IRC log started Sun Apr 18 00:00:00 1999
[msg(TUNES)] permlog 1999.0418
-:- NetSplit: king.openprojects.net split from sterling.openprojects.net [12:04am]
-:- BitchX+Deb1an: Press Ctrl-F to see who left Ctrl-E to change to [king.openprojects.net]
-:- Netjoined: king.openprojects.net sterling.openprojects.net
-:- Tril [dem@sloth.wcug.wwu.edu] has joined #TUNES
-:- mode/#tunes [+o Tril] by ChanServ
-:- NetSplit: mccaffrey.openprojects.net split from koontz.openprojects.net [01:43am]
-:- BitchX+Deb1an: Press Ctrl-F to see who left Ctrl-E to change to [mccaffrey.openprojects.net]
-:- SignOff _QZ: #TUNES (BRiX [http://www.qzx.com/brix] :: sleep)
-:- Downix [down@survivoronetwothree.ne.mediaone.net] has joined #Tunes
-:- mode/#tunes [+v Downix] by Tril
-:- SignOff Downix: #TUNES (Leaving)
-:- tcn [tcn@cci-209150250093.clarityconnect.net] has joined #tunes
-:- mode/#tunes [+v tcn] by Tril
<tcn> hello
07:00am
-:- tcn has changed the topic on channel #tunes to: Reflective Computing System <http://www.tunes.org/>, Retro updated <http://bespin.cx/~tcn/retro.html>, <http://www.users.zetnet.co.uk/petef/scribbles/ppl.html>
-:- SignOff tcn: #TUNES (tcn has no reason)
-:- smkl [sami@MCCCII.rdyn.saunalahti.fi] has joined #tunes
-:- mode/#tunes [+v smkl] by Tril
-:- ProGuy [Paul@p128-60.ppp.get2net.dk] has joined #tunes
-:- mode/#tunes [+v ProGuy] by Tril
<ProGuy> Hello
-:- SignOff ProGuy: #TUNES (Segmentation fault : KVirc 0.9.0 by Szymon 'Pragma@ircnet' Stefanek <kvirc@tin.it>)
08:40am
-:- AlonzoTG [Alonzo@client-151-200-120-108.bellatlantic.net] has joined #tunes
!mccaffrey.openprojects.net!! SERVER Numeric Collision: services1.openprojects.net != services10.openprojects.net
-:- _QZ [brand@p0wer.qzx.com] has joined #tunes
-:- mode/#tunes [+v _QZ] by Tril
-:- SignOff smkl: #TUNES (urgh)
-:- binEng [bineng@dialup243-2-47.swipnet.se] has joined #tunes
-:- mode/#tunes [+v binEng] by Tril
* AlonzoTG/#tunes greets "Salutations binEng!" :-)
<binEng> hi
11:10am
<AlonzoTG> this linking problem is really pissing me off...
<AlonzoTG> I read the libraries to MINIX and it looks like it interfaces with the system at link time...
<AlonzoTG> But I can't figure out how to do that...
<AlonzoTG> I guss it is just a lack of confidence because I havn't maniplated binaries before as data...
<AlonzoTG> :|
<binEng> Don't ask me... I'm just reading GEB
<AlonzoTG> geb=?
<AlonzoTG> godel escher bach?
<binEng> yep
<AlonzoTG> what do they have to say for themselves?
<binEng> whadda mean?
<AlonzoTG> what is their hypothesis?
<AlonzoTG> talking about incompleteness?
<binEng> I haven't finished the book yet...
* AlonzoTG/#tunes ph33rs y0ur 1337n3$$
<binEng> partly about incompleteness, yes. And "strange loops"... AI...
<AlonzoTG> :P
<AlonzoTG> AI: what kind of theories do they present?
<binEng> uhh read the book ;) well, it sounds pretty much like refelection
<binEng> reflection
<AlonzoTG> kule...
<AlonzoTG> that is what I was saying in my most recient post to the TUNES list...
