00:00:00 --- log: started forth/02.02.07 00:16:12 * aaronl is away: :P 01:06:57 maybe isforth seperate head space will work - not sure yet 01:07:24 im NOT defining a .text section and now my .data section can reference labels in my user defined .head section 01:07:30 gtg zzz 01:07:36 --- quit: I440r ("!") 03:57:52 --- join: aum (~david@l76-145.world-net.co.nz) joined #forth 04:10:12 --- quit: aum () 07:42:24 --- join: futhin (thin@h24-67-116-201.cg.shawcable.net) joined #forth 09:45:19 --- quit: futhin ("bbl") 10:58:07 --- join: futhin (thin@h24-67-116-201.cg.shawcable.net) joined #forth 11:06:30 --- join: futhin_ (thin@h24-67-116-201.cg.shawcable.net) joined #forth 11:06:37 --- quit: futhin_ (Remote closed the connection) 11:14:06 --- join: MrReach (~mrreach@209.181.43.190) joined #forth 11:14:29 hihi 11:14:55 hihi 11:15:20 how are you? 11:15:31 : hihi >r ! swap >r @ dup r> ; 11:16:15 heh, that'll crash 11:33:35 --- log: started forth/02.02.07 11:33:35 --- join: clog (nef@bespin.org) joined #forth 11:33:35 --- mode: carter.openprojects.net set mode: +n 11:33:35 --- names: list (clog) 11:34:31 --- log: started forth/02.02.07 11:34:31 --- join: clog (nef@bespin.org) joined #forth 11:34:31 --- mode: carter.openprojects.net set mode: +n 11:34:31 --- names: list (clog) 11:37:38 --- log: started forth/02.02.07 11:37:38 --- join: clog (nef@bespin.org) joined #forth 11:37:38 --- mode: carter.openprojects.net set mode: +n 11:37:38 --- names: list (clog) 11:45:00 --- log: started forth/02.02.07 11:45:00 --- join: clog (nef@bespin.org) joined #forth 11:45:00 --- topic: 'do drop >in, if you don't, you won't be able to join FNRC (Forth New Ruling Class) | -- Extra Extra - Read all about it - (;code) (;uses) and does> all now work in IsForth - shock horror!' 11:45:00 --- topic: set by I440r on [Tue Feb 05 07:35:57 2002] 11:45:00 --- names: list (clog oxygene futhin aaronl MrReach @ChanServ) 12:07:34 thanks 12:07:45 --- join: futhin_ (thin@h24-67-116-201.cg.shawcable.net) joined #forth 12:07:51 wb 12:08:20 how's it going? 12:09:01 ok 12:09:12 the whole net seems to be having a seizure right now 12:09:21 naw 12:09:25 i restarted my router 12:09:28 bought a router yesterday 12:09:29 yay 12:10:43 routers are very useful 12:11:01 especially when you start buying mid to high-end printers 12:11:18 which router did you get? 12:12:01 sohoware 12:12:13 i was considering netguard, but sohoware seemed to have more features 12:12:27 my router doesn't have a built in print server though .. 12:12:51 they seldom do ... and it defeats the purpose of a router 12:13:08 eh? there were a few routers that had built in print servers 12:13:09 many printers come with a 100BT connector as standard equipment 12:13:19 smc, us robotics, and maybe another.. 12:13:25 the purpose of a router is to move packets 12:13:42 to add functionality is to reduce reliability 12:14:58 i went with sohoware because it has stateful packet inspection which is supposed to be leet 12:23:41 --- quit: futhin (Read error: 110 (Connection timed out)) 12:23:46 --- nick: futhin_ -> futhin 13:26:18 --- join: I440r (~mark4@1Cust41.tnt2.bloomington.in.da.uu.net) joined #forth 13:26:18 --- mode: ChanServ set mode: +o I440r 13:26:57 --- topic: set to 'seperate headers now works - found a way to trick nasm - nasm is STILL braindead' by I440r 13:27:04 ls 13:27:06 who 13:27:08 ugh 13:27:19 bitchx hates me :P 14:12:22 --- join: aum` (~david@l76-134.world-net.co.nz) joined #forth 14:27:34 aum!!! 14:27:52 isforth is slightly less broke than it was yesterday heheh 14:28:36 hi 14:29:05 time to buy i440r some beer 14:29:19 i've gotten a bit lazy with netForth 14:29:21 get drunk! it'll help you finish isforth! :P 14:29:39 forth coders are lazy! that's why we have to use forth :P 14:29:45 something that's put me off is that a fibbonacci benchmark revealed it was 50 times slower than C 14:30:18 was it gforth or something? 