00:00:00 --- log: started forth/03.07.28
00:00:21 <kc5tja> OK that changes some things.  I don't know how to help in that case.
00:14:30 <kc5tja> Well, I'm thinking I should get to bed.
00:14:41 --- quit: kc5tja ("THX QSO ES 73 DE KC5TJA/6 CL ES QRT AR SK")
00:43:04 --- quit: Serg_Penguin ()
02:09:55 --- join: Serg_Penguin (Serg_Pengu@212.34.52.140) joined #forth
03:11:32 --- quit: Serg_Penguin ()
03:36:37 --- join: mur (murr@baana-62-165-186-178.phnet.fi) joined #forth
04:27:35 --- join: Serg_Penguin (Serg_Pengu@212.34.52.140) joined #forth
04:27:45 <Serg_Penguin> hi
04:28:20 <Serg_Penguin> mur: i played PIRATES yesterday, but i found no way around ;(
04:29:37 <Serg_Penguin> i could not fence, sail, battle etc.. - i did it but blindly, purposeless, trying to determine controls on the fly ^((
04:31:19 <mur> :P
04:41:54 <Serg_Penguin> would you consult me after work ?
04:42:26 <mur> me?
04:42:35 <Serg_Penguin> today i gotta run strict in time to my girl (or, better say, lady :)
04:42:48 <Serg_Penguin> oops.. u said you played ?
04:42:50 <mur> ;)
04:43:15 <mur> sorry, i dont remember anything of pirates, it has been 5 years when i played it some 
04:43:30 <Serg_Penguin> ;((
04:43:41 * Serg_Penguin remembers old games pretty good
04:44:04 <Serg_Penguin> Prince of Percia, F-19 ...
04:45:24 <mur> and i even played it on amiga with joystick
04:46:28 <Serg_Penguin> ;)
04:48:05 <Serg_Penguin> would i be near, you WOULD recall ;)
04:48:39 * Serg_Penguin is certified Dianetics auditor ;))
04:48:50 <mur> ;)
04:50:16 <Serg_Penguin> whadda u know of it ? ;)
04:50:29 <mur> of pirates?
04:50:44 <mur> only that you can sail rob boats and do business, which is not what you supposed to do :)
04:52:13 <Serg_Penguin> of Dianetics ;) 
04:52:24 <Serg_Penguin> i got what you forgot Pirates ;)
05:04:50 <Serg_Penguin> mur: what's your current/las Forth project ?
05:05:21 <Serg_Penguin> mine is rewriting my ASCII editor in kc5tja's fashion
05:08:39 * mur has not programmed for ages
05:08:44 <mur> cept one irc client
05:08:53 <mur> i have been doing grpahics only
05:37:58 --- join: w1k1_ (~w1k1@pD9E59528.dip.t-dialin.net) joined #forth
05:38:11 <Serg_Penguin> one my gfx was a scull and 2 crossed bottles
05:38:41 <mur> hah
05:45:21 --- quit: rO| (Connection timed out)
05:46:31 --- quit: w1k1 (Read error: 60 (Operation timed out))
05:46:33 --- nick: w1k1_ -> w1k1
05:48:25 <Serg_Penguin> mur: ever thought of own gfx editor/renderer ?
05:48:33 --- join: rO| (~rO|@pD9E59528.dip.t-dialin.net) joined #forth
05:48:34 <mur> hmm yes
05:48:39 <Serg_Penguin> in what do you draw ?
05:48:58 <mur> but that woudl require entirely new operating system context :)
05:49:03 <mur> photoshop usually
05:49:31 <Serg_Penguin> how do u make sprites match ?
05:49:43 <mur> what is sprites matching?
05:50:15 <Serg_Penguin> 1) making picture repeat seamlessly
05:50:26 <mur> oh pattern
05:50:44 <Serg_Penguin> 2) making many tiles join seamlessly in many number of combinations
05:50:46 <mur> when i design or when i want to paint with pattern?
05:51:22 <Serg_Penguin> then u draw tiles
05:51:37 <mur> i use eyes to do match :)
06:12:15 --- join: rO|_ (~rO|@pD9545372.dip.t-dialin.net) joined #forth
06:12:20 --- join: w1k1_ (~w1k1@pD9545372.dip.t-dialin.net) joined #forth
06:20:45 --- quit: rO| (Read error: 60 (Operation timed out))
06:20:47 --- quit: w1k1 (Read error: 60 (Operation timed out))
06:20:50 --- nick: w1k1_ -> w1k1
06:30:46 --- quit: Serg_Penguin ()
07:49:06 --- join: Herkamire (~jason@h0030657bb518.ne.client2.attbi.com) joined #forth
09:03:13 --- join: wossname (wossname@HSE-QuebecCity-ppp82003.qc.sympatico.ca) joined #forth
09:16:42 --- join: gilbertdeb (gilbert@fl-nken-u2-c3b-73.miamfl.adelphia.net) joined #forth
09:29:38 <rO|_> gilbertdeb: re
09:30:23 <gilbertdeb> hi rO|_
09:32:07 <rO|_> long time ago i saw you here, how are you doing?
09:32:40 <gilbertdeb> fine. mostly lurking.
09:33:26 --- quit: wossname (niven.freenode.net irc.freenode.net)
09:33:26 --- quit: rO|_ (niven.freenode.net irc.freenode.net)
09:33:26 --- quit: hefner (niven.freenode.net irc.freenode.net)
09:33:26 --- quit: GnuVince (niven.freenode.net irc.freenode.net)
09:34:15 --- join: wossname (wossname@HSE-QuebecCity-ppp82003.qc.sympatico.ca) joined #forth
09:34:15 --- join: rO|_ (~rO|@pD9545372.dip.t-dialin.net) joined #forth
09:34:15 --- join: hefner (hefner@pool-151-196-177-226.balt.east.verizon.net) joined #forth
09:34:15 --- join: GnuVince (~vince@modemcable158.188-200-24.mtl.mc.videotron.ca) joined #forth
09:35:38 --- join: gilbertbsd (gilbert@fl-nken-u2-c3b-73.miamfl.adelphia.net) joined #forth
09:35:49 <gilbertbsd> hmm.
09:36:19 <gilbertbsd> gilbertdeb must have died.
