00:00:00 --- log: started forth/03.10.09 00:34:25 hi mmanning ! 00:34:40 hi mur ;)) completion err 00:34:42 privet 00:34:47 he 00:34:49 h 00:37:54 Hi Serg_Penguin 00:37:58 And murmanning 00:41:01 robushka 00:41:06 Boo 00:41:10 Time for school... 00:41:19 learn much 00:41:30 I won't. 00:41:30 Bye 00:58:12 --- quit: Serg_Penguin () 00:59:06 --- join: LOOP-HOG (TofuMonste@207.191.240.11) joined #forth 01:13:33 --- part: njd left #forth 01:25:28 --- quit: LOOP-HOG () 01:37:58 --- quit: onetom (Remote closed the connection) 02:37:29 --- join: Serg_Penguin (~z@212.34.52.140) joined #forth 02:40:47 --- join: onetom (~tom@cab.bio.u-szeged.hu) joined #forth 02:43:02 --- nick: mur -> mureat 02:51:35 --- join: schihei (~schihei@p5085DD0D.dip.t-dialin.net) joined #forth 03:55:27 --- join: augur (~hulla@200.217.158.174) joined #forth 03:56:26 --- join: segher (~segher@blueice1a.de.ibm.com) joined #forth 04:11:15 --- nick: augur -> augur-off 04:17:24 --- quit: Serg_Penguin () 04:55:05 --- quit: I440r_ () 05:13:57 --- quit: augur-off (Read error: 60 (Operation timed out)) 06:32:13 --- nick: mureat -> mur 07:26:28 --- join: yasam (~sam@210-55-151-39.dialup.xtra.co.nz) joined #forth 07:26:41 --- quit: mmanning (Read error: 104 (Connection reset by peer)) 07:35:47 --- join: mmanning (~mmanning@saturn.vcsd.com) joined #forth 07:51:18 --- join: richard (~richard@adsl-19-180-125.shv.bellsouth.net) joined #forth 07:51:24 --- nick: richard -> snowrichard 07:51:43 hey dudes be right back 07:52:21 --- quit: mmanning ("User pushed the X - because it's Xtra, baby") 07:52:45 I need a beta tester. :) 07:54:02 anybody awake? 07:54:14 no 07:54:18 i wish i was sleeping 07:54:21 work not yet done 07:54:43 what's up? 07:54:50 can't betatest 07:54:53 * mur at uni 07:55:44 too bad mur. email me at richardsnow@bellsouth.net if you get any volunteer to help out. Name in credits, glory all that 07:56:08 :) 07:56:18 heh 07:56:43 new michael release 07:57:04 I defined a HELP word that cranks up mozilla and brings up a web page. cool eh? 07:58:00 so my language is sort of internet enabled already 07:58:49 manual not written yet but I have a quickstart guide 08:00:04 michael, I havent even heard of it, url? 08:01:56 http://veteransinaction.org/quickstart.html 08:05:40 veteransinaction.org/michael/michaelV0.0.tar.gz 08:05:47 source code tarball 08:06:11 this: http://www.veteransinaction.org/michael/index.html goes to the right place? 08:08:12 wrong index page hold on 08:09:58 http://veteransinaction.org/michael/index.html 08:11:44 if you didnt know, it seems to work under Mac OS X 08:13:53 http://veteransinaction.org/michael/index.html 08:15:50 if test.f should output 45 that is :) 08:20:52 well cool then contribute a MAC OS X install guide and I'll give you credit 08:21:06 richardsnow@bellsouth.net 08:21:10 the installation is identical 08:21:14 good 08:21:27 I'll make a note of that 08:22:11 michael2>"mozilla http://veteransinaction.org/michael/quickstart.html" SYSTEM 08:22:31 should bring up the new quickstart guide 08:23:35 404 08:23:37 I will write a tutorial type manual and announce on freshmeat as V0.1 08:23:58 I meant index.html for the correct index page sorry 08:24:24 but you can bring up mozilla on os/x ? cool! 08:25:44 quickstart guide is : http://veteransinaction.org/quickstart.html 08:26:39 actually no I just tried the link, mozilla command not found 08:28:42 ok well sub your browser of choice the system command should work though 08:29:25 try "ls" SYSTEM and see if you get a directory 08:29:58 yes 08:30:01 cool 08:30:38 so you can define a michael word that runs a unix command script ... 