<tcn> I used it for an hour or so
<iStormy> 10mb resident
<iStormy> i doubt i'll use it that long
<iStormy> just wanna tinker with it
<tcn> try writing a program
<tcn> heh
<tcn> squeak gives persistent storage a bad name
<tcn> stores everything in one big image file and makes it hard to export anything
<tcn> haha
<tcn>  On The Mosquito Upgrading.
<tcn> huh huh
<AlonzoTG> well it's smaller than a fly! 
<tcn> bad fucking pun man
<tcn> :)
<_QZ> alonzo
<tcn> you can do that in an academic paper though. I've seen it before
<iStormy> i locked it up
<tcn> haha
<tcn> maybe it's just being slow
<iStormy> and again
<iStormy> i click too fast, and it sits at 90% cpu
<_QZ> abi: alonzo
<abi> hooked on phonics worked for me and it can work for alonzo too!!
07:50pm
<tcn> AlonzoTG: why use FAT?
<_QZ> two words: dos clone
<tcn> In some ways he's getting on the right track
<_QZ> i hate form failures :)
<tcn> html forms?
<_QZ> Name: Config Default.
<_QZ>    Why: Used to recover form failures due to configuration problems.
<AlonzoTG> my sendmail server is down!!! :(((((
<AlonzoTG> Porotype = use old technology whenever possible.
<_QZ> Preforms: Reads or generates ....
<tcn> AlonzoTG: use someone else's mail server
<tcn> most servers refuse relaying now.. thanks to fucking spammers
<AlonzoTG> huh?
-:- dreycon [jgraham@polly.rap.ucar.edu] has joined #Tunes
<_QZ> did u hear that some luser from hotmail has joined CAUCE
<iStormy> welp, i'm getting nothing accomplished with squeak.
<tcn> cauce?
<tcn> iStormy: is there a better smalltalk implementation?
<tcn> (not a toy :)
<iStormy> tcn: i dunno...i've never used smalltalk
<_QZ> Coalition Against Unsolicited Commercial E-Mail
<AlonzoTG> om
<_QZ> Hotmail director Randy Delucchi has been appointed to the board of CAUCE
08:00pm
<_QZ> Delucchi was 
<_QZ> later quoted saying that he is "looking forward to sharing" his vast 
<_QZ> knowledge of spam prevention with the rest of CAUCE's board.
<_QZ> maybe he could start by killing off hotmail
<tcn> haha
<tcn> ahhhhhhhh.. college is pissing me off
<tcn> english 220.. fucking bs course
<_QZ> been there done that, dont gotta do it anymore :)
<tcn> same here
<tcn> 1 more week
<tcn> err, 2 more days
<tcn> and a final
<_QZ> no its not the same, if it was then u wouldnt be bitching
<AlonzoTG> my ISP is on buck an hour rate....
<AlonzoTG> gotta quit. :*(((((
<AlonzoTG> final comments before I go?
<_QZ> AlonzoTG: u lie
<tcn> how come u were on here for 10 hours the other day?
<AlonzoTG> that was before I realized I was nearing my threashold...
<AlonzoTG> :(
<_QZ> AlonzoTG: if that was true then u would be paying more per month that i do
<AlonzoTG> I hit my threashold about an hour ago...
-:- SignOff dreycon: #TUNES (Leaving)
<tcn> oh, fucking bellatlantic
<tcn> those bastards
<AlonzoTG> yep!
<AlonzoTG> goodnyte!
<tcn> switch
<_QZ> ya switch
<_QZ> right now
<tcn> that's my final advice for the day :)
<AlonzoTG> or upgrade to ISDN....
-:- AlonzoTG [Alonzo@client-151-200-126-16.bellatlantic.net] has left #tunes []
<_QZ> get aol
<tcn> haha
<tcn> u calling him a lamer?