<binEng> yes. coincidences...
<binEng> That book has gotten me interested in AI. It's very interesting, but I should really be doing school job instead :P
11:20am
<AlonzoTG> hehe
<AlonzoTG> nah.
<AlonzoTG> keep doin what u are doing... :)
<AlonzoTG> lemme put that book on my list...
<binEng> do that
-:- PP [micorreo@193.152.225.91] has joined #Tunes
-:- mode/#tunes [+v PP] by Tril
<PP> thanks tril!
<PP> Bueno alguien habla espaņol?
<AlonzoTG> he's afk
<_QZ> si
<AlonzoTG> he has his client set to autovoice anybody who doesn't say "om"
<PP> Menos mal!, que tal _qz?
<AlonzoTG> (om)
-:- SignOff PP: #TUNES (Leaving)
<_QZ> hmm, he left before i could talk to him
-:- SignOff binEng: #TUNES (Ping timeout for binEng[dialup243-2-47.swipnet.se])
-:- binEng [bineng@dialup41-3-6.swipnet.se] has joined #tunes
-:- mode/#tunes [+v binEng] by Tril
11:30am
-:- binEng2 [bineng@dialup145-1-39.swipnet.se] has joined #tunes
-:- mode/#tunes [+v binEng2] by Tril
<binEng2> I'm beginning to hate my ISP.
-:- SignOff binEng: #TUNES (Ping timeout for binEng[dialup41-3-6.swipnet.se])
-:- binEng2 is now known as binEng
11:50am
-:- tcn [tcn@cci-209150250115.clarityconnect.net] has joined #tunes
-:- mode/#tunes [+v tcn] by Tril
<tcn> hey
<tcn> AlonzoTG: Why are you looking at MINIX?
<tcn> ;)
-:- SignOff tcn: #TUNES (phone)
<AlonzoTG> kuz its there
12:00pm
<AlonzoTG> I haf the source directly accessable instead of having to hunt around through these arcane archives to get it...
>>> binEng [bineng@dialup145-1-39.swipnet.se] requested PING 924462280 from #tunes
12:10pm
-:- timestamp [dancer2@207.16.36.12] has joined #tunes
-:- ServerMode/#tunes [+v timestamp] by drexelhill.pa.us.opirc.nu
-:- SignOff binEng: #TUNES (Ping timeout for binEng[dialup145-1-39.swipnet.se])
<_QZ> any1 know of any servers on OtherNet?
01:10pm
-:- SignOff AlonzoTG: #TUNES (Have Nice Day :))
-:- tcn [tcn@cci-209150250116.clarityconnect.net] has joined #tunes
-:- mode/#tunes [+v tcn] by Tril
<tcn> hey, anyone here?
-:- SignOff tcn: #TUNES (tcn has no reason)
02:30pm
-:- tcn [tcn@cci-209150250120.clarityconnect.net] has joined #tunes
-:- mode/#tunes [+v tcn] by Tril
<_QZ> hello
<abi> salut, _QZ
<tcn> hey
<tcn> not much happened today, eh?
<_QZ> nope
<tcn> I spent half the day wiring up FF's.. that shit is a pain..
<_QZ> why?
<tcn> for a course
08:10pm
<tcn> usually it's microcontrollers, this week it was counters & shift regs.
<_QZ> i love that shit
<_QZ> i built a small cpu with that kinda stuff
<tcn> it beats english, chemistry, physics.. :)
<tcn> hehe.. how many breadboards did it take?
<_QZ> just a couple
<tcn> 4 bits?
<_QZ> 8
<tcn> I might try that with PGA's
<tcn> gonna do that in a couple weeks
<tcn> what opcodes did it have?
<_QZ> very simple ones
<tcn> enough to do anything?
<_QZ> nothing productive :)
<tcn> I think of it as practice for building real CPU's
<_QZ> i designed a real cpu
<tcn> you need an accumulator.. load & store instructions.. an ALU that does a few different ops at the same time, instructions to transfer the result of one ALU op back to the accumulator..