09:36:33 --- quit: gilbertdeb (niven.freenode.net irc.freenode.net)
09:36:33 --- quit: TreyB (niven.freenode.net irc.freenode.net)
09:37:43 <mur> hmm?
09:37:50 <gilbertbsd> terve mur
09:37:54 <mur> terve
09:37:55 --- nick: gilbertbsd -> gilbertdeb
09:38:05 * mur was confused for a moment
09:38:24 <gilbertdeb> a netsplit occured.
09:49:54 --- join: TreyB (~trey@cpe-66-87-192-27.tx.sprintbbd.net) joined #forth
10:02:24 --- quit: wossname ("ggs intruder quit")
10:46:17 --- join: kc5tja (~kc5tja@ip68-8-206-137.sd.sd.cox.net) joined #forth
10:46:17 --- mode: ChanServ set +o kc5tja
10:46:28 <gilbertdeb> re:
10:46:51 <kc5tja> 'morning.
10:46:55 <Robert> Hi :)
11:35:03 --- quit: skylan (Read error: 104 (Connection reset by peer))
11:37:53 * kc5tja is torn -- I want to do some actual FS/Forth programming for a change, but I also need to work up additional advertising material too...  So many decisions....
11:38:23 <gilbertdeb> kc5tja: is biz picking up?
11:38:55 <kc5tja> Not really, but if it's going to happen, it won't happen right away.
11:39:05 <kc5tja> I made my first dollar last week on Thursday.
11:39:27 <gilbertdeb> wow. what did you do?
11:41:55 <kc5tja> I connected a printer to the computer using a USB cable.
11:42:03 <kc5tja> (Yes, I am absolutely, positively serious about this)
11:42:13 <gilbertdeb> Hmmm. there is hope for me ...
11:42:25 <gilbertdeb> I just need to pick up some spanish soon ;)
11:44:52 <Herkamire> kc5tja: they payed you a dollor for that ;)
11:47:12 <Herkamire> I wish I got cash for my first income so I could tack it to the wall
11:47:28 <gilbertdeb> how do you pay for the tack? :)
11:47:51 <Herkamire> gilbertdeb: it's already bought and paid for ;)
11:48:05 <Herkamire> I invested in several :)
11:48:19 <Herkamire> that, or I stole several, don't remember :)
11:48:56 <gilbertdeb> it is kinda hard to tack ones' first dollar on the wall and just look at it.
11:49:07 <gilbertdeb> I'd rather take a photo, put it on the web and go spend it wisely :)
11:53:25 <kc5tja> No, they paid me $30.
11:54:08 <kc5tja> I hadn't the heart to charge my full hourly rate.
15:32:19 --- quit: gilbertdeb ("<--- off to play in the dirt.")
16:32:12 * kc5tja will stay home from aikido today.  Not feeling all that well.
16:33:43 --- join: suprdupr (CrowKilr@HSE-Windsor-ppp251430.sympatico.ca) joined #forth
16:33:47 <suprdupr> hi!
16:33:53 <kc5tja> greetings
16:34:12 <suprdupr> i just met a guy on the bus, i talked him about my forth project of course, he's in javascript dev, hes 17 and learning
16:34:42 <suprdupr> he already knew about the benefits of design phase (thinking about what you're gonna write) and wants to join a community
16:35:06 <kc5tja> What am I writing in this case?  :)
16:35:34 <suprdupr> i really start to believe in the idea thta if you believe in something and you dont change your mind, well the universe's gonna send it to you
16:35:46 <suprdupr> kc5tja: lowlevel drivers I hope ;o)
16:36:26 <suprdupr> i discovered the bourque engine and a new kind of electric motor
16:36:33 <suprdupr> just by checkin g my mail =)
16:37:43 <kc5tja> I believe one has to work hard to achieve what he wants.
16:37:49 <kc5tja> And even then, it's no guarantee.
16:39:38 <suprdupr> i dont believe its already done, just that it got a kickstart
16:42:49 <kc5tja> You know, I should actually get some design work done on my Forth, shouldn't I, seeing as how I'm taking tonight off from Aikido while my stomach heals a bit?
16:43:11 <suprdupr> can only agree with that =)
16:43:28 <suprdupr> btw its the bourke engine, for the interested
16:43:39 <suprdupr> http://bourke-engine.com/b/faq1.htm
16:51:44 <suprdupr> brb
16:51:56 --- quit: suprdupr ("i think i got something better for my ;")
16:54:35 --- join: suprdupr (CrowKilr@Ottawa-HSE-ppp3653577.sympatico.ca) joined #forth
16:55:15 <suprdupr> i thought that i didnt need the optimization pointers and to look back in code for actual "call rel32" opcodes
16:55:31 <suprdupr> since i use almost a call token i should only look for him
16:55:45 <suprdupr> code get simpler and nicer
16:56:35 <suprdupr> also about stacks in registers
16:57:07 <suprdupr> I was thinking about stacks in registers, let say you have a 2 unit stacks, it takes 1 mov to dup or drop
16:57:15 <suprdupr> a 3 unit stack takes 2 moves etc
16:58:18 <suprdupr> 3 unit would be ok, since on x86 movs can pair
16:58:26 <suprdupr> and each mov is 2 bytes
16:58:33 <suprdupr> so 2 movs fit in one cell =)
16:58:58 <suprdupr> my current "universal" dup and drop primitives are 5 bytes long
16:59:14 <suprdupr> if a forth would need only a two deep datastack, one could use registers
16:59:32 <suprdupr> but three starts to be ok
16:59:37 <suprdupr> 2 is very limitative
16:59:44 <kc5tja> The minimum useful stack depth is actually 3 elements deep.
16:59:59 <suprdupr> but waht do you think of the idea?
17:00:06 <kc5tja> The Steamer16 MISC CPU uses a 3-deep data stack.
17:00:14 * kc5tja doesn't know.
17:01:21 <suprdupr> a fetch would be 4 instructions, a store only two
17:01:27 <suprdupr> no
17:01:44 <suprdupr> 3 instructions
17:02:38 <suprdupr> anyway you get the picture
17:02:51 <suprdupr> riscs could have the return stack in regs
17:05:32 <kc5tja> Return stacks can get rather large.  Even an 8-deep return stack would take more clocks to maintain than keeping the return stack in RAM.