<_QZ> :)
<tcn> iF 0n1y h3 w08ld qu1it ta1k1ng l1k3 th1$
<_QZ> haha
<tcn> aNd thiS
<_QZ> he even does that in his specs
<tcn> didn't anyone ever teach him, always use lowercase
<_QZ> and specs are suppose to be a tool to sale yer ideas to others
<tcn> d33z d00deZ
<tcn> heh
<_QZ> (This beats the shit out of
<_QZ>    LiNuKs)
<_QZ> and dont dis linux
<tcn> hey if he types like that the only people receptive to his ideas will be waR3z pyRat3z
<_QZ> ucna point out the flaws with it but u cant dis it
<_QZ> s/ucna/u can/
<tcn> yeah, it's the best thing we got going
08:10pm
<_QZ> i cant wait to use the REED function for the console module :)
<tcn> hey if u can REED an object and u can MURDER it, can you REEM it?
<_QZ> hahaha
<tcn> instead of saying kill -9, u say LOUIMA
<tcn> or whatever that guy's name was, in nyc
<_QZ> nyc?
<tcn> u know, the guy who got butt-shanked by the cops?
<_QZ> uhh
<tcn> they talk about it on the radio all the time here in NY
<_QZ> i dont live in ny
<tcn> i thought it was national news
<_QZ> i dont watch much tv
<tcn> you know how they shot some guy 40 times this year?
<_QZ> the police in slc have shot about 10 ppl this year
<tcn> a couple years ago, they sodomized a guy with a plunger
<_QZ> they shot to kill 10 ppl that didnt need to be shot
<_QZ> they put 22 rounds into this guy that had a pair of scissors
<tcn> the nurses at the hospital said he had like, a torn colon, anal splinters, etc..
<_QZ> he wanted to commit suicide and he was swinging them around and they all opened fire 
<tcn> damn
<tcn> suicide by cops?
<_QZ> assisted suicide
<tcn> haha
<_QZ> they shouldnt have shot at all
<_QZ> and if they outright had to it shoulda been in the knee
<_QZ> and only one bullet not 22
<_QZ> and this other cop who is now in jail pulled this guy over, walked upto the car and blew him away
<tcn> I heard the spanish call them 'los tiros' - the shooters :)
08:20pm
<tcn> oh well
<tcn> i'm callin it quits
<_QZ> a couple cops get there ass killed by a couple ak47 toten criminals and now the cops have happy trigger fingers
<tcn> hey, i caught my insurance co. trying to dick me over today
<_QZ> get a better company
<tcn> listing me as 'under 21'
<tcn> i'm gonna ask them to list me as a married female over 30
<_QZ> hahahaha
<_QZ> how much u pay on what kinda car?
<tcn> saturn $1250
<_QZ> no way
<tcn> you?
<_QZ> $684 on a 1999 sunfire
<tcn> saturn's like the cheapest car to insure, too
<tcn> what company?
<_QZ> allstate
<tcn> liberty mutual
<_QZ> allstate is the best
<tcn> geico seems good
<tcn> someone insured under geico hit my sister's car.. geico helped her out and her own company didn't do a fuckin thing
<_QZ> do u have a bad driving record?
<tcn> nope
<tcn> clean
<tcn> they raised the premium about 15% this year, too. fucking bastards
<_QZ> i was hit serveral years ago and the chicks insurance paid for my car and gave me a 95 grand am for a month
<tcn> sweet
<_QZ> oh i guess it woulda been in 95 so the car was new too
<tcn> my sister got a 98 tracer.. ok, but no balls
<_QZ> why did u get a saturn?
<tcn> oh, my mom got one in 93 i think.. it was pretty good..
<_QZ> did u know that pontiacs dont break down?
<tcn> ahhahaa
<tcn> pontiacs make fords look good :)
<_QZ> and the way saturn brags about the millions of repair places all over it sounds like they break down alot :)
<_QZ> pontiac doesnt have repair places cuz they never break
<tcn> the saturn's allright, but kinda flaky.. rattles alot.. some annoying electrical problems..
<tcn> tell that to my friend who has a 94 grand am
<_QZ> my mother has a 94 grand am
<tcn> the main good thing about saturns is, they don't have many places that can rust.. plastic sides, stainless exhaust..
<_QZ> what did u pay for it and what year?