<tcn> and branching
<_QZ> when brix is usable i will have an asic made
<tcn> I'm thinking DPGA's now..
<tcn> Or maybe like the Transit project recommended, a combo DPGA/misc cpu chip
<_QZ> do a very simple cpu with enuf instructions, very little silicon
<_QZ> then put tons of em on the same die
<_QZ> and crank up the clock
<tcn> yeah
08:20pm
<_QZ> u have a cpu that can handle tons of threads
<tcn> Use only combinational logic
<tcn> should it have microcode?
<_QZ> a simple paging unit per thread cpu and some cache
<_QZ> no
<_QZ> the ops should be simple and not need microcode
<tcn> maybe explicit cache control
<tcn> so you're not always waiting for external RAM
<_QZ> well
<_QZ> some simple cache is ok
<tcn> How fast is ram now?
<_QZ> the ram in this machine is 103mhz and 6ns
<_QZ> my L2 cache is 463mhz and 5ns
<_QZ> or 4ns
<_QZ> cant remember
<tcn> You familiar with transputers?
<_QZ> ya
<tcn> Hey, on a P-400, is there an internal clock running at 800 mhz or so?
<tcn> 4 ghz?
<_QZ> uhh
<tcn> it's gotta have microcode..
<tcn> say it takes 5 uclocks to execute an instruction that takes 1 clock
08:30pm
<tcn> what's the 'ns' refer to in memory speed?
<_QZ> access speed
<_QZ> NanoSeconds
<tcn> and the 'mhz'?
<_QZ> if u invert the access speed u get the maximum transfer rate
<_QZ> so 6ns ram can handle 166mhz
<tcn> ok
<_QZ> dimms are 64bits wide so the bandwidth is 64bits * 166mhz
<tcn> what's a dimm? Is that what SDRAM comes in?
<_QZ> 1.3gig/sec
<_QZ> ya
<_QZ> simms are 72pins, dimms are 168pins
<tcn> and the old EDO DRAM, that's 72pins/32bits..
<_QZ> u can get edo and sdram dimms
<tcn> damn, that's a good reason to get a new MB
<tcn> My RAM is 50 or 60 ns
<_QZ> the pentium and up uses a 64bit data bus so u need pairs of simms
<_QZ> but with dimms u only need one
<tcn> ahh
<_QZ> they didnt make 50ns
<_QZ> 70ns fp dram or 60ns edo dram
<tcn> what's fp?
<_QZ> sdram ranges from 6ns to 10ns
<_QZ> fast pages
<_QZ> fast paged
<tcn> oh yeah, that's what my old ram was
<tcn> edo.. what's that stand for?
08:40pm
<_QZ> enhanced data output
<tcn> sdram.. static-dynamic ram?
<_QZ> heh not
<tcn> :)
<_QZ> synchronous dram
<tcn> does it always run synchronously?
<_QZ> no, it usually runs asynchronously
<_QZ> :)
<tcn> so it's not _really_ running at 6 ns
<_QZ> 6ns is the ACCESS speed
<_QZ> when the cpu says "give me the data at this address" the 6ns is how long it takes the ram to respond
<_QZ> it always takes only 6ns
<tcn> sounds nice
<tcn> you know where I can find the pinout for some *old* TTL ram?
<tcn> i got a bunch lying around
<_QZ> u wanna make dram cells?
<_QZ> or sram cells
<tcn> dram
<_QZ> oops i didnt read that entire line
<_QZ> thought u asked for schematics
<tcn> haha
<tcn> not masks.. I just have these chips
<_QZ> i coulda drawn up a d or s ram cell from my head
<_QZ> what chips?
<tcn> 4256-15 and 4264-15
<_QZ> company?
<tcn> TI
<_QZ> this is yer lucky day
<_QZ> i have a ti ttl book right here :)
<tcn> cool. I got a textbook that tells how to hook up drams
08:50pm
<_QZ> bummer this book doesnt have the 42 series
<tcn> got any other companies' databooks?