14:30:23 i saw that benchmark i think 14:30:26 but gforth is slow 14:30:28 no - it is a FICL derivative 14:30:34 and was the code any good? 14:30:42 the code was optimal imo 14:30:44 the fibbonacci code might've been poor 14:30:48 hmm 14:30:53 benchmark it yourself or something 14:30:55 don't trust others :P 14:30:59 i did 14:31:07 what platform did you use? 14:31:07 i wrote the C and Forth code 14:31:08 linux? 14:31:19 make mine a jd 14:31:20 w--doze - does it matter? 14:31:38 which forth implementation? 14:31:43 which c compiler? 14:31:45 netForth - a FICL derivativ 14:31:49 gcc 14:32:07 so your netforth is slow, that doesn't mean other forths are ... 14:32:16 i never suggested they were 14:32:24 yeah ok 14:32:24 is netforth direct or indirect threaded 14:32:28 or is it written in c 14:32:35 is it subroutine threaded ? 14:32:36 indirect threaded, written in c 14:32:42 theres the problem 14:32:49 1 its indirect threaded 14:32:56 so to get teh xt you have to do @ @ execute 14:33:05 fetch the pointer to the pointer to the code 14:33:17 and teh fact that its written in c makes it automaticaly crippled 14:33:27 do yuor tests again in linux using isforth when its ready 14:33:50 how good is your syscall interface? 14:33:57 actually, the feb03 version of isforth could run your tests prolly 14:34:15 but they source would have to be typed in by hand 14:34:34 i dont realy have a GENERAL interface to syscalls 14:34:47 the only syscalls i use are key and an fcntl 14:34:52 int fib(int n) { return (n < 3) ? 1 : fib(n-1); } 14:35:43 : fib dup 3 < if drop 1 else 1 - dup recurse swap 1 - recurse + ; 14:36:13 i would use 1- not 1 - 14:36:29 iirc correctly, i was :) 14:36:52 which is fair because the c compiler probably optimises (n-1) to dec ebx or something heh 14:37:40 i cannot corelate the forth with the c 14:37:49 the c code only recurses in one location 14:38:14 oops 14:38:39 int fib(int n) { return (n<3) ? 1 : (fib(n-1) + fib(n-2)); } 14:38:46 right you are 14:39:17 and the forth is doing 1- in both places :P 14:39:24 that benchmark is handy for highlighting the overhead of function calls 14:39:43 show me a corrected forth 14:40:30 : fib dup 3 < if drop 1 else 1- dup recurse swap 1- recurse + then ; 14:41:30 why does the c code do n-1 in one location and n-2 in another and the forth does 1- in both 14:41:54 i thought it would be more optimal 14:42:38 but the 2 are different hehe 14:42:46 : fib dup 3 < if drop 1 else dup 1- recurse swap 2 - recurse + then ; 14:42:50 and i dont understand the diff heh 14:43:14 well, the C version runs 50 times faster than the forth version 14:43:32 turn optimizations off for the c 14:43:46 what's the gcc flag for that? 14:44:01 -o0 ? 14:44:04 -O0 ? 14:44:09 i fergit :P 14:44:25 not that i think this is a fair way of comapring the 2 versions 14:44:35 the NATURAL way teh c would be IS optimized 14:45:04 * aum` installs gforth on debian 14:45:24 ugh heh 14:46:52 ill compile it under isforth feb03 14:47:38 which will take a while because im copying civctp to a loop device over the network heh 14:47:51 so i can burn a playable cd 14:48:35 instacll civctp on my laptop and it uses 60% of the drive :P 14:49:42 cd .. 