17:07:31 --- quit: mur ("MURR! save the http://rainforest.care2.com/")
17:09:07 <suprdupr> kk but for the datastack it could work
17:10:06 <suprdupr> since its around 3-4 items, but only three items would make it worthwhile, since after code size is 6 vs 5 for memory based stack, and you get less registers
17:16:42 <kc5tja> Well, like I said, 3 is the minimum stack depth to do anything useful.
18:56:47 --- join: rk (~rk@ca-cmrilo-docsis-cmtsj-b-36.vnnyca.adelphia.net) joined #forth
19:45:03 <kc5tja> back
19:45:21 <kc5tja> for now at least -- I actually didn't get around to doing much design work.
19:54:43 <kc5tja> http://news.bbc.co.uk/1/hi/world/americas/3060579.stm
19:58:42 <Herkamire> hehe :)
20:00:33 <kc5tja> See, that just makes me want to go out and purchase a rubber duck, just so I can have a continued memory of that article.
20:30:00 <suprdupr> i tried to get people interested
20:30:08 <suprdupr> in my project by the bruteforce approach
20:30:13 <suprdupr> didnt quite work very well
20:30:14 <suprdupr> ;pp
20:57:37 * rk is away: snooozes
21:11:41 --- join: skylan (sjh@vickesh01-4572.tbaytel.net) joined #forth
21:11:52 --- quit: Herkamire ("bedtime")
21:17:30 --- quit: rk (Read error: 110 (Connection timed out))
21:21:27 --- quit: suprdupr ("work tommorow, see ya everyone")
21:49:43 <kc5tja> back
22:04:37 --- join: Serg_Penguin (Serg_Pengu@212.34.52.140) joined #forth
22:04:44 <Serg_Penguin> hi !
22:05:32 <kc5tja> howdy
22:05:53 <Serg_Penguin> i finally wrote your idea 4 my gp-forth
22:06:06 <Serg_Penguin> but in 'normal' mode, no self-loader
22:06:17 * kc5tja nods
22:06:36 <Serg_Penguin> i realized one difficulty anyway:
22:06:49 <Serg_Penguin> i need words what are independent of MODE
22:07:14 <Serg_Penguin> like "exit program w/ Y/N ?"
22:07:23 <Serg_Penguin> or "gimme Forth console"
22:08:13 <Serg_Penguin> even mode switch words would be more convenient 'modeless'
22:12:50 <Serg_Penguin> whadda you think of this ?
22:13:03 <kc5tja> I don't know what you mean.
22:13:59 <Serg_Penguin> i mean what if i hit ESC, i'll get the question "Are you sure to exit [Y/N] ?" in _any_ mode
22:14:20 <kc5tja> : $$i1B query-quit ;
22:14:24 <kc5tja> : $$c1B query-quit ;
22:14:27 <kc5tja> (or something similar)
22:14:36 <Serg_Penguin> doubling words ?
22:15:11 <Serg_Penguin> hmm...
22:16:04 <kc5tja> That's how I'd do it at least -- it would eliminate a bit of special-case code.
22:17:37 * Serg_Penguin thinks of special FIND
22:18:36 <kc5tja> That could perhaps work as well.
22:18:40 <Serg_Penguin> so i can have 6-bit mask of modes
22:18:55 <Serg_Penguin> if all 6 bits up, word will be executed in any mode
22:19:15 <Serg_Penguin> if all down - never ;)
22:19:26 <kc5tja> Yow!  You have six modes?  Egads!
22:19:42 <Serg_Penguin> i have 2 by now, only idea demo
22:20:02 <Serg_Penguin> in a game, it may be:
22:20:09 <Serg_Penguin> 1) Sail on sea map
22:20:15 <Serg_Penguin> 2) Sea battle
22:20:22 <Serg_Penguin> 3) Sword fencing
22:20:32 <Serg_Penguin> 4) Town walking
22:20:36 <Serg_Penguin> 5) Trade
22:20:54 * kc5tja nods
22:21:03 <Serg_Penguin> and spare one, maybe system options ;)
22:21:42 <Serg_Penguin> oops, + NPC talking ;))
22:22:23 <Serg_Penguin> whats for editor, i prefer to have no modes at all
22:22:26 <kc5tja> Looks like you'll be needing an 8-bit mask after all.  :)
22:23:26 * Serg_Penguin will prefer dict recycling 4 every game mode
22:24:13 <Serg_Penguin> see: i fight on guns in open sea in 3D
22:24:47 <Serg_Penguin> i hit F8 (used to it in Sea Legends) to see local map and give orders to friendly ships
22:25:00 * kc5tja nods
22:25:48 <Serg_Penguin> : #F8 ( ;) " include fitemap.frt" LOAD ; \ ;)))
22:26:22 <Serg_Penguin> it has some control redefined, and shows the same units data in another, 2D format
22:27:17 <Serg_Penguin> also, it has 
22:27:46 <Serg_Penguin> no, no !
22:28:04 <Serg_Penguin> : #027 " unplug" load ;
22:28:17 <Serg_Penguin> in plugins, UNPLUG is bare MARKER
22:28:34 <Serg_Penguin> but the "root" UNPLUG asks if i wanna exit ;))
22:29:12 * kc5tja nods
22:29:44 <Serg_Penguin> btw, here we need 'late redefine'
22:29:46 --- join: gilbertdeb (jqu@24.53.192.88) joined #forth
22:30:40 <Serg_Penguin> for example, 2D map will surely redefine 'SHOW', and i want mainloop to use new one
22:31:42 <Serg_Penguin> the only way i see to do so, is damn weird:
22:31:51 <Serg_Penguin> : SHOW L? ... ;
22:32:58 <Serg_Penguin> : L? R@ 2- >NAME FIND DROP RDROP 2+ EXECUTE ;
22:33:02 <Serg_Penguin> kinda like this
22:33:37 <Serg_Penguin> utter mess, yeah ?
22:33:58 <kc5tja> I don't know the details of how gp-forth works, so I can't tell.
22:34:07 <kc5tja> But I do not disagree with the idea of manipulating the return stack.
22:34:16 <kc5tja> As long as a clear benefit is attained from doing so.
22:34:17 <gilbertdeb> has forth been used in writing device drivers ever?