<_QZ> do u know what kinda ram it is
<tcn> hey, I found it on the net
<tcn> ahh crap, they don't have the pinout for that one
<_QZ> what kinda ram is it
09:00pm
<_QZ> s or d, size, databus size, addrbus size
<tcn> hang on
<tcn> 256k pagemode 150ns
<tcn> dram
<_QZ> ahh
<_QZ> thats why none of my books have it
<_QZ> they are all for embeded crap with 32k or less
<_QZ> libc6-static means i can use it on a libc5 system doesnt it :)
<tcn> ?
<tcn> i dunno
<tcn> anyways it's from an 80's vintage 8086 box.. it uses banks to address 640k
<_QZ> banks?
<tcn> yeah, I think that's what you call them
<tcn> there's some external logic chips to interface the RAM to the 20-bit address bus
09:10pm
<tcn> 16 pins.. would would they be? Vcc, gnd, in, out, r/w, ras, cas, A0-A8...?
<tcn> hehe
<tcn> The 3b1 faq has a cross-reference of compatible drams.. the ones I've got are from a AT&T 6300
<_QZ> do u have netrek?
<_QZ> this libc crap blows
<_QZ> in brix u could have libc5 and libc6 both working on hte same system
09:20pm
<tcn> got it
<tcn> the pinout, i mean
<_QZ> what is it
<tcn> don't have netrek
<tcn> http://www.blee.net/chipdir/pin/41256.txt
<_QZ> 41256?
<tcn> I would never have guessed it
<_QZ> u said 4256
<tcn> same thing
<_QZ> i have info on the 41 series
<tcn> in that 3b1 faq it list about 10 chips
<tcn> 41256 was made by NEC/okidata/samsung.. but it's the same
<tcn> found the 64k*1bit pinout also
<tcn> 8kb.. haha
<tcn> so you just set RAS, send the first 9 address bits, set CAS, send the next 9 bits
<_QZ> ya
<tcn> holy shit, it's 12:30
09:30pm
<_QZ> just take 64 of those chips and hook em to a pentium
<tcn> haha
<_QZ> and youll have 2meg of ram
<tcn> 150ns.. that means they can only run at 6.66 mhz?
<_QZ> ya
<tcn> interfacing it should be interesting :)
<tcn> well
<tcn> see ya
<_QZ> cya
<tcn> hey will this shit really fry if I touch it? :)
09:40pm
-:- SignOff tcn: #TUNES (tcn has no reason)
* Tril/#TUNES is back from the dead. Gone 195 hrs 1 min 43 secs
>>> Tril [dem@sloth.wcug.wwu.edu] requested PING 924498084 944444 from #TUNES
<Tril> anyone here
10:00pm
<_QZ> ya
<_QZ> i hate cheap makefiles
<Tril> go read http://tunes.org/~dem/tunes/Specifications.html, or at least the first part of it
<_QZ> ok
<_QZ> its about time :)
<Tril> yeah I finally got organized enough to work on tunes.
10:10pm
<_QZ> why is it hard to hide id numbers?
<Tril> maybe hiding them is not hard. But renumbering every object when the number of objects grows or shrinks might be.
<_QZ> growing is not a problem, shrinking is
<Tril> uhh?
<Tril> if you have a 32bit object number, and use them all up, you have to get a bigger number.
<_QZ> oh
<_QZ> well that is something u can handle when the time comes :)
<_QZ> do u isolate objects into spaces for users?
<Tril> hmm
<Tril> what do you mean spaces for userS?
<Tril> forget the caps there
10:30pm
<_QZ> well each user would have that entire range in the 32bit number
<_QZ> so then u would have another 32bit number for the user ID
<_QZ> now u have a 64bit object ID
<_QZ> and a user cant access another users objects
<Tril> actually, the object numbers i mean are ones for a single local system.  All the users share the local object store.  Restricting access to the objects is done by the typesystem, not by giving them separate stores. although I guess it could be.
