IRC log started Sat Apr 17 00:00:01 1999
[msg(TUNES)] permlog 1999.0417
<Fare> gakuk again
03:40am
* Fare/#Tunes fell asleep yesterday
-:- rau [mbalazs@server-gw-1.phos.server.ro] has joined #Tunes
-:- mode/#tunes [+v rau] by Tril
-:- SignOff rau: #TUNES (BitchX No this is not a cybersex client)
-:- smkl [sami@DXI.rdyn.saunalahti.fi] has joined #tunes
-:- mode/#tunes [+v smkl] by Tril
-:- AlonzoTG [Alonzo@client-151-200-126-129.bellatlantic.net] has joined #tunes
<smkl> hello AlonzoTG
<AlonzoTG> om
11:20am
-:- SignOff AlonzoTG: #TUNES (Have Nice Day :))
<Fare> gakuk
11:40am
-:- Iepos [root@d3.t1-3.tecinfo.com] has joined #TUNES
-:- mode/#tunes [+v Iepos] by Tril
<Fare> gakuk, Iepos!
<Iepos> gakuk!
<Iepos> I read some of your lambdaND again
<Fare> sorry about yesterday: I literally fell asleep
<Iepos> heh heh
<Iepos> i forget... you live in France, right?
<Iepos> that was late at night then, i guess.
<Iepos> Anyway, I'm having some trouble getting past some of the terminology...
<Fare> it's rather left than right
<Fare> Iepos: go ahead. The article was rejected, so there must be a lot to enhance...
<Iepos> do you know where i can find definitions of some terms?
<Iepos> anyway, what is a recursively enumerable set?
11:50am
<Iepos> ummm.... you're not asleep again, are you, Fare?
12:00pm
-:- SignOff Iepos: #TUNES (Iepos)
-:- Iepos [root@d3.t1-3.tecinfo.com] has joined #TUNES
-:- mode/#tunes [+v Iepos] by Tril
-:- SignOff Iepos: #TUNES (Iepos)
-:- GMOL [gmol@24.66.11.51] has joined #tunes
-:- mode/#tunes [+v GMOL] by Tril
<GMOL> DOes anyone know if the 4MB page optimization on linux works?
-:- SignOff GMOL: #TUNES (Leaving)
12:30pm
-:- Iepos [root@d15.t1-3.tecinfo.com] has joined #TUNES
-:- mode/#tunes [+v Iepos] by Tril
<Fare> hsfthrth
<Iepos> hsfthrth ?
<Fare> Iepos: take any book on the basics of computation theory
<Iepos> yeah.
<Iepos> i'm trying to find a place to get Combinatory Logic still...
<Fare> Raymond Smullyan has written some.
<Iepos> okay...
<Iepos> what is a recursively enumerable set anyway?
<Fare> There is ~rbjones' page on combinatory logic
<Fare> (and logic in general`
<Iepos> yeah, I've seen part of it
<Fare> you'll find lots of pointers and bibliography at his
<Fare> recursively enumerable set == a recursive program can enumerate the set.
<Iepos> enumerate? does that mean list all the elements?
<Fare> yup
<Iepos> ok then
<Fare> though not necessarily in any particular order
<Iepos> next, what's a computational domain?
<Fare> that's not as well defined
<Iepos> oh
<Fare> in my article, it's a set of objects whose semantics is somehow equivalent to that of computable functions
12:40pm
<Iepos> I'm going to finish reading it now ...
<Fare> that is, you have injective recursively enumerable mappings both way between this domain and the domain of recursively enumerable functions on integers
<Iepos> let's see... injective means many-to-one ? or other way around, i forget?
<Fare> Iepos: do not hesitate to suggest enhancements...
<Fare> injective means one-to-one
<Iepos> oh...
<Fare> but not all elements may be reached
<Fare> I think I define injective somewhere in the text...
<Fare> (although it's a fairly well-known concept)
<Iepos> after reading some of this, relations (which I've never really thought much about until now) are starting to sound more fundamental than functions (deterministic functions i mean) ...
<Fare> Iepos: depends on what formalism you use.
<Iepos> although I'm still not sure I get the idea of a non-deterministic function
<Fare> In type theories and categorical logic, deterministic functions are sure more primitive.
<Iepos> yes...
<Iepos> It seems like if "f" is a non-deterministic function, then "f x" could have different meanings, so in some cases "f x != f x" ... this seems contradictory
<Iepos> but I guess this is like the english "you != you" it certain contexts...
<Fare> Depends if "=" compares _expressions_ or _values_
<Fare> we have "f x" = "f x", but f x != f x
<Iepos> I meant "!=" in the sense of meaning, not representation.
<Iepos> so, value, not expression i guess.
<Fare> just like "this word" = "this word" but this word is not this word.
<Iepos> this seems like a problem though in a language, if two identical forms can have different meanings.