22:34:33 <Serg_Penguin> L? does:
22:34:36 <kc5tja> gilbertdeb: For something like Linux?  No.  But certainly for embedded devices, yes.  And for OpenFirmware.
22:35:13 <gilbertdeb> kc I'm asking because I saw a jeremiad about the fact that the source code for some of these really ordinary devices are about 5000 lines long!
22:35:35 <kc5tja> What's a jeremiad?
22:35:36 <gilbertdeb> and it got me wondering about how many lines it might take in forth.
22:35:49 <Serg_Penguin> hehe, Chuck's code 4 IDE HDD is 4 lines
22:35:54 <gilbertdeb> jeremiah (TM) complainer par excellence and whiner too 
22:35:58 <Serg_Penguin> but i did not grok it
22:36:02 <gilbertdeb> of book of jeremiah fame.
22:36:13 <kc5tja> Serg_Penguin: It made perfect sense to me. 
22:36:30 <kc5tja> However, remember that Chuck's code does not employ DMA either, so it's also the slowest possible IDE code there is.
22:36:55 <gilbertdeb> serg I just noticed what you wrote...
22:36:57 <gilbertdeb> 4 lines???
22:36:58 <kc5tja> If it used DMA, I'm sure it'd be just a block or two in size.  Still pretty small, but certainly larger than the 4 lines it currently is.
22:37:07 <gilbertdeb> I guess he'll say 'whats the hurry?'
22:37:11 <kc5tja> gilbertdeb: Basically, yes.
22:37:16 <kc5tja> gilbertdeb: Pretty much.
22:37:41 <Serg_Penguin> it's the way how free soft made: 1st - anything working, 2nd - make it work better
22:38:10 <gilbertdeb> kc why do you suppose some of the device drivers have such lengthy code, aside poor factoring etc?
22:38:17 <gilbertdeb> I mean, are these devices THAT complicated?
22:38:31 <gilbertdeb> eg a NIC, or a modem, or a floppy drive or some such?
22:39:39 <Serg_Penguin> floppy - not much more than HDD, and it has no DMA by nature ;)
22:41:28 <kc5tja> Much of the complexity in a device driver is in *scheduling* what will be done.
22:41:46 <kc5tja> A large portion of most device drivers is dedicated to the interface it has with the rest of the system.
22:42:06 <kc5tja> Another large portion is the management of interrupts, a requisite for any kind of real-time response.
22:42:19 <kc5tja> Finally, another large portion is the management of when user requests are completed.
22:42:33 <Serg_Penguin> kc5tja: *scheduling* is 4 core kernel, once 4 all things, interrupts - too
22:42:36 <gilbertdeb> and these are so complicated as to need that many lines of code?
22:42:45 <kc5tja> In a multitasking operating system, you usually cannot just "drop everything" and tend to an applications' request -- that sort of logic has been tried, and usually with horrible results.
22:42:52 <kc5tja> Serg_Penguin: Not true.
22:43:07 <kc5tja> The kernel schedules tasks, but a task is not necessarily an application request.
22:43:33 <Serg_Penguin> so what does driver shedule ?
22:43:43 <kc5tja> Depends on the driver, obviously.
22:43:50 <kc5tja> For disk drivers, it'd be read and write requests.
22:44:03 <kc5tja> To maximize throughput, it's clear that the heads should move as little as possible.
22:44:25 <kc5tja> So an 'elevator algorithm' is used to pack as many I/O requests per cylinder as possible to keep data flowing as quickly as possible.
22:44:33 <kc5tja> The result is substantially improved disk I/O performance.
22:44:54 * Serg_Penguin never digged so deep
22:44:55 <kc5tja> In the case of a networking driver, the driver will schedule when to transmit packets.
22:45:15 <Serg_Penguin> happily, SCSI does it in controller ;))
22:45:25 <kc5tja> Through the use of probabalistic techniques, you can avoid collisions on an Ethernet network by timing packets out, actually improving average throughput.
22:45:33 <kc5tja> Serg_Penguin: Another misconception -- no it doesn't.
22:46:24 * Serg_Penguin has to reboot 4 CHKDISK, dammit !
22:46:25 <kc5tja> The controller usually maintains a cache, but it doesn't enforce any drive access policies.
22:46:38 --- quit: Serg_Penguin (Read error: 104 (Connection reset by peer))
22:47:18 <kc5tja> Operating systems are not simple pieces of software, and much of the code found in device drivers is simply making the device "friendly" to the environment an OS presents to its users.
22:47:41 <gilbertdeb> until I write a device driver and find out exactly why they take up so much code space, I guess I can only sit and wait until all devices have drivers written for them.
22:48:31 <gilbertdeb> I'm having the most annoying time getting plan9 to install on random hardware :(
22:48:43 <kc5tja> When you think about it, 5000 lines isn't a lot of code.
22:48:46 <kc5tja> Not for C anyway.
22:49:07 <gilbertdeb> kc but the UNIX V 6 was about 11,000.
22:49:09 <kc5tja> Much of those lines will be devoted to declarations of data structures, and administravia from the language itself.
22:49:26 <gilbertdeb> that 5000 lines is about the size of an 80 paged novel app.
22:49:40 <kc5tja> gilbertdeb: Unix 6th edition didn't have any device drivers.  It was all hard-coded and hard-wired into the OS.
22:49:44 <gilbertdeb> kc5tja, ha so _managing_ the language itself!
22:50:02 <gilbertdeb> re: administrivia from the language itself.
22:50:07 * kc5tja nods
22:50:13 <gilbertdeb> Hmmmmmm.
22:50:15 <kc5tja> And let's not forget comments too.
22:50:25 <gilbertdeb> grep -v # :)
22:50:25 --- join: Serg_Penguin (Serg_Pengu@212.34.52.140) joined #forth
22:50:36 <kc5tja> grep -v \/\*
22:50:47 <gilbertdeb> that too.
22:51:14 <kc5tja> Remember too that devices back when Unix 6th Edition was around were all standardized to a very large extent -- Unix *only* ran on PDP-7 and PDP-11 machines back then.
22:51:42 <kc5tja> Unix then was a very, very different beast than it is today.
22:51:55 <gilbertdeb> Mkay, I want a pdp-10/11.
22:52:01 <gilbertdeb> I hear they're still being manufactured!