14:49:43 ls 14:50:39 --- quit: futhin () 14:52:33 --- join: futhin (thin@h24-67-116-201.cg.shawcable.net) joined #forth 14:54:21 ok - gforth is 10 times slower than gcc 14:54:43 what starting param did u use 14:54:49 about 13 times slower than optimised C 14:54:56 starting param was 39 14:55:06 gforth is slower because it's not optimized, poorly coded, coded ontop of c, etc 14:55:11 machine is K6-2/500, running debian potato 14:55:28 im 550 here 14:55:31 running woody 14:55:38 let me get this compiled 14:55:44 theres a fsckup in how im doing this heh 14:56:07 just so i'm understood - i'm not knocking forth as such, i'm in need of a forth compiler that has abundant support for high-level stuff - eg sockets, gui etc 14:56:48 an example of a forth app is the PSST encrypted instant messaging prog - http://netforth.sf.net/psst 14:57:26 netForth i mean 14:58:40 ok i got it compiled 14:58:46 what parameter do you want me to pass to it 14:59:07 im gona build a sockets library for isforth 14:59:11 SOURCES that is 14:59:34 what parameter did you pas to fib ? 14:59:39 it needs a starting parameter 14:59:42 5 fib 14:59:44 500 fib 14:59:45 what ? 15:00:01 everybody is waiting for i440r to finish isforth, then soon forth code will be actively developed! a massive programmer base will develop and engul the world!!! yes!!! 15:00:10 engulf 15:00:15 yea 15:00:17 exaktlee 15:00:19 39 fib 15:00:24 a lot of money is riding on you i440r :P 15:00:24 ok 15:00:40 just a sec 15:00:49 ive got problems entering numbers on the command line 15:00:53 a bit of LSD might facilitate easier coding :) 15:00:55 i can see multiple ways how isforth will be the champion for Forth 15:00:58 ill make a dummy : def to do 39 fib heh 15:01:01 brb 15:01:03 bbl 15:02:18 fuck i segfault heh 15:02:27 might be a stack issue 15:02:39 ill make 64k stack each heh 15:03:59 nope - still segfaults :( 15:04:05 how much stack will 39 fib use ? 15:04:11 and rstack 15:06:13 hehe nevermind 15:07:21 would you consider it cheating if i just branched to teh start of the definition instead of recursing ???? heheh 15:09:49 of course it's cheating 15:09:50 erm wait 15:09:51 heh 15:09:58 i know it would be hehe 15:10:01 im not cheating 15:10:05 im segfaulting still 15:10:08 for some reason 15:10:13 im checking it out now 15:10:14 try with smaller parm 15:11:24 no 15:11:28 how long did yours take ? 15:11:41 either mine is in infinite loop or its taking a while heh 15:11:52 im nit timing it yet, jhust makeing sur it works 15:12:08 not timing 15:12:15 how loong did the c version take 15:12:19 with 39 fib 15:14:13 ok it didnt enter an infinite loop 15:14:14 what is "fib"? 15:14:16 it finished the loop 15:14:20 File Input Buffer? 15:14:22 no 15:14:29 no - fibbonacci sequence 15:14:30 fibonacci benchmark 15:14:33 from aum 15:14:39 ah! 15:14:44 i hand coded it into a def in feb03 isforth 15:14:45 it thrashes the function calling 15:14:51 yea 15:14:55 yep, big time 15:14:59 how long did it take witha param of 39 15:15:10 I've seen it ... common benchmark 15:15:13 i want to compare isforth with that other forth :P 15:15:18 HOW LOING !!!!!!!!!!!!!!!!11 15:15:21 grrr :P 15:15:24 long evn heh 15:15:40 on netForth, 39 seconds 15:15:41 i wanna time isforth and win :P 15:15:45 woah 15:15:51 let me check 15:16:00 erm ... what are the system configs? 15:16:07 wait, i was using gforth - *that* took 39 secs on K6-2/500 15:16:08 us that user time? clock time? 15:16:13 user time 15:16:22 ok 15:16:35 I440r: what is your machine? smallish? 15:16:49 it might take a minute or two 15:17:13 and yours is bound to run somewhat slower than gforth 15:19:53 550 mhz 15:20:00 and it took 3 minutes 15:20:12 mine is a k6-2/550 15:20:23 is gforth a native code compiler ? 15:20:28 optimized forth ? 