00:00:00 --- log: started forth/02.04.22 01:05:47 --- quit: Stepan ("Client Exiting") 01:26:04 --- quit: onetom (Read error: 110 (Connection timed out)) 04:11:52 --- join: Fare (fare@samaris.tunes.org) joined #forth 05:19:45 --- join: rob_ert (~robert@h237n2fls31o965.telia.com) joined #forth 09:03:27 --- join: herkamire (~jason@ip68-9-58-81.ri.ri.cox.net) joined #forth 09:24:55 --- join: I440r (~mark4@1Cust87.tnt1.bloomington.in.da.uu.net) joined #forth 09:25:04 hi 09:25:14 im trying to add vocabularies to isforth in nasm 09:25:29 its goign to make the macros file a gordian knot tho i think :P 09:26:02 but not having vocabs is starting to become a drag and the 'knot' will untie itself once we get a meta compiler 09:39:13 how many different asm instructions do you use? 09:41:16 robert did a test of that i think (i think it was him) and he said im using about 380 diff instructions ? 09:41:36 crap 09:41:43 What Robert? 09:41:49 you wasnt it ? 09:41:53 or was it onetom 09:41:57 Hehe 09:41:58 i think it was tom heh 09:42:02 I'm innocent! :) 09:42:05 but ill blame you 09:42:12 OK, good. 09:42:17 * rob_ert is the punching bag. 09:42:50 I think PPC only has 190 instructions total (not including the vector unit) 09:44:47 actually of those 380 different instructions SOME are the same instruction but used differently 09:45:31 yeah, I wasn't counting those 09:45:49 ppc also has a simpler syntax at least most of the time. (I think...) 09:47:04 ppc is better 09:47:32 motorola makes far better microprocessors than intel but intel makes FAR FAR superior microcontrollers to motorola 09:52:27 for PPC there's about 34 "fake" instructions (assembly instructions you type in like "li" that actually compile to a different instruction (in this case addi)) 09:54:17 eh ? 09:54:26 sounds silly hehe 09:54:44 i like the 68k nop instruction - its not realy a nop, id DOES something heh 09:55:22 what sounds silly? 09:56:20 having multiple names for the same instruction - unless its a 'context' thing 10:00:58 it's not silly 10:01:19 do you have a "load immediate" instruction? 10:03:59 mov eax,5 ? 10:05:49 does that put 5 in a register for you? 10:05:56 yes 10:06:02 --- join: davidw (~davidw@ppp-128-17.25-151.libero.it) joined #forth 10:06:19 does mov do anything besides put a constant in a register? 10:06:50 no flags are affected by it 10:08:23 PPC asm has something like that: li r3, 5 (puts the value 5 in register 3) 10:08:53 does it assemble the same opcode as load r5, 5 ? 10:08:54 makes sense. but it does not make any sense for the precessor to have an extra instruction for this when it already has one that can do the job (addi) 10:09:57 same opcode as addi 10:10:14 addi destination-register, source-register, constant-value 10:10:42 addi r3, r0, 5 is the same as: li r3, 5 10:11:15 er addi sounds like an addition not a move hehe 10:11:23 (when math is involved r0 usually tells tho processor to use zero instead of the value of the first register) 10:11:56 yes. addi adds the seccond and third arguments and stores the result in the first 10:12:23 PPC assembly put's the result in the first argument gennerally, it's confusing for a while if you're used to it the other way around. 10:12:50 i stands for "immediate" which means a constant value like 5 10:16:08 what does mov do? 10:16:25 never mind 10:16:33 --- join: tathi (~tathi@ip68-9-58-81.ri.ri.cox.net) joined #forth 10:16:37 heh 10:16:43 mov is like load 10:16:50 everything in x86 is a mov :P 10:17:17 pain in the ass when you're trying to write an assembler :) 10:17:58 tell me about it :P 10:18:00 though actually it probably isn't any worse than trying to assemble any other mnemonic... 