22:52:10 <kc5tja> Not the machines, but their CPUs are.
22:52:25 <gilbertdeb> the cpus? 
22:52:30 <gilbertdeb> they folded it all into a chip?
22:52:38 <kc5tja> But I don't see why you'd want one -- a Motorola 68010 is overwhelmingly similar in machine language and more powerful overall.
22:52:56 <kc5tja> Yes.
22:53:11 <gilbertdeb> just to say 'mind the pdp-10' to my non-existent future guests :)
22:53:20 <kc5tja> Heheh :)
22:53:47 <kc5tja> And here is my coffee table -- it's a PDP-10 running an operating system written in Forth, and is currently running a home-made SETI@Home client.
22:54:28 <gilbertdeb> :)
22:55:15 <kc5tja> But, seriously, my Amiga 500 is more powerful than a PDP-11 -- it addresses more than 16 times the memory, runs about as fast, and can natively process 32-bit data.
22:55:24 <kc5tja> (PDP-11 and earlier are all 16-bit CPU designs)
22:56:13 <hefner> I thought they were 18-bit. Actually, I thought some of them were 36-bit, but I don't know them that well.
22:56:20 <kc5tja> Nope
22:56:23 <kc5tja> Not the PDP series.
22:56:33 <gilbertdeb> so which were the 36 bits?
22:56:36 <kc5tja> The first time DEC breached the 16-bit barrier was with the VAX-11 processor.
22:56:46 <kc5tja> gilbertdeb: IBM's CPUs tend to have really weird sizes.
22:56:53 <kc5tja> CYBER processors were interesting.
22:56:57 <gilbertdeb> the 360?
22:57:19 <kc5tja> The 360 was their first "normal" CPU, I think.  I'm think S/36 here.  :)
22:58:18 <kc5tja> From the 360, IBM later released the 370, and from that their research into RISC was made with the S/800.  The S/800 was canned, though, because it would kill off their existing mainframe business.
22:58:25 * Serg_Penguin wants Forth hobby computer
22:58:25 <Serg_Penguin> if under $100, i would buy right now ;))
22:58:40 <gilbertdeb> as described in Automatic Data Processing by blaauw and Iverson with APL?
22:58:46 <kc5tja> But, the S/800 was resurrected in semiconductor form for their IBM PC/RT system, and from that came the RS/6000 line.  Then, the PowerPC, and there you go.
22:59:10 <gilbertdeb> s/blaauw/brookes
22:59:48 <kc5tja> Serg_Penguin: I'm interested building one myself.  I'm looking currently at using the 65816 as the central processor for it.
23:00:00 <gilbertdeb> kc5tja, have you ever read the book by blaauw and brookes on computer architecture?
23:00:16 <kc5tja> gilbertdeb: Nope.
23:00:39 <gilbertdeb> they describe the machines in APL.
23:00:42 <Serg_Penguin> kc5tja: u need to design it w/ great modularity
23:00:49 <gilbertdeb> Brookes is the same brookes from Mythical Man MOnth.
23:01:06 <Serg_Penguin> so it maybe breaking win as IBM PC
23:01:21 <kc5tja> Serg_Penguin: I'm not convinced that modularity truely helps.
23:01:40 <kc5tja> But even so, nobody anywhere will ever build peripheral equipment for it except me, so it's a moot point anyway.
23:01:55 <Serg_Penguin> here in RU, folks knee-make Speccy modular way ;)
23:02:08 <Serg_Penguin> despite it was 1-board originally
23:02:13 <kc5tja> Serg_Penguin: Making hardware modular requires money.
23:02:37 <Serg_Penguin> hmm..
23:02:38 <kc5tja> And the speccy wasn't a terribly interesting computer anyway.
23:03:02 <Serg_Penguin> i would never design anything to be condemned to 1 unit
23:03:03 <gilbertdeb> whats the speccy ?
23:03:09 <Serg_Penguin> ZX Spectrum
23:03:30 <gilbertdeb> from sinclair?
23:03:38 <kc5tja> Serg_Penguin: Despite that single units are known to be easier for people to operate, more reliable overall, and are overwhelmingly more compact.
23:03:49 <Serg_Penguin> oops
23:03:57 <Serg_Penguin> i meant...
23:04:10 <kc5tja> Yes, the Timex/Sinclare ZX-1000 Spectrum, IIRC.
23:05:04 <Serg_Penguin> if i would design a comp, i'll try to make it a great hit to be built by many hobbysts
23:05:26 <Serg_Penguin> kc5tja: think BIG
23:05:29 <kc5tja> Serg_Penguin: Check out http://www.6502.org -- many have beat you to it.  :)
23:05:39 <kc5tja> Serg_Penguin: Thinking BIG gets EXPEN$IVE.
23:06:40 <Serg_Penguin> BIG i mean not 'beat P4 in power', but 'beat any hobby home-made comp in quantity of units built worldvide'
23:06:56 <kc5tja> Why?
23:07:02 <kc5tja> That's not my goal.
23:07:17 <gilbertdeb> serge to starve out the other designs?
23:07:30 <kc5tja> My goal is to build a small PC that I can be proud of, that has adequate capabilities for my needs, that doesn't suck even my Amiga's 25W of power, and which lets me program every nuance of the machine to its fullest potential.
23:07:50 <Serg_Penguin> even if it'll remain yours only, it will give u GOOD design guidelines
23:07:52 <gilbertdeb> ah kc, you want Asm programming :)
23:07:59 <kc5tja> No, it won't.
23:08:02 <kc5tja> In fact, it'll ruin them.
23:08:07 <kc5tja> "No hooks." -- Chuck Moore
23:08:17 <kc5tja> "YAGNI -- You Aren't Gonna Need It" -- Extreme Programming
23:08:54 <Serg_Penguin> it'll make you analyse WHY hits became hits, and so educate self greatly
23:09:01 <kc5tja> gilbertdeb: Assembly, Forth, doesn't matter.  Forth is every bit as good about that as assembly is.
23:09:13 <kc5tja> Serg_Penguin: It has nothing at all to do with it.
23:09:28 * kc5tja points to the Apple II series, which had an ISA-like expansion bus YEARS before IBM PC did.