08:30:44 I dont have a browser which I can invoke from the command line with an url as argument 08:30:51 cool 08:31:00 : dir "ls" SYSTEM ; 08:32:01 have you done any progamming for your computer? 08:32:43 yes 08:32:52 feel free to submit customization for mac platform I don't have one to test on 08:33:27 that version can have your name in copyright notice and docs of course 08:33:28 ok, Im new to the unix side of things, but am learning 08:34:26 richardsnow@bellsouth.net would like to chat later need to take break. 08:34:27 --- join: andreou (~panx@195.130.107.57) joined #forth 08:34:34 email 08:34:39 hey 08:34:47 brb 08:34:48 --- join: rO|_ (~rO|@pD9E596BE.dip.t-dialin.net) joined #forth 08:38:06 back 08:38:32 fixed the dl/link http://www.veteransinaction.org/michael/index.html now had it 08:38:38 correct tarball link 08:39:43 thanks for help 08:42:59 --- quit: rO| (Read error: 60 (Operation timed out)) 08:43:17 no problem, looks very interesting, I dont know If Ill get the time to muck around (I hope so), if I do Ill be sure to report bugs etc 08:54:45 --- join: mmanning (~mmanning@saturn.vcsd.com) joined #forth 09:02:41 even better, even I can understand the source ;) 09:04:34 :) 09:18:19 --- join: augur (~hulla@200.217.158.174) joined #forth 09:20:50 you can read my code? great! 09:21:26 need to put it into CVS and do a commenting session marathon 09:21:57 most comments that are there now are as a result of bugs :) 09:23:21 if you can't laugh at your self when programming it is not fun 09:24:57 need to log off for lunch and then do manual. later 09:25:21 --- quit: snowrichard ("Client Exiting") 09:45:58 --- quit: andreou ("ta kanonika paidia pai8ainoyn kanonika. ta diaforetika paidia pe8ainoyn eley8era. (A)") 10:02:06 --- nick: augur -> augur-off 10:46:00 --- quit: yasam ("dumdumdumdum") 10:51:54 --- quit: augur-off ("...veritas virus est") 10:52:42 --- join: augur (~unknown@200.217.158.174) joined #forth 11:30:55 --- quit: segher ("catching bus") 11:32:07 --- quit: schihei (Client Quit) 12:19:50 --- join: snowrichard (~richard@adsl-19-180-125.shv.bellsouth.net) joined #forth 12:20:04 hey 12:20:07 how is it going? 12:22:02 Hi 12:22:14 I'm leaving soon, for a math seminar paid by my school. :) 12:22:16 y0 12:28:27 --- quit: augur ("...forth rlz") 12:37:00 hey still here/ 12:37:02 ? 12:37:12 i was in #lunar 12:37:28 they have added a michael beta package to their distro 12:37:55 manual started, a version uploaded with a glossary 12:38:49 freshneat announcement submitted, not reviewed yet 12:39:16 freshmeat 12:39:32 brb 12:40:21 --- quit: snowrichard ("Client Exiting") 13:06:00 y0 13:09:58 --- join: augur (~unknown@200.217.158.174) joined #forth 13:13:39 --- join: tathi (~josh@pcp02123722pcs.milfrd01.pa.comcast.net) joined #forth 13:19:57 hrm 13:20:04 GPf is getting weird 13:22:05 GPf? 13:24:35 --- join: augur_ (~unknown@200.217.158.174) joined #forth 13:25:35 --- quit: augur_ (Client Quit) 13:25:58 general protection fault 13:26:00 gpf-comics.com 13:27:11 ah 13:28:30 --- quit: augur (Read error: 60 (Operation timed out)) 13:28:58 its great :) 13:29:24 cool 13:29:30 most online comics suck :) 13:37:46 --- quit: tathi ("Lost terminal") 14:00:46 --- join: schihei (~schihei@p5085DD0D.dip.t-dialin.net) joined #forth 14:04:59 --- join: augur (~augur@200.217.158.174) joined #forth 14:17:23 --- join: augur_ (~augur@200.217.158.174) joined #forth 14:22:02 --- quit: augur (Read error: 60 (Operation timed out)) 14:23:34 --- nick: augur_ -> augur 14:37:10 --- quit: schihei (Client Quit) 14:45:42 --- quit: augur (Read error: 60 (Operation timed out)) 15:01:08 --- quit: arke ("My damn controlling terminal disappeared!") 