10:18:10 on second thoughs DONT tell me about it 10:18:18 * I440r goes and cries in the corner 10:18:29 * tathi comforts I440r 10:18:36 you trying to write one? 10:18:43 yes 10:18:59 actually rite now im obfuscating my nasm macros file 10:19:05 adding vocabularies if i can 10:19:49 ah 10:21:44 meanwhile im rtfm'ing every single assembler source i can lay my hands on 10:21:57 actually a lot of the stuff is pretty similar 10:22:44 I think if you can write code to handle operands and generate the SIB and R MOD M bytes or whatever they're called, 10:23:01 then the rest should mostly just be tedious 10:23:05 mod r/m 10:23:22 its knowing when to generate sib 10:23:26 when to generate a mod 10:23:30 when to generate a r/m 10:24:04 is there any consistency to that, or is it per mnemonic? 10:24:16 no 10:24:28 this is x86 were talking about 10:24:37 you want consistency go for 8051 :P 10:24:42 :) 10:30:30 --- join: MrGone (~mrreach@209.181.43.190) joined #forth 10:30:35 or 6502 !!! 10:30:39 heh 10:30:46 --- nick: MrGone -> MrReach 10:31:02 hihi! 10:31:20 thats consistent :) 10:31:20 hello mrreach 10:31:31 hey mrreach 10:31:39 im trying to add vocabs to isforth and i noticed somethin weird in fpc :) 10:31:44 hang on let me get to it 10:31:54 oh? why doesn't that surprise me? 10:32:12 lol 10:32:50 btw, Ericson's assembler link vocabs just fine ... erm ... was it A86? 10:33:00 but that's 16 bit 10:33:13 eric isaacson 10:33:29 a86 is 16 bit 10:33:34 a386 is 32 bit 10:33:42 i regged so i have a386 :P 10:34:23 will it produce ELF output? 10:34:46 it can do binary - so you can create the elf shit yourself hehe 10:34:55 ah! figures I'd FUBAR the guy's name 10:34:57 i already thunked of that 10:35:22 ok. 10:35:36 : definitions context @ current ! ; 10:35:39 we all understand this 10:35:49 but why does fpc's create do a 10:35:54 context @ current ! 10:36:15 any time you define a word in a vocab it makes sure that vocab is in the search order 10:36:19 thats silly 10:36:38 er i did it backwards 10:36:45 create does current @ context ! 10:36:47 grr heh 10:37:11 : (:) current @ context ! blah blah blah ; 10:37:21 im also confused about HIDE 10:37:23 --- join: bugslayer (~thin@h24-64-174-2.cg.shawcable.net) joined #forth 10:37:28 i understand reveal - but hid confuses me 10:37:34 its already hidden 10:37:38 for what? 10:37:43 its not added to the vocabularies hash chain yet 10:37:49 reveal and hid for forth ? 10:38:02 er, should be "hide" i'd think.. 10:38:03 : new-word blah blah [ this word is still hidden ] blah blah ; 10:38:22 semi did a reveal and added it to some chain 10:38:34 hide is what i said 10:38:57 ok, lemme reread/ponder for a moment 10:39:04 k 10:39:48 : hide last @ dup n>link y@ swap current @ yhash ! ; 10:40:06 ok, which do you want me to address first? 10:40:07 : reveal last @ dup n>linik swap current @ yhash ! ; 10:40:11 hide 10:40:22 wtf is it doing 10:40:28 --- quit: herkamire ("Client Exiting") 10:40:34 (:) also calls hide on the word being defined 10:40:37 ok, hide is used to clean up a vocab after a package/module is loaded 10:40:38 but its already hidden 10:40:44 arg! 10:40:48 not in this context its not 10:41:04 : (:) 10:41:08 !csp 10:41:12 current @ context ! 10:41:14 hide and reveal look like they could be factored out into an additional word or two.. because y@ is the only difference 10:41:19 I see that it is the word that actually makes the entry in the hash table 10:41:32 header ,jmp 10:41:39 xhere paragraph + 10:41:44 dup xdpseg ! 