08:30pm
<tcn> oh, it's a 92, paid $5000 a few years ago
<_QZ> oh my god
<tcn> what
<_QZ> a 92 and u pay $1200?
<tcn> yeah!
<_QZ> u got a few dui's under yer belt or what?
<tcn> clean
<tcn> I think it's because of New York
<_QZ> damn
<tcn> they dick you hard in this state
<_QZ> oh ya u have all those drive by shootings
<tcn> I live 3 hours from NYC
<tcn> everything costs too much here
<_QZ> do u have dsl?
<tcn> taxes, insurance, phone, electric.. ELECTRIC
<tcn> no dsl
<tcn> electric - $.15/kW
-:- core [core@mcp.suntech.fr] has joined #tunes
<_QZ> what do u pay for car taxes?
<_QZ> core!
<core> people hi
<tcn> hey core
<core> hey Brand :)
<core> hey Tom :)
<core> wow, actual people in here for once :)
<tcn> car taxes? huh?
<core> abi kind of has a limited conversation
<_QZ> do u pay car tax?
<tcn> some way or other, I pay
<_QZ> licensing
<tcn> what's the tax in utah?
<core> french people pay tax on [insert anything you can imagine]
<tcn> hehe
<tcn> Europe is like NY only 10 times more :)
<core> we'll probably pay taxes on taxes soon
<tcn> haha
<core> tcn: some like that :)
<_QZ> i was paying $45 for my last car but im told it will now be $150
<tcn> $45/year?
<core> qz: get a bike :)
<_QZ> ya
<core> qz: or release brix, become rich and famous, and you can afford the tax :)
<tcn> well we pay 8% sales tax up front
<_QZ> 6.3%
<core> tcn: france is 20.6%
<_QZ> core: u lie
<core> (vat)
<core> qz: no, that's how much vat we have.
<_QZ> vat?
<tcn> In sweden I hear you pay more than half your income in taxes :)
<tcn> value added tax
<core> qz: value added tax. sales tax
<_QZ> my god
<core> tcn: for a $2K salary a month, employers shell out $3K in france; we're getting there :)
<tcn> you know how companies are always talking about 'value added products' these days? like that ;)
<_QZ> nevada has no tax
<core> qz: well i'm in the US right now, so ha :)
<_QZ> its weird to just buy shit for what it costs
<tcn> core: That's what income tax is here
<core> tcn: what, 20.6%?
<tcn> 32% (plus state tax) if you make more than $30,000 a year
<core> tcn: that's not that bad
<tcn> but sales tax is "only" 8%
<core> i guess i'll just have to become rich with clementine </sarcasm>
<_QZ> move to nevada
<_QZ> they dont pay any tax
<tcn> move to Christmas Island :)
<_QZ> they got all the lusers at the casinos
<tcn> oh yeah, tax for morons!
<tcn> (in NY we have the lottery)
<_QZ> we have bingo here :)
08:40pm
<tcn> we have indian casinos now too
<tcn> sellouts :)
<core> what about all the three of us buy a country and declare no tax for OS developers?
<tcn> I heard there are some islands near Maine that seceded from the US and never got taken over :)
<_QZ> how do we get our money back that we bought the island with
<tcn> don't buy it, take it :)
<core> qz: we write code that's impossible to understand, then we charge people for support?
<tcn> oh yeah
<_QZ> ok lets get all the os developers to collect arms and build bombs and shit
<tcn> we'll become arms developers
<core> or we can make the OSes non-year-2038 compliant and then charge for year 2038 compliance
-:- hcf [nef@me-portland-us839.javanet.com] has joined #tunes
<_QZ> heh
<core> although i'll be 62 by then, so it's kind of late
<_QZ> hcf!
<hcf> _QZ!
<tcn> Just say 2112-compliant cause that's when the Maya calendar ends.. people will pick up on that in about 10 years..
<tcn> howdy hcf
<core> lol tcn
<core> hi hcf
<hcf> hoy ppl
<core> tcn: well, that's a bit late for charging for fixing stuff. what about 2003-compliance?
<tcn> hehe
<core> we'd need to count system ticks starting from april 17, 1933 or something, but it can be done.