23:09:43 <kc5tja> Yet, the all-in-one Commodore 64 utterly *destroyed* the Apple II's sales records.
23:09:59 <kc5tja> The PC didn't win on technical merit.
23:10:08 <kc5tja> It won because IBM sold it.
23:10:16 <kc5tja> And nobody ever got fired for using IBM equipment.
23:10:20 <gilbertdeb> the pc won by sheer momentum.
23:10:30 <gilbertdeb> the word juggernaut comes to mind.
23:10:35 * kc5tja nods
23:10:40 <kc5tja> Like Microsoft of today.
23:10:54 <kc5tja> Still, the Commodore 64 is, to this day, the all-time best-selling home computer ever.
23:10:55 <Serg_Penguin> so, maybe i wrong
23:11:25 <Serg_Penguin> so, try to analyse and repeat ham radio hits - ...
23:11:40 <Serg_Penguin> the units repeated by hobbysts in most quantitys
23:11:50 <kc5tja> which are all single-unit devices.
23:11:51 <kc5tja> :)
23:12:07 <kc5tja> The TenTec Pegasus is the rare exception.
23:12:17 <gilbertdeb> but kc aren't these things though so much more complicated in terms of their circuitry?
23:12:30 <gilbertdeb> n/m it is all boolean factoring.
23:12:38 <kc5tja> gilbertdeb: No, actually, homebrew QRP rigs are pretty doggone simple.  Chuck Moore would be quite proud.
23:12:57 * kc5tja isn't against modularity.
23:12:59 <gilbertdeb> kc5tja, I bet he's built one or two in his time.
23:13:12 <kc5tja> But I'm not going to design in modularity at some cost to me, when I know full well I won't get any return on my investment.
23:13:29 <kc5tja> gilbertdeb: I can't comment.  But something tells me, he didn't.
23:13:36 <Serg_Penguin> did u read 'Thinking Forth ?'
23:13:37 <gilbertdeb> heh Forth?
23:13:38 <kc5tja> He doesn't seem to be that kind of guy.
23:13:55 <kc5tja> Serg_Penguin: I cannot locate a copy of the book.  
23:14:03 <gilbertdeb> even on ebay?
23:14:10 <gilbertdeb> I have a copy here I"m treasuring :D
23:14:28 <kc5tja> However, even so, it's from the mind of Leo Brodie, not Chuck Moore.  There are some philosophical differences.
23:14:28 <gilbertdeb> I dust it off every now and then and coo admiringly at the line drawings.
23:14:46 * kc5tja has a copy of Starting Forth lying around here somewhere though.
23:15:04 <gilbertdeb> much of that book has been re-written and is somewhere on the web.
23:15:04 <Serg_Penguin> this book contains excellent example of how hardware modularity won over single-unit unique design
23:15:14 <kc5tja> Ebay has no such book.
23:15:45 <Serg_Penguin> factoring RULES ;)
23:15:47 * kc5tja must be talking to a wall.
23:15:56 <kc5tja> Custom design does not mean unfactored!!!!!
23:16:29 <kc5tja> I am not going to invest even $15 worth of electronics that I *KNOW*, ahead of time, I'll get *ZERO* payback from.
23:16:33 <kc5tja> It's a WASTE.
23:16:56 <Serg_Penguin> thinking this way, any hobby is waste
23:17:01 <kc5tja> I don't know if you know, but I have no idea if I can even be online in another month or two, because my income is non-existant right now.
23:17:33 <kc5tja> After paying this month's rent, I'll be $100 or so in the hole.  I have to recoup $100 just to bring my account balance to zero
23:17:45 <kc5tja> And I have no idea where that $100 is going to come from.
23:17:56 <Serg_Penguin> so, definitely, put any kind of hobby in a "long-time box" and work ;((
23:17:58 <kc5tja> Well, $70 now -- I made my $30 from a customer last week.
23:18:15 <gilbertdeb> school is expensive unfortunately.
23:18:17 <kc5tja> Serg_Penguin: But that's besides the point.  Obviously, I'm not spending money on hobbies right now.
23:18:31 <kc5tja> But let's say I have enough cash to spend.
23:18:41 <kc5tja> Does this justify me making my computer more expensive than it needs to be?
23:19:11 <kc5tja> If fitting everything on a single PCB saves me $100, you better believe I'll do it.
23:19:18 <kc5tja> Especially at the CPU speeds I'm intending on running this thing.
23:19:27 <gilbertdeb> kc5tja it took woz to build the Apple1 whereas it took a team to build the ibm/pc.
23:19:38 <gilbertdeb> I think the constraints improve creativity significantly.
23:19:47 <kc5tja> gilbertdeb: The Apple-I was a single board computer too, I might add.
23:20:11 <kc5tja> Moreover, he was limited to running the CPU at 1MHz.
23:20:17 <kc5tja> I'm planning on driving mine at least at 12.5MHz.
23:20:21 <gilbertdeb> single board, one man, craftsmanship.
23:20:25 <kc5tja> That's a huge jump in clock speed, one not to be taken lightly.
23:20:36 <gilbertdeb> Thats the word 'craftsmanship'.
23:20:41 <gilbertdeb> hehehe
23:22:15 <kc5tja> And then there is the VGA display for it.  And no, I won't go TV -- too complex, and TVs are big, bloated, power-hungry hogs that give me a headache because, unlike most humans, I CAN hear the 15750Hz tone their flyback transformers give off.
23:23:12 <Serg_Penguin> cellphone LCD ?
23:23:22 <kc5tja> Not suitable for what I want to do.
23:23:28 <kc5tja> WAY too low a resolution.
23:23:59 <kc5tja> (and way too slow a screen update too -- the pixels transition very slowly)
23:24:07 <Serg_Penguin> IMHO, 320x240 75Hz is optimal
23:24:25 <kc5tja> VGA is clean, simple, efficient, and . . . did I mention simple?
23:24:28 <gilbertdeb> kc it is MORE complicated to make it display on a TV than on a monitor?
23:24:40 <kc5tja> gilbertdeb: absolutely.
23:24:57 <Serg_Penguin> TV fries eyes, i won't look in it
23:25:03 <kc5tja> gilbertdeb: The TV signal must encode both luminence and chroma signals into a single waveform, plus syncing information and color burst information.