15:03:47 --- join: Sonarman (~matt@adsl-67-113-234-137.dsl.snfc21.pacbell.net) joined #forth 16:14:34 --- join: kc5tja (~kc5tja@66-91-231-74.san.rr.com) joined #forth 16:14:34 --- mode: ChanServ set +o kc5tja 16:15:15 --- join: thin (thin@bespin.org) joined #forth 16:15:23 re thin 16:15:37 * kc5tja just got back from, ironically, doing Falvo Technical Solutions work. :/ 16:15:56 Had to install Windows 98 on this guy's box, on a new harddrive. 16:15:59 ironically? 16:16:15 Yeah, because I'm going to be closing Falvo Technical Solutions down. 16:16:19 ah 16:16:22 (to make room for the kit business) 16:16:52 so how many hours is the in-n-out taking of your time? 16:17:06 in-n-out always makes me think of the sexual innuendo.. 16:17:13 (aspect) 16:17:43 18 16:18:14 does that even cover your living expenses? 16:18:39 No, but it helps. 16:18:47 It's better than NO job. 16:19:02 are you also doing school? 16:19:26 not much time to do FTS is there? 16:19:32 Yes. 16:20:02 The biggest problem is scheduling around my school. It was hard for us to hook up, because he has UCSD, and I have Mira Costa, and our schedules were completely incompatible. 16:20:14 Plus, he's a pro musician, so he also has practice and touring to do. 16:20:26 Needless to say, it took me 1.5 weeks to schedule a drop-off time with this person. 16:20:32 The job took an hour. :) 16:20:55 hmm, usually takes me a bit longer than an hour to install win98 16:21:03 unless the comp & hdd is faster i guess 16:21:21 7200 RPM harddrive. 16:21:22 60GB 16:21:32 yeah prolly faster than my comp 16:21:42 i reinstall win98 once a year :D 16:21:55 that's why i got a 1 gig partition for win98 and all the standard apps like microsoft office 16:22:02 and a 39 gig partition for all my content, anime, etc 16:23:03 most weenies would need more than 1 gig partition for win98 and all the standard apps 16:23:08 but i am leet 16:23:15 and i don't install stupid stuff :P 16:29:52 :) 16:29:53 Yeah. 16:30:09 I miss proper partitioning. 16:30:16 The PCs make it virtually useless. 16:30:35 On AmigaOS, it was not only considered good practice, but it was considered the ideal thing to do. 16:30:37 beb -- phone 16:47:15 --- topic: set to 'comp.lang.forth.fast' by thin 16:47:19 --- quit: thin ("Leaving") 16:49:46 back 16:49:50 doh 17:30:18 --- join: TheBlueWizard (TheBlueWiz@207.111.96.69) joined #forth 17:30:18 --- mode: ChanServ set +o TheBlueWizard 17:30:22 hiya all 17:33:04 Howdy 17:33:54 hiya kc5tja...how're things going? 17:34:14 Not well. 17:34:18 I'm burning cash. 17:34:28 Job at In-n-Out isn't covering things. 17:34:30 ouch! 17:34:34 It's HELPING, but it isn't covering things. 17:34:43 is that a restaurant? 17:35:55 Yes 17:36:02 But a reasonably well-paying restaraunt. 17:36:09 It's a burger place. 17:36:12 I see 17:36:33 http://www.in-n-out.com 17:36:34 :) 17:36:50 presumably better than McDonald's...though of course that ain't covering the rent 17:37:20 In-n-Out's quality is so high that it's nonsensical to even compare them to McDonalds. 17:37:33 Carl's Jr, Burger King, yeah, those can be compared to McDonalds. 17:37:36 But not In-N-Out. 17:37:41 And I'm not saying this beacuse I work for them. 17:37:52 I'm saying this based on the 6+ years I've lived in California and eaten there. :D 17:38:02 And here's the deal -- nothing in the store is ever frozen. 17:38:03 Ever. 17:38:42 hmm...I know nothing about In-n-Out, so... 17:39:58 and with Arnold elected to the office, maybe there will be a turn around...though I wouldn't hold my breath on it...so many hand tyings 17:40:24 * TheBlueWizard waves to I440r and mur...in case they're awake :) 17:42:04 I don't believe anything at all will come of it -- in fact, I only expect the situation to worsten. 