<_QZ> year 2001 compliance
<tcn> someone my mom knows got a letter from their town.. "the power company has informed us to expect 2-3 days of outages in 2000, but possible 2 to 6 months"
<core> heh
<core> time for them to figure out how to install a linux or *BSD cdrom?
<tcn> makes you wonder
<_QZ> utah is ahead of all other states in y2k compliance
<core> qz: you're cheating, they don't have electricity
<_QZ> except for the city of protection
<tcn> I hope New York loses all our records in 2000
<core> there are lots of amish people here, i suppose they didn't pass year 1900 compliance, they're stuck in 18xx.
<_QZ> hahaha
<tcn> in france??
<core> tcn: no, where i am right now; iowa :) (for a few more days anyway)
<tcn> oh.. hehe
<tcn> no wonder you're on this late at night :)
<core> yeah.. if i was up at 4:39 AM on a week day you could start being worried about me
<_QZ> come to utah
<core> that's not that far away
<core> tcn: still planning to use xcom? :)
<tcn> core: yeah
<core> tcn: hmm, i suppose i should send you updated stuff then :)
<tcn> sure, the format, the NASM patch, etc...
<core> i could send you a whole Clementine archive but it's doing all sorts of things at the moment :)
<core> sure :)
<core> there's also gdb source-level debugging support if that's useful :)
08:50pm
<tcn> Never used gdb :)
<core> i'm more of a printf debugging person but some people like it i guess
<tcn> same here
<core> i don't use it much, only to show off :)
<iStormy> i've always used prints too...saves learning gdb
<tcn> hahaha
<iStormy> and you always think it'll just take a sec to add that print...and you end up putting in 40
<core> starting clementine, breaking into the debugger and having a linux box linked on the serial port, displaying all internal structures in ddd, is kind of amusing
<core> istormy: that's why i use DEBUG and ASSERT.
<core> istormy: the prints are there, they are just conditionally compiled or assembled.
<tcn> hmm
<tcn> I should make macros that do the $ifdef/$endif automatically
<core> tcn: that's quite timesaving
<tcn> right now I don't do anything special
<core> tcn: i used to remove the debug code after i was done with some part, but i usually put it back in sometime later when i broke it, so.. :)
<tcn> hehe, yeah
<core> for some reason, samuel l. falvo is writing an xcom linker, that should come in handy for multiple archs and stuff.
<tcn> is he using C?
<core> i suppose.. i mean linker as in, like, gnu ld
<core> that would run on *nix, cross-platform.. like i currently use gnu ld.
<tcn> I'm thinking of a 'loader'
<core> oh.. well, everyone will write their own xcom loader i guess
<core> it's quite simple
<tcn> yeah.. I'll use asm
<tcn> Shouldn't be much harder than the RDf loader
<core> not really
<core> you probably saw the xcom loader in the clementine shuttle component, it's quite small
<tcn> first I'll write it as an RDF module, test it out, move it into the bootloader, and stop using RDF entirely
<core> yeah.. that's probably a good plan
<core> you can even maybe try to tinker with it under linux :)
<tcn> what's ld do with xcom's.. just combine them into libraries?
<core> oh.. no, it links several xcom objects into one xcom object
09:00pm
<core> like gcc and nasm generated files into one
<tcn> oh ok
<core> that's kinda useful :)
<tcn> just being able to use gcc will be a big deal :)
<core> i know _qz will disagree :)
<core> yeah.. i moved much faster once i was able to do C
<core> i remember that was my milestone #1
<tcn> i don't wanna fall back on C, though.. that's mainly just to import GGI drivers
<core> my C is what fare calls "portable assembler"
<tcn> I wanna build everything from the ground up, as much as possible
<core> it almost compiles 1:1 to assembly opcodes :)
<core> yeah, i understand :) i'll probably rewrite a few bits as assembler once most algorithms are settled.. the C version still being here for other archs until they get their own.
<core> although gcc does a pretty good job if raped properly
<tcn> haha
-:- billyboof [hatefull@nrwc-sh8-port175.snet.net] has joined #tunes
<billyboof> whats up everyone
<core> most people don't know you can tell it to pass parameters in registers for internal functions and things like that :)
<core> hi billyboof, not much.