23:25:32 <kc5tja> gilbertdeb: VGA, OTOH, has five signals: horizontal sync (TTL level), vertical sync (TTL level), and three analog direct-drive signals: Red, Green, and Blue.
23:25:37 <kc5tja> (1V peak)
23:25:53 <hefner> are you going to build your own hardware to generate the video?
23:26:20 <kc5tja> hefner: Of course.  Commercial VGA chips are so heavily geared towards PC interfaces that it renders them useless.
23:26:29 <gilbertdeb> kc5tja, so basically the Apple1 was a needlessly complicated machine :)
23:26:37 <kc5tja> Plus, they're horrifically unorthogonal.
23:26:48 <hefner> I've heard of people yanking the chips off old CGA cards for their homebrew projects, must be a simple interface.
23:26:49 <Serg_Penguin> unorthogonal ????
23:26:53 <kc5tja> gilbertdeb: Compared to a VGA monitor, yes.
23:27:03 <hefner> probably you want something more sophisticated, though :)
23:27:15 <kc5tja> Serg_Penguin: The PC's video cards generally cannot handle such "advanced" things as raster compare interrupts, vertical sync interrupts, etc.
23:27:53 <Serg_Penguin> IIRC, EGA had _line_scan interrupt
23:27:59 <kc5tja> Two things: one, the CRTC chip is a standard component: the Motorola 6845.
23:28:10 <Serg_Penguin> VGA didi not, so some old games wait it 4ever
23:28:10 <kc5tja> Serg_Penguin: Not all of them.  IBM made provision for it, but it wasn't universally supported.
23:28:22 <kc5tja> Serg_Penguin: Again, it was optional for VGA.
23:28:28 <kc5tja> Not all SVGA cards have them either.
23:28:50 <Serg_Penguin> mine old S3 had not  ;(
23:28:58 <kc5tja> But the 6845 doesn't do serialization of video data -- it simply handles (in a very crude way) display refresh.
23:29:07 <kc5tja> And it patently doesn't support raster line compare interrupts.  :/
23:29:18 <Serg_Penguin> line compare ??
23:29:29 <kc5tja> Yes
23:29:43 <kc5tja> You can tell the video card, "When drawing screen line #45, interrupt me."
23:29:54 <Serg_Penguin> ok
23:29:59 <kc5tja> This is how split-screen and multi-color-palette affects were done on the 8-bit computers of the time.
23:30:28 <kc5tja> A related technology was used in the Amiga to make draggable screens, each with their own palette and video resolution.
23:30:36 <kc5tja> You could drag physical screens around the screen like windows.
23:30:44 <Serg_Penguin> will u make co-processors, like Amiga blitter ?
23:31:01 <kc5tja> Serg_Penguin: Not sure.  Depends on how much it costs to do so.  While it's nice, it is not absolutely necessary.
23:31:53 <Serg_Penguin> in what cases chips will be ? DIP ?
23:32:06 <kc5tja> For ease of assembly, quite likely.
23:32:22 * kc5tja may go with PLCC for the CPU though, considering its high clock rate.
23:33:36 * kc5tja may also go with the Cray-like construction technique of building it in a cylinder form instead of a box form, too.
23:33:53 <kc5tja> This is to keep lead lengths to an absolute minimum.
23:34:14 <gilbertdeb> if you wanna go ALL cray, the machine would have to be part of the airconditioning of your apartment.
23:34:25 <kc5tja> At 12.5MHz, it'll be smack in the middle of the HF shortwave spectrum.
23:34:56 <Serg_Penguin> ha-ha, comp clocked by TRX ;))
23:35:00 <kc5tja> gilbertdeb: Cray I was independent of any air conditioning unit -- it had a fully integrated LN2 cooling system.
23:35:34 <gilbertdeb> kc5tja, hence Cray 2. 
23:35:39 <gilbertdeb> 1 was an error ;)
23:35:50 <gilbertdeb> j/k
23:35:52 <kc5tja> The whole point of building it (semi-)cylindrically is to minimize RF hash, which seems to be an annoying constant of PCs made today.
23:36:02 <kc5tja> The Cray-I was a wonderful piece of fine computing equipment.
23:36:15 <gilbertdeb> that and the CDC 6600.
23:36:19 * kc5tja nods
23:36:28 <kc5tja> The 6600 reminds me of the 6502, actually, only a 64-bit version of it.  :)
23:36:43 <gilbertdeb> it was one of the first RISC's wasn't it?
23:36:46 <gilbertdeb> the 6600?
23:36:58 <kc5tja> Not quite, but it certainly was quite a bit simpler than its competitors.
23:37:09 <Serg_Penguin> ham men here learned to shield QRN from PC below air noice ;))
23:37:09 <kc5tja> It was also the first multi-threaded architecture IIRC.
23:37:12 <kc5tja> And the first pipelined.
23:37:19 <gilbertdeb> ah yes, pipelining.
23:37:28 <gilbertdeb> thats what it majorly inspired from what I read.
23:37:30 <kc5tja> Serg_Penguin: I'm talking RF noise.
23:37:49 * kc5tja nods
23:37:51 * Serg_Penguin means RF too, not acoustic noice
23:37:54 <kc5tja> It had four pipeline stages.
23:37:58 <kc5tja> It also had four execution units.
23:38:23 <kc5tja> It worked by continually pumping the state of the four execution "units" through the four pipeline stages.
23:38:36 <gilbertdeb> kc5tja, the remarkable thing about the Cray1 and I think the 6600 is how few people it took to make the whole machine!
23:38:38 <gilbertdeb> 13!
23:38:42 <gilbertdeb> including Cray.
23:38:54 <kc5tja> So a single context woudl go from instruction fetch stage, to decode stage, to execute stage, to write-back stage, and then back to instruction-fetch stage.
23:39:03 * kc5tja nods
23:39:35 <gilbertdeb> he said he was designing his latest machine on a mac at a time when apple was designing their latest machine on a cray :)
23:39:49 <gilbertdeb> he was a craftsman, they were a factory.
23:39:49 * kc5tja nods
23:40:08 <kc5tja> Which brings to mind my utter distaste and disdain for our modern educational system.
23:40:20 <Serg_Penguin> kc5tja: whadda u think of making all new CPU from PLM ?