17:42:14 hmm 17:42:24 if so....not good then! 17:42:33 Republicans are conservatives -- by definition, they want things "the old way." 17:42:56 Expect strong support for DMCA and monopoly. 17:43:20 generally true...but not always...and then again, Arnold is no regular politician 17:43:39 Yeah, he calls himself a "Moderate" Republican. 17:43:40 maybe 17:44:09 (re: supposed support for DMCA and monopoly) 17:44:35 just too much unknowns right now 17:45:12 * kc5tja nods 17:46:45 but I, as an outsider, must admit I really enjoy the spectacle :) 17:47:26 He 17:47:27 Heh 17:47:29 even 17:48:25 is there such a newsgroup comp.lang.forth.fast? 17:48:39 I'm not aware of any. 17:48:40 * kc5tja checks 17:49:14 i tried to find it, but it wasn't there (not with my isp, anyway) 17:51:50 Nope. 17:51:55 IT doesn't exist. 17:52:10 anyway) 17:52:33 sorry, was playing with the mouse again, and middle clicked :) 17:52:52 :) 17:53:04 Dang it, I wanted to head to the bank today. 17:53:09 Gahh....just too much crap going on! 17:53:26 * kc5tja will go tomorrow. 18:23:23 kc5tja: remember the discussion about COM? I remember you said you wrote a version of it for Amiga Inc. 18:24:28 --- quit: fridge ("http://lice.codehack.com") 18:25:24 anyway, I am wondering about your opinion of COM technology...is it good or bad? 18:27:10 well, gotta go...bye all 18:27:25 --- part: TheBlueWizard left #forth 18:35:20 --- join: suprdupr (CrowKilr@HSE-Windsor-ppp252334.sympatico.ca) joined #forth 18:35:32 hi 18:37:12 What have you guys been discussing lately? any breakthrough? ;p 18:38:35 hmm ill ask clog instead ;p 18:39:51 re 18:40:24 Nothing of any significant consequence. :) 18:40:52 im trying to develop a nocall compiler 18:41:22 No-call? 18:41:25 call call call is less efficient than push push push ret or push push push jmp 18:41:53 i use the data stack at compile time to store compiled word's addresses 18:42:03 i stack em on top of zero 18:44:06 so i use a test on the first item on stack after doing the 4, if its zero i convert it to a jump 18:44:17 i optimize for 2 bytes jumps 18:44:24 OK but how does the data get put on the stack to begin with? 18:44:32 tokens 18:44:40 Let me rephrase. 18:44:45 my green tokens simply fetch from the Index on the stack 18:45:01 When invoking a "word" in your environment, what places the list of words to execute on the return stack at run-time? 18:45:39 ah ok, the runtime is a classical forth vm, only the compiler use the data stack that way 18:46:12 So the compiled image of a word still uses call call call call then... 18:46:58 nope, only "pushes jumps" and for when calls are necessary a push FORWARD push push jump 18:47:20 Then I repeat my question: at run-time, how does one invoke a word if you have no calls? 18:47:24 forwards means after the jump is compiled the label is overwritten with a forward absolute adress 18:48:05 hmmm the call mechnism still exists, but no call opcodes are used 18:48:15 i dont have a runtime prompt yet 18:48:18 OK, that's what I was getting at. :) 18:48:36 I was confused by that. 18:48:41 but i dont want to play with bytes anymore 18:48:51 so i try a new trick to align 18:49:23 90, 6690 666690 and ultimately 66666690, the bytes are in hex, lsb first 18:49:34 66 is on the x86 a size prefix 18:49:38 and 90 is the nop 18:49:45 the trick is froman engineer at amd 18:49:53 found it while googling 18:50:07 brb 18:50:11 * suprdupr now almost perceives google as startrek's "computer" invocation 18:53:23 i want to work with dword aligned primitives so that lookback optimization can be implemented without "list" (Last InSTruction) variable like in colorforth 18:53:40 Yeah 18:54:13 Problem is, though, some instructions require 5 to 7 bytes, depending on what prefixes are used and what operands you supply to it. 18:54:28 (actually, the longest instruction length for the x86 is 15 bytes!! But those are rare indeed...) 