<tcn> sorry, I gotta leave soon
<core> tcn: the xcom archive is on bespin if you missed that
<tcn> core: I got it
<tcn> thanks
<core> ok :)
<core> tell me if you have trouble with it.. hopefully i packaged everything you need
<core> i'll give you a full clementine archive RSN
<tcn> allright, cool
<tcn> I've got enough here to write the loader.. hmm, prolly after I write an IDE driver
<billyboof> tcn: ypi
<billyboof> err
<billyboof> your writing an os?
<core> yeah.. you'll have the clementine components as reference soon :)
<tcn> yup.. it's coming along pretty good
<billyboof> yeah... same here... how far are you?
<core> billyboof: any documentation/anything online?
<tcn> see http://tunes.org/~tcn/retro.html
* core/#tunes enters os-review-maintainer mode
<tcn> hehe
<billyboof> core: no .. i haven't put anything online... mostly just my own comments for myself... :)
<core> billyboof: ahh.. ok.. any general principle behind it or just tinkering with it for fun?
<billyboof> or all in my head... haven't spent time on documenting anything for others yet... mostly working on research/coding...
<billyboof> what do you mean general principle?
<core> tcn: btw don't worry much about GGI, you mostly have to write one file and you're set
<tcn> well, I gotta hit the sack. too much shit to do for college..
* iStormy/#Tunes is working on his language again...learning smalltalk at the moment to broaden his horizons.
<tcn> see ya all
<core> billyboof: like a brand new idea on how to implement or organize it :)
<core> bye tcn, goodnight :)
<iStormy> i'm kinda like fare in that way...i wanna write an os.  so first i have to design a computer worth programming.  (c;
09:10pm
-:- SignOff tcn: #TUNES (tcn has no reason)
<billyboof> well i got many of my idea's from a book i had borrowed from a friend... but for the most part haven't been implemented...
<core> istormy: yeah, that makes sense, it's just a titanesque job
<billyboof> going to be document based... none of this shit organization windows/linux has...
<billyboof> for the filesystem anyhow...
<billyboof> internally the root dir will have the bootloader and two directries.... one for code and one for docs....
<billyboof> to the user the root dir will be the documents dir...
<billyboof> and the programs shit won't be a part of the normal filesystem....
* iStormy/#Tunes has considered doing away with the hierarchal filesystem...but he can't conceive of any other way to do it.  (c;
<core> i don't know, i think a filesystem is a pretty nice abstraction
<billyboof> i'm not doing away with the hierachy, just lessening it by a lot...
<core> persistent storage is needed, but that shouldn't mean to do away with an actual FS
<billyboof> also, "plugins" instead of applications.... 
* iStormy/#Tunes growls...dangblasted typing injuries.
<billyboof> universal doc format which will hold objects of different types <images/rich text/tables/etc>
<core> yeah, that seems to be the direction most new projects are going (components)
<billyboof> dunno what else to say... just working on memory management at the current time..
<billyboof> then going to work on an IDE driver...
<core> paged? linear?
<core> there are some documents at tunes.org about mm if you need..
<billyboof> paged
09:20pm
<core> ah ok :)
<billyboof> and segmented... :)
<core> aw, isn't that nonportable/more complex?
<billyboof> i'm not looking to make this portable...
<core> why not? current ia32 arch will last ~5 years
<billyboof> segmentation won't go away though..
<billyboof> neither will intel machines... will just change slowly...
<billyboof> for major things like that anyways...
<core> i'm not sure ia64 will have segmentation :)
<billyboof> i think it will simply because of this....
<billyboof> data verses code... one can be read/write, other read/execute...
<billyboof> it's too much to define this for each page...
<core> paging does that too :)
<core> not really, if you use paging, it's embedded into page tables anyway
<billyboof> hmm... i don't know that much about paging.. just got it to work .. :)
<billyboof> anyways... here's what will be segmented:
<billyboof> for a task: 1 code seg, 1 dataseg, 1 segment for each file open...