10:41:49 xset @ - , 10:41:53 xdp off 10:42:01 HIDE ;uses nest ,-x 10:42:12 ugh the definition has no ; on it either 10:42:31 thats the definition for (:) 10:42:36 ignore it ... write your forth more sensibly 10:42:38 it calls hide on the 'last' word 10:42:42 i will heh 10:42:43 but 10:43:00 yes, it is odd 10:43:00 : some-word .... somewhere in here we HIDE this word blah blah ; 10:43:09 but its already hidden 10:43:21 and hide only works on LAST 10:43:30 so it HAS to be used ONLY during compilation 10:43:35 you cant do 10:43:36 : foo blah blah ; 10:43:38 hide 10:43:42 HIDE (along with REVEAL) contains the code that actually links into the hash table 10:44:02 which is really wierd, you're right there 10:44:05 i dont think isforthy is going to have hide 10:44:29 well, HIDE is fine, but it should have behavior sensible to its name 10:44:32 last is a place holder for the current definition so that ; can link it into what aver chain 10:44:44 little in fpc does :P 10:45:21 also, I don't really understand resetting CURRENT in colon, either 10:46:10 well, IMO FPC is an excellent forth, but it suffers from its own history in some cases 10:46:20 who would take the vocabulary being built OUT of the search order ? 10:46:20 lol 10:46:42 yes 10:46:56 isforth is very heavilly based on fpc heh 10:47:59 I would rather that you thought through the stuff on your own, and use FPC (or any forth) only as an example of how some task _might_ be accomplished 10:48:49 I'm gonna get upset if I see X, in your sources @:^> 10:49:01 http://cagle.slate.msn.com/news/StayskalGUNS 10:49:13 not a chance hehe 10:49:25 ir y@ 10:49:30 nor 10:49:42 you're bting the bullet on the assembler? 10:49:56 biting 10:50:05 i am doing things ALOT different 10:50:11 not 100% sure on that yet :) 10:50:24 im still very very very tied in knots over it 10:50:43 but if i can find a sword sharp enough ill cut the knot and write the assembler 10:50:43 yes, x86 is such a joy to write for 10:50:58 that's why optimisers work better than humans on x86 10:51:09 which is a sad state of affairs 10:51:41 :P 10:51:50 well you know my opinion of optimizing :P 10:53:26 yep 10:53:41 lol 10:53:50 what is your opinion of optimizing? 10:53:51 some day i might make isforth do some sort of optimizing 10:54:00 its a total waste of effort 10:54:12 really 10:54:16 that it bastardizes perfectly good code 10:54:26 optimize your code today and tomorrow intel tweaks this and your code is slower that it would have been if you hadnt optimized 10:54:29 removes clarity 10:54:50 and then some :) 10:54:57 ok 10:54:58 i Do optimize my code 10:55:01 but not seriously 10:55:06 its like i posted on clf 10:55:28 a badly written radix counting sourt is going to out perform a heavilly optimized quick sort every time 10:55:36 :) 10:56:06 oh, I know ... I've seen Hendrix and Ertl both dig into clocks, pipelining, cache borders, and benchmarking so deeply that I got sick of it 10:56:25 hendrix is a master at it :) 10:56:41 but as a result, they write some of the fastest forths in existance 10:57:07 I haven't wanted to do something where speed is much of an issue so far 10:57:08 yes but i would rater figure out how FPC works :P 10:57:16 lol 10:57:36 it's really ironic, because the machines that NEEDED optimising were the old i386SX16MHz machines, and they didn't have a enough memory for full-fledged optimising 10:57:42 yeah 10:58:04 the ex too 10:58:10 i386Ex 10:58:21 now that we've got P4s, the optimisers are inhumanly effecient 10:58:27 DX? 10:58:31 EX 10:58:35 its an embedded 386 10:58:35 never ehard of it 10:58:38 oh! 10:59:06 is that the one they're selling now with the keyboard chip in it and the whole bit? 10:59:40 i dont think it has a keyboard chip in it 10:59:43 unless they upgraded it heh 10:59:55 now THERE is a place where FPC might have really shown in the compiler marketplace 11:00:19 yes 11:00:27 by using a multi-pass optimiser and overlays, it might have optimised better than anything on the market 11:01:08 the only two really competing were Borland and Microsoft 11:01:29 TCOM did pretty darn good 11:01:38 and it was *SO* primitve 11:01:44 hell yes 11:01:52 i want something like tcom for isforth ! 