18:55:07 but primitives almost always use 8 bytes, and if im optimizing only the preceding instruction 18:55:11 it might work 18:55:19 * kc5tja nods 18:55:44 FS/Forth maintains its last instruction pointer in a traditional VARIABLE instead of in a register. 18:56:07 No use wasting a register on it when I/O is going to overwhelm compilation speed. 18:56:20 i resosrted to use EBP as a data stack pointer, although the adrssing use one byte more 18:56:34 main arg: ebp is in the stack segment 18:57:05 Well, unless you're working in a segmented environment, it really doesn't make any difference. 18:57:10 (especially on AMD processors) 18:57:37 * kc5tja uses ESI as the data stack, and I intend on using EBP as a pointer to the dictionary base. 18:57:50 Right now FS/Forth development is kind of stalled though. 18:58:09 I'm burning cash, and my job at In-N-Out burger, while helping to slow the burn, isn't covering it in its entirety. 18:58:39 i used this arangement also, but thought afterwards about Diesel forth and its register usage 18:58:47 yeah life is expensive 18:59:06 im glad i work only 4 hours a week, aside from school 18:59:16 I currently work 18 hours a week. 18:59:25 That's part of the reason why it doesn't make enough to cover living expenses. 18:59:33 minimum salary? 18:59:37 No 18:59:56 $8.25/hr to start, but as I advance through each training stage, I earn more and more. 19:00:10 In-n-Out burgers pays their people very well for the type of job it is. 19:01:41 * suprdupr nods, and hope he doesnt infringe on any trademark by writing this ;p 19:03:27 :) 19:03:35 * kc5tja is going to open an electronics kit business though. 19:03:43 great! 19:03:52 --- join: njd (~junk@njd.paradise.net.nz) joined #forth 19:03:52 One of the projects is a 6502-based single board computer. 19:04:01 That computer will run a dialect of FS/Forth ported to the 6502. 19:04:06 cool! 19:04:31 (out of the box; of course, if you equip it with your own EEPROM (yes, EEPROM) image, it will run anything you want) 19:04:37 Sonarman: It will run at 1MHz. 19:04:48 nostalgia :) 19:04:51 This is the entry-level version of a whole line of SBC kits. 19:05:22 hmmmm i read an article about molten plastics under an electric field that align their inner dipoles toit so when they are cooled down there's a static potential 19:05:30 i thought maybe homemade computers.... 19:05:55 * kc5tja is also making plans for building a simple headphone guitar amplifier too. 19:06:02 kc5tja: simple conditionals handling? i have trouble finding THE way to do it 19:06:15 More likely than not, the guitar amplifier will preceed the SBC, for no other reason than it's cheaper to build and market. 19:06:33 suprdupr: ?? 19:06:42 www.unquantum.com, refutes quantum models 19:06:52 * suprdupr is in brainstorm mode 19:07:08 the link is that 19:07:14 i work with young kids 19:07:20 8 to 12 years old 19:07:30 i tried to teach them optics the other day 19:07:36 i devised a lot of experiment 19:07:43 witn refraction/reflexion 19:07:49 and i thought some of em 19:07:51 like computers 19:07:58 but the learning curve is too high 19:08:10 After a while, I am planning on releasing a more advanced SBC, which I so-far call the SBX, which includes an 8MHz 65816 CPU, equipped with some expansion slots that you can plug cards into. Fully auto-config capable, yet still hacker friendly. 