15:20:35 dunno 15:20:48 are you saying that isforth took 3 minutes? 15:20:54 yes 15:21:01 hers the definition i coded in 15:21:07 tell me if i got it right 15:21:11 user or clock time? 15:21:17 colon '(fib),pfib 15:21:25 dd dup, three, less 15:21:37 gforth is a forth written in C, specifically designed to utilize the GCC code optimiser 15:21:44 dd qbranch, .L1 15:21:56 dd drop, one 15:22:10 dd branch, .L2 \ could just make this an exit 15:22:14 .L1: 15:22:18 can you try it with gforth on your machine? 15:22:22 dup oneminus, pfib 15:22:26 and/or the C version? 15:22:33 swap twominus, pfib 15:22:40 plus 15:22:46 .L2: 15:22:48 exit 15:23:01 no - i dont have gforth on this laptop 15:23:09 arg! gforth isn't installed on my linux machine 15:23:18 insert dd's on those lines heh 15:23:23 is that definition correct ? 15:23:34 * MrReach laughs. 15:24:16 that 2- is wrong 15:24:26 when the if parts returns a 1 15:24:33 it should be returning a 2 15:24:42 drop 1 15:24:45 should be drop 2 15:24:50 if we do 2- 15:25:35 write it in fpc :) 15:25:46 this is a non optimized forth 15:26:00 actually, I440r is in a good position to write a pure assembly version of fib for comparison 15:26:33 ill do a coded def for it :) 15:26:35 within isforth 15:26:42 unfortunately, he's not so good at optimizing for Pentium or better @:^> 15:26:46 I440r: i thought isforth was meant to be kickass fast ?!?!? 15:26:50 not not good 15:26:52 dont bother 15:27:00 it is 15:27:05 for a non optimised forth 15:27:15 non optimizing 15:27:22 there are limitations on any generalized forth 15:27:57 the nest/unnest optimises differently on a Pentium and a Pentium III 15:28:07 they look very different 15:28:41 3 mins 12 seconds 15:28:50 also, Aton Erle discovered that on some processors, direct threaded is faster than indirect threaded ... and on other processors in the same line the opposite is true 15:29:04 what about x86? 15:29:12 i dont like optimized 15:29:16 direct is faster 15:29:22 nearly all of ertl's work is on x86 (linux) 15:29:25 i cannot see how indirect could be faster 15:29:37 @ @ execute 15:29:42 instead of @ execute 15:29:59 mov ebx,[ebx] 15:30:02 he found that the differences in the caching and lookadhead often favored one threading or another 15:30:03 jmp [ebx] 15:30:48 yes, it was a real eyebrow raiser to me, also 15:35:00 so, there's no way that I440r can produce THE FASTEST threaded forth w/o a bazzilion of conditional assemlies in the kernel 15:35:27 but isforth should be relatively fast, and it's hella simple, which can be of more value 15:36:30 ( this doesn't even consider optimized native code forths like Swiftforth and VFX ... which should be roughly 10 faster than isForth) 15:36:38 10x 15:40:17 my asm verwsion isnt timeable 15:40:21 it returns immediatly 15:40:31 with the correct asnwer? 15:40:49 anser ? 15:40:52 theres an answer ? 15:40:54 heh 15:41:00 code '(fib2)',pfib2 15:41:00 cmp ebx, 3 15:41:00 jge .L1 15:41:00 mov ebx,2 15:41:00 ret 15:41:01 .L1: 15:41:03 push ebx 15:41:05 dec ebx 15:41:07 call pfib2 15:41:09 xchg ebx,[esp] 15:41:11 sub ebx, 2 15:41:13 call pfib2 15:41:15 add ebx,[esp] 15:41:17 add esp,4 15:41:19 ret 15:43:09 erm 15:43:09 no 15:43:11 i lied 15:43:15 its not returning immediatly 15:43:22 it segfaulteed heh 15:43:28 haha! 15:44:40 which might be it finished and my hex output of the result fubared 15:45:01 is the above code correct ? 