<billyboof> the files open will be within the task's LDT
<core> well you have a read/write bit and an user(ring3)/supervisor(ring0) bit in every page entry
<core> what about memory mapped io or shared memory regions?
09:30pm
<billyboof> that's something that seem's like it's going to be a pain in the ass...
<core> well if you stick to segmentation you need additional selectors for it
<billyboof> it seems like every used page will have two page table entries... one for the task itself, and one for the os
<core> why? the os (if you mean kernel code) and a task usually share the same address space
<billyboof> for file io i basicly plan on having an OpenFile proc that'll give the task a new entry in it's LDT, and basicly do demand paging on the file....
<billyboof> wouldn't that make it so any task could screw up any other task, as long as they were the same priv-level?
<core> no, every task has its own address space; i meant, the kernel code is present in every at the same address, in pages that can't be touched from usermode
<billyboof> so basicly i should reserve the first page table directory for the kernel <4mb> and have every main page directory point the first page dir to the kernels list....
<billyboof> and remap extra space to higher addresses...
<core> if you plan to limit your kernel to 4 mb that could work yes
<billyboof> and then for a task have the second and on point to it's own memory space...
<billyboof> i mean for now... that could change of course :)
<core> yeah.. usually paranoid OSes like NT or linux use a 2G/2G or 3G/1G scheme on ia32
<billyboof> 2g/2g?  3/1 ?
<billyboof> what do you mean?
<billyboof> like the first 2gig of addresses reserved for the kernel??
09:40pm
<core> the last 2 or 1 gig reserved for the kernel yeah
<core> that leaves 3 GB for every task
<billyboof> hmmm.... that's a different way to think about it...
<billyboof> paging is kinda hard for me to understand i guess...
<billyboof> i for some reason i think like for my system, 4mb reserved = 28mb left for tasks.. :)
<billyboof> keep forgetting that leaves 4092mb left for tasks :)
<billyboof> i think i can handle setting aside like 128meg or so :)
<core> well, that's 1 virtual gigabyte, not actual physical memory
<core> you don't have to map all of it :)
-:- SignOff hcf: #TUNES (Leaving)
<core> probably as the system runs, task addres spaces will be made of physical pages grabbed from memory all over, not contiguously
<billyboof> so nt/linux have page tables for the whole 4 virt gigs?
<billyboof> yeah i figured that... :)
<core> they add page tables when they need.. otherwise that'd mean using 4 MB for every address space
<core> most tasks don't need anywhere as much.. and if you have some way of verifying code (like java, although that's a poor example) you could even run them in the same space
<billyboof> actually i was thinking that since everything does need to be run in a segment, just use the main 4mb page tables for everything, and actually use the segment offset/limits...
<billyboof> for an apps cs/ds...
<core> well, if you use segments and don't plan to use more than 4 MB per task, you don't even need different address spaces
<billyboof> no... i ment 4mb's the page tables take up ... like the same memory map..
<billyboof> ie cr3 on the kernel = cr3 in a task
<core> well, you don't need to actually use those 4 MB at once.. that'd be mapping the entire 4G
<core> basically if you're about to map pages and the table for them doesn't exist, then you create it
<core> but yeah cr3 for the task and the kernel is the same. you can limit access either through segmentation or through paging; there are bits in page tables that restrict access to ring 0-2 code.
09:50pm
<billyboof> yeah... i think i'm going to use the segmentation for access limitations..
<billyboof> ok ... so for keeping track of physical memory, should i just have like a bit table to say if the page is mapped or not?
-:- SignOff iStormy: #TUNES (iStormy has no reason)
<core> that's one of the ways to do it yeah
10:00pm
<billyboof> or have a dword for each page of physical memory that a timer of when it was last used, and reserve like 0xFFFFFFFF for unused
10:10pm
<billyboof> oh well i'm going to sleep ... talk to you tomorrow...
-:- SignOff billyboof: #TUNES (zzzzzzzzzzzz)
10:20pm
<_QZ> crap
<_QZ> core: wanna make some money?