11:01:59 it was very nice! 11:02:07 well, ummm ... you'd have to subroutine thread 11:02:44 TCOM? 11:03:08 long ago, TCOM was a "compiler" for forth 11:03:13 in the true sense of the word 11:03:18 ah 11:03:30 start it with a source file, and you get a .COM as output 11:03:52 it takes your sources and builds a kernel comtaing only those primatives required at run time by your target 11:04:03 and extends it with your sources :) 11:04:07 but it used forth architectures in the runtime, and the result turned out to be incredibly fast 11:04:55 it mostly did subroutine threading, with code-copying and a few simple code replacements 11:05:51 nice 11:06:25 it was. tho i never realy used it :) 11:06:52 i still cant figure out why tz went DOZE on us. why didnt he go linux like the rest of the sane world ? 11:07:10 his clients were using windows machines 11:07:23 Linux at the time didn't have the services that it has now 11:07:50 he has CLIENTS ? 11:08:15 yes, that's how he makes money, silly 11:08:36 both FPC and Win32Forth are production Forths 11:09:09 and it shows, he spent a minimal amount of time writing a kernel, then just ported his existing tools, patching as he needed 11:09:20 i thunked he worked for someone 11:09:47 fpc runs fast ?? 11:09:57 fairly fast, yes 11:09:59 fpc is optimised? 11:10:12 it's indirect threaded for 16bit DOS 11:10:43 no. its direct threaded 11:11:00 the major improvement for speed in FPC is direct screen writes to video memory 11:11:04 oh, ok 11:11:41 yes and i doubled its speed 11:11:45 his code SUCKED there heh 11:12:19 i440r: and you won't share your modified fpc to anybody? :P 11:12:31 its currently b0rked hehe 11:12:39 i made a mod then discoverd linux 11:12:44 hard to find a machine to run it on 11:12:44 didnt finish the mod :P 11:12:53 it runs in a dos box :P 11:12:55 quite nicely 11:13:03 could use DOSEMU on linux, but then you miss out of the really good stuff 11:13:09 linux has a dosemu that actually runs command.com.. 11:13:28 like networking and pipes 11:13:48 yea but the dosemu sets up a bullshit environment 11:13:58 last i knew it couldnt access any linux partitions 11:14:04 just a virtual dos partition 11:14:07 bleh 11:14:09 heh 11:14:35 THAT must be I440r's dream ... FPC on linux 11:14:43 even no 11:14:45 not realy 11:14:50 i want isforth to be readable :P 11:15:04 readable to who? 11:15:06 asm coders? 11:15:12 or non-asm forth coders? 11:15:22 asm coders and forth coders 11:15:29 i.e. everyone in here but you :P 11:15:34 lies 11:15:49 i know basic asm too 11:15:58 i coded this really elite asm code 11:16:03 lol 11:16:07 leet: 11:16:08 nop 11:16:09 ret 11:16:31 that went thru a specific area of memory, and converted the lowercase to uppercase and vice versa 11:16:49 the loop that did the work was only 3 instructions.. it was l33t 11:16:52 used x0r ! ;) 11:17:17 yeah, i coded hello world a bunch of times too 11:17:21 coded it in debug.com too ;) 11:17:27 bugslayer given registers ax and bx swap their contents without using memory or another register :P 11:17:30 no pushes or pops 11:17:34 no stores to memory 11:17:41 no xfer to a different register 11:17:46 swap contents of ax and bx 11:18:02 hmm.. lemme think :P 11:18:05 mrreach should know this :) 11:18:12 if he dont i would be very supprised :) 11:18:36 brb - making coffee 11:18:44 if you dont have it by the time i get back ill show you 11:18:46 is there any instruction that swaps them or swaps half of ax and half of bx ? 11:18:52 erm yes \ 11:19:01 but thtas another one i forgot to mention 11:19:03 no xchg :) 11:19:30 no using any memory, not even on chip memory? 