15:45:16 yea i think it segfaults when its done 15:47:06 7 emit 15:49:51 teh : definition returns 0x78a1d44 as a result 15:49:57 is that the correct result 15:56:26 i fout out why the asm was crashing heh 15:56:42 why? 15:56:50 i need (FIB2) 15:57:00 AND IT (oopts caps) has to be able to CALL itself 15:57:08 so... the last thing in it is a ret 15:57:10 so 15:57:15 wheres the next man heh 15:57:22 had to make a _pfib2 15:57:25 call pfib2 15:57:26 next 15:57:27 heh 15:57:42 asm returns same result as forth 15:57:45 ill time it now 15:58:50 about 47 seconds 15:58:51 wtf 15:58:54 i gtg store 15:58:55 brb 16:04:37 --- join: ult (~saturn@149.149.200.56) joined #forth 16:54:00 --- quit: aum` () 17:12:12 so what result should that fib function return :P 17:12:47 no clue 17:13:18 thats because ur too lame to calculate the result in your head :P 17:13:29 fine! 17:13:31 the answer is 42! 17:13:54 no thats he answer to life the universe and everything 17:14:25 hmm 17:14:29 fine, the answer is 2 17:14:48 close - but like i said - ur way too lame to get it right :) 17:14:49 hehe 17:14:53 :P 17:15:03 well i haven't been paying attention 17:15:11 maybe if you give me the math i'll calculate it in my head 17:15:21 heh 17:16:06 quick, gimme math ! 17:16:09 :) 17:17:05 r = 1/r + 1 17:17:12 what is the real value for r 17:18:34 --- quit: ult (Read error: 110 (Connection timed out)) 17:19:25 u scared him off! 17:21:30 In[1]:= Solve[r == 1/r +1] 17:21:32 1 - Sqrt[5] 1 + Sqrt[5] 17:21:32 Out[1]= {{r -> -----------}, {r -> -----------}} 17:21:32 2 2 17:21:38 golden ratio!! 17:22:04 r is 1.61803499 17:22:05 ish 17:22:11 wow 17:22:18 and i was thinking 1.3 17:22:19 damn 17:22:34 :P 17:22:38 just do 17:22:40 if r = 2 17:22:47 1/2+1 s not 2 17:22:54 so what is 1/2+1 17:22:56 yeah whatever, i wasn't thinking, chatting to other ppl y'know :P 17:22:56 1.5 17:22:57 so 17:23:01 take r = 1.5 17:23:04 shh! :P 17:23:09 i can do math :P 17:23:13 1/1.5+1 is not.... 17:23:19 heh 17:23:34 we are pentium of borg >??? heh 17:33:15 --- quit: I440r ("BitchX-1.0c18 -- just do it.") 17:56:57 --- join: I440r (~mark4@1Cust41.tnt2.bloomington.in.da.uu.net) joined #forth 17:56:58 ... it is easy to be blinded to the essential uselessness of them by the 17:56:59 sense of achievement you get from getting them to work at all. In other 17:57:00 words... their fundamental design flaws are completely hidden by their 17:57:02 superficial design flaws. 17:57:04 -- The Hitchhiker's Guide to the Galaxy, on the products 17:57:06 of the Sirius Cybernetics Corporation. 17:57:08 :) 17:57:11 ah! you're back 17:57:18 i think he was talking about isforth :) 17:57:37 heh, not very confident in your work? 17:57:41 heh 17:57:48 the "bastard creation" syndrom? 17:57:51 no i just LIKE its flaws :) 17:58:14 i wonder if thas where the bastard disassembler gets its name from :) 17:58:23 how close are you to getting INTERPET to work? 17:58:31 or maybe its because its such a bastard to get to compile :P 17:58:39 +R 17:58:42 interpret works. but it works WRONG 17:58:59 i need to rewrite query, expect, interpret, quit etc 17:59:13 compilation will be done INSIDE interpret 17:59:18 once you get to a point where you can type in a line and it produces a result, things will get 10x easier for you. 17:59:27 they do that now 17:59:42 but word for instance doesnt place a copy of the found string at here 17:59:46 its going to 17:59:50 I usre wish you had chosen to metacompile 17:59:57 except its going to be at hhere (header space) 18:00:00 ah! ok 18:00:04 i have chosen to 18:00:13 but metacompilation is an extention to forth 18:00:24 i WANT a meta compiler on TOP of forth 18:00:30 have you found anyone to write a P4 assembler yet? 18:00:30 not forth on top of a metacompiler hehhe 18:00:38 no :( 18:00:48 without that metacmpilation is a dream 18:00:50 what level of assembler do you want? 18:00:56