-:- carlosflo [nfer@206.84.93.152] has joined #Tunes
-:- SignOff carlosflo: #TUNES (Leaving)
<_QZ> i need someone to kill a couple basketball players and some refs
-:- SignOff _QZ: #TUNES (BRiX [http://www.qzx.com/brix] :: sleep)
11:10pm
-:- Flash [speedy2@36001.radcaixa.tsai.es] has joined #Tunes
-:- ruiner [nate@ppp373.wi.centuryinter.net] has joined #tunes
<Flash> alguien habla espaņol?
<ruiner> ack!
<ruiner> get out of here you foreigner!
<ruiner> just kidding
<Flash> por que?
<Flash> mas vale que te vayas a acostar
<ruiner> No hablo espaņoles
<Flash> i donīt speak english with ignorants
<ruiner> heh heh
<ruiner> sorry
<Flash> ok
<Flash> Iīm foreigner...
<Flash> Iīm from Spain
<Flash> ...
<ruiner> ...
<Flash> byeeeeeeeee
-:- Flash [speedy2@36001.radcaixa.tsai.es] has left #Tunes []
11:20pm
-:- billyboof [hatefull@nrwc-sh1-port80.snet.net] has joined #tunes
<billyboof> hello...
<ruiner> greetings
<billyboof> hows it going?
<ruiner> alright
<ruiner> how're you?
11:30pm
<billyboof> i'm doing good... just really bored...
<ruiner> same here
<billyboof> are you running mirc?
<ruiner> you play core war?
<ruiner> yeah
<billyboof> check out my /whois
<ruiner> hatefull.....hmmmm.......hateful is spelled with one l
<ruiner> T1? must be nice
<billyboof> i don't have t1
<billyboof> i ment @#Tunes
<billyboof> you see that right?
<billyboof> yeah the email addr was taken though ... soo... :(
<ruiner> what was I supposed to see in your whois?
<billyboof> what'd you get for the channels i'm in?
<ruiner> tunes
<billyboof> @#Tunes ?
<ruiner> yeah
<billyboof> ok ... do it now
<billyboof> look at the channel list..
<ruiner> this is all
<billyboof> <yes i'm very very board...
11:40pm
<ruiner> brb, gotta pee
<billyboof> bored too...
<billyboof> ok...
<ruiner> heh heh
<billyboof> that's a pretty nice little bug of mirc...
<billyboof> it shows the color codes in /whois lists...
<billyboof> so i just /join #Tunes^K0blah and it shows @#Tunes ... then the rest is white <which is hidden to most people>
<ruiner> what projects are you working on?
11:50pm
<billyboof> i'm writing an os
<ruiner> ah
<ruiner> features?
<billyboof> currently or planned? :)
<ruiner> both
<billyboof> currently it features a bootsector which loads into pmode <not grub, but much like it, i wrote myself> ..
<billyboof> the pmode code which is loaded sets up a new perm GDT/IDT/page tables/enables a20/paging and runs 4 tasks
<billyboof> currently have timer/keyboard irq's written....
<ruiner> hmmmmmm......
<billyboof> neither are close to final... but ... the timer just loops through the tasks... the keyboard only does normal display keys and shifts...
<billyboof> umm... plans.... i don't know where to start...
<billyboof> document based...
<billyboof> total seperation of programs from visible filesystem...
<billyboof> programs will be plugins that'll be visible as componets in a kinda thought bubble chart... not as dir's and files....
<billyboof> the desktop dir will be the /
<ruiner> I was thinking to something along the same lines
<ruiner> but I figured it would be too limiting
<billyboof> limiting how?
<ruiner> apps as plugins for the os? well.....its pretty obvious
<ruiner> plus I'm assuming there would be lots of calls back and forth
<billyboof> that's not set in my mind....
<ruiner> I say, just load the instructions into memory and let fly
<billyboof> definately more integrated than applications as we know it...
<billyboof> know what i mean?
<ruiner> true
<ruiner> but anything that dependant on the os isn't going to be easy to port
<ruiner> so you're not going to get lots of support from people who aren't die hard users of your os
[msg(TUNES)] newlog 1999.0513
IRC log ended Thu May 13 00:00:00 1999