<Fare> depends on what you call "meaning"
<Iepos> wouldn't it just be better to use relations, where this doesn't happen?
<Fare> again, if the meaning of an expression is the set of all values it may take, then this meaning doesn't change
12:50pm
<Fare> you still have meaning("f x") = meaning("f x")
<Iepos> so, now you have deterministic functions that just returns sets of values.
<Fare> what you don't (always) have is f x = f x
<Fare> but you still have let y = f x in y = y
<Iepos> non-deterministic functions seem like fuzzy concepts to me.
<Iepos> but I should finish reading your paper...
<Fare> I cite a review of non-deterministic stuff in algebraic settings
<Fare> you _can_ interpret non-deterministic functions thanks to deterministic functions yielding sets.
<Iepos> yes ...
<Iepos> i'm printing off the thing now ...
<Iepos> i don't have a dvi or a ps viewer
<Iepos> only have a pdf viewer and ps2pdf
<Iepos> and it is really really slow :-)
<Iepos> it takes almost as long to show it on the screen as print it.
<Fare> Iepos: do you have linux?
<Iepos> yes
<Iepos> slackware, remember ? :-)
<Iepos> doesn't come with anything.
<Iepos> actually, i also have some debian cds somewhere.
<Fare> well, maybe it's time to go debian...
<Iepos> heh heh
* Fare/#Tunes thinks slackware is trying to tell Iepos it's time to upgrade
01:00pm
<Iepos> hmm ... what is a partial function? one that does not have apply to all parameters (has a limited domain?)
01:10pm
<Fare> has a limited domain
<Iepos> ok
<Fare> i.e. x |-> 1/x is not defined for x=0
<Iepos> so division is a partial function...
<Iepos> i guess all arithmetic functions are, because they only work on numbers...
<Iepos> Let's see, what is a total function. One that is not partial?
<Iepos> (I'm having trouble with the syntax used to describe total, injective, surjective, etc.)
<Iepos> (it's probably standard syntax; i just haven't see much of it...)
<Fare> Iepos: a total function is defined on the whole considered domain
<Fare> x |-> 1/x is total on R^*
<Iepos> okay, and what is surjective ?
<Fare> (the real line minus point 0)
<Fare> being surjective is reaching every point in the considered codomain
<Iepos> uummm ... ?
<Iepos> so a function is surjective if ... ?
<Fare> x |-> 1/x is surjective onto R^*, (but obviously not onto R)
<Iepos> what does * mean here?
<Fare> x |-> x^2 is surjective onto R^+
<Fare> \mathbb{R}^* is the set of invertible real numbers
<Fare> for any ring R, R^* is the set of invertible numbers in the ring
01:20pm
<Iepos> oh
<Fare> (see basic book on algebra)
<Iepos> hmmm..............
* Fare/#Tunes has problem installing linux on a friends' DELL Inspiron 7000 :( :( :(
<Iepos> :(
<Fare> and there's a web page specifically to help on that case...
<Iepos> wait ... i think i'm getting surjective ... does it mean that the domain and range are the same?
<Fare> the range is the set of actually reached points
<Fare> being surjective means the actual range == the considered codomain
<Iepos> so, what is a codomain?
<Fare> it's the considered "target" set for your function/relation
<Fare> as when you write (in LaTeX):   $f : A \rightarrow B$
<Iepos> oh ... i don't write LaTeX
<Iepos> all right... what is an involutive operation ?
<Iepos> i gotta go
<Iepos> au revoir
-:- SignOff Iepos: #TUNES (Iepos)
01:30pm
-:- AlonzoTG [Alonzo@client-151-200-125-219.bellatlantic.net] has joined #tunes
-:- hcf [nef@me-portland-us636.javanet.com] has joined #tunes
-:- mode/#tunes [+v hcf] by Tril
-:- _QZ [brand@p0wer.qzx.com] has joined #tunes
-:- mode/#tunes [+v _QZ] by Tril
-:- fishhead [shave@client-151-201-27-95.brads.net] has joined #Tunes
-:- mode/#tunes [+v fishhead] by Tril
<fishhead> what's this channel about
<_QZ> tunes operating system
<fishhead> what's what
<fishhead> oh wait
<fishhead> HAHAHAHAHAH
<Fare> look at the first part of the /topic
<Fare> (the second is of interest, too, although different :)
<fishhead> back when I used to hang on this network I used to hear someone yap about a os project callled tunes
<fishhead> I guess it turned into something
* Fare/#Tunes blushes
<fishhead> fare, your female ?