19:08:12 to let say learn assembler and do stuff right away 19:08:31 what would kick ass is a tvout support 19:08:36 simple to build 19:08:41 yet impressive for the masses 19:08:42 ;p 19:08:52 suprdupr: You'd think it's simple to build. It's anything but. 19:08:54 --- join: snowrichard (~richard@adsl-19-182-41.shv.bellsouth.net) joined #forth 19:09:08 re snowrichard 19:09:12 hello 19:09:12 kk the b/w theory is easy 19:09:26 and theres lot of pics/atmel projects out thre using tv out 19:09:38 there* 19:09:45 suprdupr: Yeah, monochrome is fairly easy to do, but you have to remember to shield it VERY well. You're dealing very much with RF frequencies, and unshielded, it can interfere with a lot of things (even itself). 19:10:01 suprdupr: Yeah, with a monsterous 28 to 32 dots per line. Useless. 19:11:01 * kc5tja has considered building an NTSC video card for the SBX machine though. 19:11:05 k then hmmm a piezzobuzzer expansion 19:11:09 --- quit: snowrichard (Client Quit) 19:11:12 i have lot of em lying around 19:11:13 Not for the 6502 SBC, but the 65816 SBC could definitely support it. 19:11:16 and i never use them 19:12:08 Well, the 6502 SBC is going to have (basically) 42 bits of digital I/O available for your exclusive control. 19:12:29 The UART is reserved for communications with the host PC at boot-time; it can be repurposed afterwards, I suppose. 19:12:55 But the 32 parallel I/O pins, 4 control pins, and 2 (non-UART) serial ports are entirely at your discretion to use for whatever purpose you want. 19:14:43 hmmm i still have some AVR chips, could make the children program it... the opcodes are clean 19:15:01 avrs are wonderful to learn assembly 19:15:09 i dont know the 6502 19:15:18 only that its in the NES 19:15:19 ;p 19:15:29 isn't it in the commodore 64, too? 19:15:42 And the Atari 8-bits, and the Commodore 8-bits, and the Apple II series (except the IIgs, which uses a 65816), ... 19:15:54 Basically, yes. 19:16:04 hmmm you could maybe use a NES as a SBC platform 19:16:06 lol 19:16:19 it got the tv output ;p 19:16:22 Most instructions execute in n cycles, where n is the number of bytes in the instruction plus the number of memory accesses it needs to make. 19:16:36 suprdupr: No, I can't, for two reasons: legal and technical. 19:16:43 Legal: It's not my design. 19:16:48 Technical: It's not produced anymore. 19:17:04 Also, NES uses a highly proprietary version of the 6502 -- ti's not 100% compatible. 19:17:15 kk 19:17:43 The 6502 can do video out via the VIA chips just like the PICs and AVRs can via its I/O ports too. 19:18:02 Maybe not with the same degree of resolution (especially with only a 1MHz clock!), but still... 19:18:39 The 6502 and 65816 do have overwhelmingly superior interrupt response capabilities to the PICs though. 19:18:43 Don't know about the AVRs. 19:19:05 avrs are almost deterministic 19:19:07 The '02 can service an interrupt in only a few clock cycles, AND it's indefinitely nestable. 19:19:09 if they awake from sleep mode 19:19:11 So ar PICs. 19:19:16 are even 19:19:31 4 cycles to get out of sleep then 4 to process interrupt 19:19:43 But the PIC chip takes many many tens of cycles to service an interrupt (last I checked), while a 6502 takes, maybe, 6? 19:20:04 I'm not talking about sleep mode here. 19:20:14 i now 19:20:23 AVR apparently takes 4 cycles to execute an IRQ, which is nice. That also beats the PICs. 19:20:31 (and beats the 6502 by 2 cycles) 19:20:34 But overall... 19:20:52 The 6502 (8-bit) at 4MHz can out-perform a (proper 16-bit) 8086 running at 5MHz. So.... :D 19:21:25 great didnt knew that 19:21:33 didnt know i think 19:21:41 anyway you know my grmmar is horrible 19:21:43 ;p 19:21:58 And spelling. :) 19:22:30 also the other day 19:22:36 The 65816 (the 16-bit version fo the 6502; yes, it still has an 8-bit external bus!) ran with 82% the speed of a 68000, both clocked at 8MHz. 19:22:42 i saw in a Zellers the Gamecube at 150$CAN 19:22:55 * kc5tja imagines a 65816-compatible processor with a proper 16-bit bus. 19:23:31 * suprdupr still has a netwinder sitting there doing nothing lol 19:24:48 01:59:01 printed with a befunge prog or something? I recall seeing a sierpinski-triangle-printer once. 19:24:48 02:09:59 That one is NYYRIKKI's brainfuck version. 