<_QZ> or they can if u add support
<_QZ> and there can also be a store ID
<Tril> but the method you said, adding more bits to specify the user, that's the same thing that would happen in a distributed environment (anywhere with multiple stores). there would be some bits to identify which system is referred to.
<Tril> right
<_QZ> i use a 96bit number to reference objects
<Tril> but I keep references differently than you do
<Tril> i.e. the object referred to has the priviledge of moving around. It keeps a list of all references to itself, and can modify them when it moves.
<_QZ> i use a redirection table for that. when the object moves it gets added to the table and whenever another object tries to access the object the system will update that object. the object is removed from the table when all references are updated
<Tril> of course, that's an implementation detail. The implementation is FREE to search the world for all references to an object, if it wants. <puke, cough, wheeze>
<_QZ> brb
10:40pm
<_QZ> back
11:00pm
<_QZ> u here?
<Tril> yeah
<_QZ> under the types:object u say u delete an object by moving it to the non-object type
<_QZ> does that mean u allow direct deletion of objects?
<Tril> sure.
<_QZ> then u get broken links
<Tril> all the links to the object get deleted at the same time.
<Tril> you probably get trouble anyway, then, but you can install type checks on the delete operation to prevent deleting in any case you want
<_QZ> i only allow references to be deleted
<_QZ> when all references are gone then the object just auto-deletes
<Tril> yah.
<Tril> in tunes, references to an object are all that the object is made of.
<Tril> so deleting all references to it, well, that's already the SAME as deleting it.
<Tril> (but that's the part I didnt' write about yet)
<_QZ> and keywords also have references to objects
<_QZ> so u must remove all keywords from the object and any references that other objects had to the object before it vaporizes
<Tril> you have to remove all references to the keyword? :)
11:20pm
<_QZ> no
<_QZ> a keyword is not an object
<_QZ> if u remove a keyword then it will remove itself from all objects that used it
<Tril> what ever happened to "everything's an object"?
<Tril> brix doesn't have that
<_QZ> keywords are atributes :)
<Tril> attributes gotta be objects too :)
<Tril> they're attribute objects
<_QZ> they are special objects
<Tril> no such thing as a special object, they're all equal
<_QZ> i have tiny objects and big objects
<_QZ> and keywords are tiny objects
<Tril> what's tiny and big?
<_QZ> a number would be a tiny object
<_QZ> an embedded big object is also a tiny object
<_QZ> a big object is like a file
<_QZ> and it contains tiny objects
<_QZ> :)
<_QZ> tiny objects are under whatever laws set by the object they are in
<Tril> all objects are equal, whatever size, and they are all under the laws set by the typesystem.
<_QZ> is an integer an object in tunes?
<_QZ> do u maintain a uid, reference count, list of all references, etc for it?
<Tril> an integer is going to be an object, without the reference overhead stuff. But that doesn't make it different from another object. Because any object, you can turn on or off reference overhead..
<_QZ> not unless its a damn special integer
<_QZ> exactly
11:30pm
<_QZ> i can put an integer in its own big object
<_QZ> or it can be inside another object without the reference tracking
<Tril> I don't go around making arbitrary distinctions between tiny and big objects, though, and make stuff depend on that. Whether something has reference counting or back references is irrelevant to most of the system.
<Tril> it's irrelevant to what the object actually DOES
<_QZ> in either case the methods that access the object dont care if its tiny or big
<Tril> you don't have to special-case the methods?
<_QZ> no
<_QZ> well im off to bed
<_QZ> cya
-:- SignOff _QZ: #TUNES (BRiX [http://www.qzx.com/brix] :: sleep)
11:40pm
* Tril/#TUNES  is away: (afk) [BX-MsgLog Off]
-:- SignOff Tril: #TUNES (Ping timeout for Tril[sloth.wcug.wwu.edu])
[msg(TUNES)] newlog 1999.0419
IRC log ended Mon Apr 19 00:00:00 1999