<_QZ> hahahaha
<fishhead> DOH
* fishhead/#Tunes remeber he has not installed the sound drives yet
<fishhead> be back another time
-:- fishhead [shave@client-151-201-27-95.brads.net] has left #Tunes []
-:- SignOff hcf: #TUNES (Leaving)
05:00pm
-:- Iepos [root@d10.t1-4.tecinfo.com] has joined #TUNES
-:- mode/#tunes [+v Iepos] by Tril
-:- SignOff Iepos: #TUNES (Iepos)
-:- tcn [tcn@cci-209150250110.clarityconnect.net] has joined #tunes
-:- mode/#tunes [+v tcn] by Tril
<_QZ> hello
<abi> niihau, _QZ
<Fare> ole
<tcn> hey
<abi> hey, tcn
<tcn> just committed changes to retro.. Forth compiler works now.
<_QZ> kewl
<_QZ> got a binary?
<tcn> yeah
<tcn> http://bespin.cx/~tcn/retro/kernel/image.1999-0417
<_QZ> yup got it
<tcn> afk
<_QZ> ugh
<_QZ> u need to fix writeboot so it can write to my nfs floppy
<tcn> oh, you don't need it anymore
<tcn> just use cp
06:30pm
<_QZ> oh good
<tcn> Now it modifies the image at compile-time
-:- SignOff AlonzoTG: #TUNES (Have Nice Day :))
<_QZ> ah good, bye works :)
<tcn> haha
<_QZ> u havent updated the news section on the retro page
<tcn> as we speak
<Fare> I still haven't dichotomy-booted retro on the laptop
<_QZ> on may 1st i will be releasing binaries for the kernel and some system modules/objects along with the source code for the linux tools and the api files
<Fare> QZ: who's your target audience?
06:40pm
<_QZ> anyone that wants to program for it
06:50pm
-:- SignOff smkl: #TUNES (sleeping ...)
<tcn> Fare: any more thoughts on this Scheme interpreter?
<Fare> tcn: the one I oughto do for this week?
<Fare> hum. looks like it's late
<Fare> (to say the least)
<_QZ> no it isnt
<tcn> it's already next week in france :)
<tcn> Fare: if you feel like writing it yourself, go ahead.. let me know what support you need from the OS
<tcn> I'm gonna work on multitasking
<Fare> tcn: do you have space for a heap?
<tcn> Fare: yep. It can be allocated on demand, if you like.
<Fare> to simplify things, I need a double address space for object
<Fare> for an object heap
<tcn> double?
<_QZ> i am surprised at how easy it is to impliment tracking of required code and why no OS has ever done it before :)
<tcn> qz: F-PC does it, if you can consider than an OS
<_QZ> fpc?
07:00pm
<tcn> it's an older 16-bit Forth environment.. real nice, though.
<_QZ> no, im talking about major OSes
<_QZ> windos, os/2, unix, etc...
<tcn> yeah
<tcn> Fare: what exactly do you mean by "double address space"?
<_QZ> when brix compiles a method it records all methods needed by that method and increments there counts. and it decrements the counts when that method is removed assuming its count was zero to be removed
<Fare> tcn: the stupidest GC algorithm takes twice the address space for objects
<tcn> oh
<_QZ> why twice?
<Fare> tcn: or do you have virtual memory management already?
<_QZ> my GC doesnt need double memory
<Fare> QZ: to live copy objects from old to new location
<tcn> Fare: You can use VMM
<Fare> tcn: what's VMM's interface?
<_QZ> live copy?
<Fare> copy live objects
<_QZ> why do u need to copy them?
<Fare> QZ: see hbaker's page on GC
<tcn> Fare: It's not much now.. what do you need to do?
<Fare> QZ: to 1) separate them logically from dead objects 2) compact them
<Fare> QZ: again, that's the stupidest idea
<Fare> QZ: in early LISP systems, the "disk" was the second memory.
<_QZ> hmm
<Fare> QZ: or tape, or whatever. Then, objects were dumped back into memory from tape
<Fare> QZ: ever read a book on the HP28/48 implementation?
<tcn> wow
<_QZ> i know how the hp48 works
<Fare> then you know about GC
<_QZ> 64bit processor with a 4bit data bus :)
<Fare> the hp48 uses a mark&sweep GC.
<Fare> does things in one space, but requires two passes on data
<tcn> Fare: What do you need the VMM to do?
<_QZ> i have no idea how GC works, i just know that nothing in brix can exist unless something else is using it :)
<Fare> tcn: I need separate features: 1) allocate (unmapped) real mem, 2) allocate virtual mem 3) mmap virtual mem to real mem (or to nothing)
07:10pm
<Fare> Actually, I'd rather have full control on allocation, and just have a list of available resources
<Fare> as for 3, well, at first it'd be just a rationalization of the x386 page table mechanism
<_QZ> the only malloc in the brix kernel is only usable by the ofs module to allocate space for objects
<_QZ> if u need memory u can either clone an existing object or create an untyped object
<_QZ> that only exists in memory
<tcn> Fare: allright, just design scheme as you see fit, and I'll adapt the VMM as needed.