23:40:32 <kc5tja> Call me old-fashioned, but the master/journeyman/apprentice system is NOT that bad a system.  In many, many, many ways, it's overwhelmingly superior to our current educational system.
23:40:34 <kc5tja> But I digress.
23:40:46 <kc5tja> What is PLM?
23:41:03 <Serg_Penguin> Prog'able Logic Matrix ?
23:41:17 <gilbertdeb> kc5tja, I read an article in smithsonian about a guild that uses the same methods to train master carpenters, sculptors, cooks, etc.
23:41:31 <kc5tja> Guilds -- that's the word I was looking for.
23:41:34 <Serg_Penguin> big chip w/ logic elements what can be softly rewired
23:41:37 <kc5tja> Serg_Penguin: I'm all for it.
23:41:45 <kc5tja> I think everyone who is even remotely interested in digital logic ought to do it.
23:42:01 <Serg_Penguin> so, u will design own CPU ?
23:42:04 <kc5tja> There are two problems with it though:
23:42:10 <kc5tja> 1.  Expensive.
23:42:13 <kc5tja> VERY expensive.
23:42:25 <kc5tja> For a *minimal* system, you're looking for an investment AT LEAST $300.
23:42:37 <kc5tja> For a single-quantity part, well, ....  That's expensive.  :)
23:42:45 <kc5tja> 2.  one CPU, one part.
23:42:59 <kc5tja> You lose the fun-factor of wiring up discrete components, and watching the system come alive.
23:43:12 <Serg_Penguin> here in RU, some ZX reincarnation is 100$, having two
23:43:21 <Serg_Penguin> 1- CPU, 2- "chipset"
23:43:28 <kc5tja> I don't feel so much like a craftsman (to use Gilbertdeb's terms) if I just slap a design in the computer, and have it do all the work for me.
23:43:43 <gilbertdeb> kc5tja, you have a Victor Frankenstein complex don't you ;)
23:43:48 <kc5tja> No.
23:43:53 <gilbertdeb> j/k
23:43:56 <kc5tja> But I take a great deal of pride in the work I do.
23:44:09 <kc5tja> For example, I'd love to implement some of my CPU ideas in an FPGA.
23:44:12 <gilbertdeb> I admired frankenstein until he abandoned his creature.
23:44:18 <kc5tja> And I do really want to make my own stack architecture CPU.
23:44:43 <kc5tja> I know for a fact a stack CPU will beat the 6502/65816 by a factor of at least 2, clock for clock.
23:45:08 <kc5tja> But I also want to build my own CPU from discrete components too.  And a stack machine is the only way to go.
23:45:22 <kc5tja> The last time I got my 4-bit CPU working with discrete logic, I was ecstatic.
23:45:47 <kc5tja> Damn, I'm hungry, and there is nothing here for me to eat.
23:45:55 <Serg_Penguin> how many chips, and what kind of ?
23:46:10 * Serg_Penguin is damn thirsty - 32 C hot here ;)
23:46:45 <kc5tja> Moderate amount -- all kinds of chips used.
23:46:59 <kc5tja> Registers, decoders, shift-registers, combinatorial logic, . . .
23:47:07 <gilbertdeb> hmmm. on a bread board?
23:47:12 <kc5tja> Yup
23:47:20 <kc5tja> It didn't run very fast -- only about 100Hz or so.
23:47:36 <kc5tja> It could easily have handled at least 750kHz though.
23:47:46 <gilbertdeb> hmm. you should put it up for all to see.
23:47:46 <kc5tja> I just didn't drive it that fast, because I wanted to enjoy watching it actually work.
23:47:54 <kc5tja> The design has been lost.
23:47:59 <kc5tja> I was a teenager when I built it.
23:48:00 <gilbertdeb> whaaa?
23:48:21 --- quit: hefner (Read error: 60 (Operation timed out))
23:48:31 <kc5tja> It didn't do anything useful anyway.
23:48:36 <kc5tja> Not really Turing complete at all.
23:48:49 <kc5tja> Though had I known about stack architectures, I could have made it Turing complete.
23:49:33 <gilbertdeb> aha:- http://www.smithsonianmag.com/smithsonian/issues96/jun96/compagn.html
23:49:40 <gilbertdeb> thats the article i was referring to re: guilds
23:49:46 <gilbertdeb> it is a 2 minute read.
23:50:02 <kc5tja> I'll read it when I have time.  :)
23:50:06 <gilbertdeb> smithsonian does not put the full article online.
23:50:09 <kc5tja> Right now, I'm arguing frivolous things on IRC.
23:50:18 <gilbertdeb> hehehe.
23:50:19 <kc5tja> :)
23:50:29 <gilbertdeb> it is a 30 second read actually.
23:50:31 <gilbertdeb> :)
23:50:45 <gilbertdeb> only the abstract :(
23:50:52 <gilbertdeb> I hate it when they do that.
23:52:13 * Serg_Penguin knows one poor boy who runs P100 w/o HDD ;((
23:52:17 * kc5tja has often considered the concept of turning the computer industry into a computer guild.
23:52:51 <gilbertdeb> kc5tja, the dreamsongs.com guy agrees with you somewhat.
23:53:03 <gilbertdeb> richard gabriel
23:53:07 <Serg_Penguin> now i'll try to make usable ramdisk bootflop 4 him ;)
23:53:27 * kc5tja enjoyed his freedom as a child, but in retrospect, I think being brought up in a guild would have been the right thing for me.
23:54:15 <gilbertdeb> that would have been awesome.
23:54:25 <gilbertdeb> a Forth guild? 
23:54:27 <gilbertdeb> :D
23:54:37 * Serg_Penguin has nostalgia of my old work in trash hardware firm ;)
23:55:02 <Serg_Penguin> i have some trash to test here ;)
23:55:07 * kc5tja doesn't know about a Forth guild.
23:55:17 * kc5tja would have to receive my certification from Moore.
23:56:07 <gilbertdeb> heheh.
23:56:12 <gilbertdeb> what would that involve?
23:56:22 <gilbertdeb> a chip colorforth?
23:58:50 * Serg_Penguin revived 1 CD-ROM - exactly the model i had in my old P166
23:59:58 <gilbertdeb> I have a 15 page paper to write by 9 am.
23:59:59 --- log: ended forth/03.07.28