19:24:48 02:12:04 ah, re befunge, just wrote an interpreter with forth. so our unofficial befunge-interpreters-in-obsolete-but-non-esoteric-languages project now has forth, fortran-77, algol-60, plus few less interesting ones. maybe should do cobol next. 19:24:48 02:12:47 I've just translated a much less well-known fractal though. Again please don't interrupt...(and sorry for monopolizing the channel...this should be the last...) 19:24:50 02:13:12 fizzie: go to #forth and say it's obsolete, i dare you 19:24:52 clog from 19:24:57 #esoteric lol 19:25:54 hehe 19:26:25 anyway, im sure calls are a luxury, unless one do a lot of isolated calls 19:27:18 * kc5tja is pretty sure they're not, but whatever. 19:27:23 * kc5tja will continue to use calls. 19:27:23 push(after) jmp(word), ok its 2 times bigger 19:27:37 im not only developping a compiler 19:27:52 im devising a new opcode design paradigm 19:28:08 a lot of the compielr job would be embodied in the chip itself 19:28:58 like call/ret are often implemented in hardware 19:29:36 "push address" means "push word", so it could only be an index into the Index array 19:29:48 instead of call i would put a Def opocde 19:30:10 Well, good luck. :) 19:30:11 align last word, terminate last one and define new word 19:31:14 or terminate, align, define, or simply terminate last and define new 19:33:16 The Steamer16 is a neat MISC CPU (only 8 instructions; not a one of them is a CALL or RET instruction). The fastest I've implemented subroutines on it was in 13 cycles. 19:34:24 yeah you reminded me of something 19:34:34 i wanted to use dwords instead of bytes 19:34:40 13 cycles to call, 10 cycles to return, total latency is 23 cycles. 19:34:52 This is one cycle faster than the 80386's 24 cycle latency. :-) 19:34:54 so by using a nop token, i could fetch 4 tokens at a time 19:35:23 but again literals are in the way 19:35:43 * suprdupr is looking at the steamer16 right now 19:36:28 In Steamer16 assembly language, here's what I came up with for a subroutine call (with no parameters): 19:36:35 rtnaddr sp @ ! 19:36:40 sp @ -1 + 19:36:47 sp ! target 0 19:36:48 zgo 19:37:11 where rtnaddr is the literal containing the return address, target contains the subroutine address, and sp is a variable containing the current stack pointer. 19:37:21 A return is this: 19:37:26 sp @ 1 + 19:37:31 dup sp ! @ 19:37:33 0 zgo 19:37:50 Oops, I forgot to include the cycles required for instruction fetches. 19:37:53 i understand the code 19:38:37 by using a clever compiler you could maybe simplify, like using constants fetch and store for return stack items 19:38:58 ? 19:39:03 "Constants fetch and store? 19:39:04 like 19:39:13 kk its badly formulated 19:39:18 instead of using sp 19:39:30 just pre-calculating litterals? 19:39:40 * suprdupr dont know if it would even be possible 19:39:48 Possible, but severely limiting. 19:40:01 It's useful for leaf words -- e.g., words that don't call any other words. 19:40:17 There is no need to advance or retard the SP in that case. 19:40:23 rtnaddr sp @ ! 19:40:25 target 0 zgo 19:40:26 ;) 19:40:32 sp @ @ 0 19:40:33 zgo 19:40:59 In this case, 9 cycles to invoke call (including instruction fetch overhead), and 7 cycles to return 19:41:02 Total: 13 cycles. 19:41:04 its been enlightning as ever, im sorry i must go 19:41:15 About the same as 6502's subroutine call overhead. 19:41:29