11:19:56 :) 11:20:22 u know how tathi? :P 11:20:26 don't tell me tho 11:20:45 just say "yes i know this" or "no i dont" or "this is total bs" :P 11:21:19 ok.. 11:21:21 i'm thinking 11:21:37 add one register to another in some sort of trick :P 11:21:43 hmm 11:22:04 mov ax, bx 11:22:07 neg ax 11:22:12 mv bx, ax 11:22:15 no 11:22:18 neg bx 11:22:19 heh 11:22:20 sorry, was int he other room, reading backscroll 11:22:25 mov ax, bx overwrites bx hehe 11:22:30 you just lost its contents :P 11:22:35 wtf 11:22:36 lies 11:22:51 heh 11:22:51 bugslayer: yeah 11:23:05 tathi its realy simple once you know it eh ?? :) 11:23:08 oops 11:23:10 yup yup 11:23:11 add ax, bx 11:23:16 nope 11:23:22 getting closer tho (not) 11:23:23 heh 11:23:43 what's the instruction for making the number in one register negative? 11:23:48 erm 11:24:07 not does a 2's complement 11:24:18 xor reg, $ffff does a 1's complement 11:24:20 took me a while to figure it out the first time though :) 11:24:37 BS as far as I'm concerned 11:24:43 it is :) 11:24:58 I'd XCHG ax,bx and move on to more important things 11:25:07 lol 11:25:10 wait a min, let me figure it out fools! :P 11:25:21 I'd think it would be useful on some little processors though 11:25:22 mrreach what if you dont have an xchg and cant use stack or memory for some reason 11:25:41 xchg ax, bx 11:25:46 is fine if you got it :) 11:25:49 xor ax, bx 11:25:51 xor bx, ax 11:25:54 xor ax, bx 11:25:54 *THEN* I would sit down and think about it, not before ... and whine about totally lame processors 11:26:07 :) 11:26:12 screw u 11:26:15 PPC doesn't have xchg 11:26:17 why did you give it ? 11:26:30 your time was up :P 11:26:41 always plenty of extra registers though :) 11:26:42 add ax, bx // add bx to ax (assume 5,2) --> (7,2) 11:26:42 sub bx, ax // sub ax from bx (7,-5) 11:26:42 add ax, bx // add bx to ax (2,-5) 11:26:42 neg bx // (2,5) 11:27:21 don't tell me that doesn't work (even if i broke some asm rules)! :P 11:28:26 try that with -5 and -2 and tell me if it works :) 11:28:45 then try 2 and -5 11:29:16 heh 11:29:38 it should :P 11:30:03 neg bx should be something like "reverse the negative or positive on bx" 11:30:14 but whatever 11:30:18 time to eat breakfast 11:31:48 time to go try make vocabs work in isforth 11:31:50 bbl 11:31:53 --- quit: I440r ("bbl") 11:32:31 how fast is xoring 3 times compared to xchng or using a 3rd register? 11:33:08 xoring is slower 11:35:45 I'll be leaving for "work" in a few moments 11:36:16 heh, have fun 11:36:27 yay for work 11:36:41 yah, hurrah 11:36:56 but don't start neglecting quality time with your family :P 11:37:27 ha! 11:37:39 my wife will be glad to get me out of the house 11:37:54 awww, why? :) 11:38:28 hm 11:38:37 i like to be alone whenever i have a ton of friends 11:38:40 around 11:38:56 or when i'm alone, i like to have a ton of friends around 11:39:01 paradox 11:41:47 hm 11:41:50 --- quit: bugslayer ("bye") 11:44:18 --- nick: MrReach -> MrGone 11:44:41 Bye MrGone. 12:01:23 --- join: bugslayer (~thin@h24-64-174-2.cg.shawcable.net) joined #forth 12:06:17 --- nick: bugslayer -> futhin 12:06:41 --- nick: futhin -> bugslayer 12:22:02 --- join: davidw_ (~davidw@ppp-128-17.25-151.libero.it) joined #forth 12:23:33 --- quit: davidw (Read error: 104 (Connection reset by peer)) 14:05:32 --- join: onetom (tom@adsl52032.vnet.hu) joined #forth 14:06:19 Hi 14:06:35 heya onetom 14:06:45 hi guys 14:12:33 --- nick: davidw_ -> davidw 14:14:21 b back in ~1hr 14:37:58 --- join: Speuler (~l@195.30.184.4) joined #forth 14:38:04 g'day 14:43:17 :D 14:52:49 me back 14:54:49 :) 14:54:50 wb 15:04:18 some papers 2 read 4 the idle 1s amongst us: http://www.jfar.org/volume7.html 15:04:56 personally, i will read the state machines article, think 15:09:43 read the stack machine book written by loopman 15:11:40 I$$)r: have you found your assembler? 15:11:59 he's not here.. 15:13:07