<Fare> tcn: ok
<Fare> tcn: hum. How am I meant to find what are available memory zones?
<Fare> tcn: to simplify, let's make a first version where I assume a contiguous real mem space w/o vmm
<Fare> tcn: plus this will make it more portable.
<tcn> allright
<tcn> We can give a big chunk (1M sound good?) to scheme.
07:20pm
<_QZ> hmm
<Fare> tcn: let's give it half the memory
<Fare> or something like that
<tcn> 1 meg ought to be enough for anyone
<tcn> hey, why not just give it 640k? :)
<Fare> hum, what will you do with the rest of the memory, anyway?
<tcn> right now Forth gets a chunk, too.. but that'll be dynamically allocated soon
<_QZ> why does forth need a chunk? what if i dont want forth on it
<Fare> QZ: FOO!
<tcn> qz: that'll be dynamically allocated soon
<Fare> QZ: FORTH *is* is OS, so it can have everything
<Fare> since, you allocate things from FORTH, anyway!
<Fare> tcn: what are your current stack policies?
<Fare> your current USER and DICTIONARY policies?
<tcn> Fare: EBP is the Forth stack pointer, and it's push-down
<tcn> Fare: If you don't use Forth, EBP is available
<tcn> Fare: I'm using the object system to handle the dictionary. You use 'insert' and 'search' to access it.
<tcn> (see odb.inc)
<Fare> how's the object system used?
<Fare> what are root objects?
07:30pm
<tcn> root:kernel contains all the external variables & routines in the kernel modules (kernel/*.rdf).  root:forth contains all the forth words (it's the dictionary).
<Fare> i.e. what are global variables
<Fare> and how do they interfere with threading?
<tcn> there is no threading yet :)
<Fare> yeah, but what are conventions for thread-dependent things as opposed to global things?
<_QZ> ya, retro is dos + forth :)
<tcn> Fare: I lost ya :)
<_QZ> when u start to do threading u will knowwhat he means :)
<tcn> let me put it this way: If you type a word in Forth, it searches under root:forth.  If you declare 'extern print' in a module, the linker searches under root:kernel
<_QZ> why?
<tcn> because
<tcn> that's how it evolved
07:40pm
<Fare> tcn: you should read about LISP implementations -- that would open your mind
<Fare> tcn: get LiSP by C. Queinnec, and read the articles by hbaker
<tcn> Fare: can you recommend something about a *simple* LISP implementation, maybe a really old one?
<Fare> the LISP concept corresponding to threading context is "dynamic context"
<Fare> LiSP, the book, by C. Queinnec will really explain everything in depth, but GC.
<Fare> as for GC, there is a book specifically on it by someone & Lins
<Fare> and/or you can read the implementation notes of RScheme.org
<Fare> and you should read hbaker's article.
<Fare> (and if you begin to like it, the cmucl internals doc)
<Fare> so at least, read the impl. notes of RScheme.org
<Fare> but LiSP, the book, is great to have read, too
<tcn> what's hbaker's homepage?
<Fare> see Languages.html
<Fare> look for baker there
<tcn> is it supposed to be FTP only?
<Fare> hbaker's stuff is ftp, but you can view it with a browser
<tcn> allright, *which* article?
<Fare> LiSP is a book, although the accompanying code is available by ftp
<Fare> tcn: pretty everything he wrote is interesting.
<Fare> tcn: see the subjects that you prefer
<tcn> hmm
<tcn> parsing lisp
<Fare> but his is not tutorial material
<Fare> it supposes you already know what it's about (although not necessarily that you're very good at it)
<_QZ> damn i have about $50 in change sitting on my desk. time to clean it up :)
<Fare> so maybe start with the docs of RScheme.
<Fare> QZ: buy a book :)
<_QZ> i just bought a book
<_QZ> besides i am not the kinda person who counts and rolls change. i will just throw it in the garbage
<_QZ> after i pick out the quarters :)
07:50pm
<tcn> they should get rid of nickles & dimes & pennies
<_QZ> they should
<tcn> then companies would *have* to stop pricing things at $1.99 and just use $2
<_QZ> tax
<tcn> they should make it even after tax
<_QZ> they would have to charge $3
<tcn> I wonder why nobody does that
<_QZ> for a $2 item :)
<tcn> .
>>> tcn [tcn@cci-209150250110.clarityconnect.net] requested PING 924403677 426039 from #tunes
08:00pm
<Fare> gotaa go sleep
<Fare> bye!
<_QZ> cya
<tcn> later
<tcn> I think I'll leave too
<tcn> see ya
-:- SignOff tcn: #TUNES (tcn has no reason)
08:10pm
[msg(TUNES)] newlog 1999.0418
IRC log ended Sun Apr 18 00:00:00 1999