00:00:00 --- log: started forth/02.03.09 00:20:31 --- join: Soap` (~flop@210-55-149-150.dialup.xtra.co.nz) joined #forth 00:29:12 hi 03:20:00 --- join: rob_ert (~robert@h173n2fls33o898.telia.com) joined #forth 03:55:02 --- quit: Soap` ("sleep") 06:13:20 --- join: Etaoin (~david@ljk2-11.sat.net) joined #forth 06:14:14 armed with a few hours of sleep, I return to do battle with forth 06:30:03 --- join: aum (~david@210.55.76.154) joined #forth 06:30:13 hi 06:31:25 hi there 07:09:25 hi 07:09:46 aum, do u have some time 2 discuss a problem? 07:09:51 a theroetical 1 07:10:15 Etaoin: hi 07:10:34 how does yr vactor stuff advances? 07:10:49 imean vector :) 07:11:26 going okay 07:11:32 i wanted 2 mention yesterday 07:11:38 I just woke up a few minutes ago 07:11:42 onetom: i have a few minutes ... go ahead 07:11:47 that u shouldnt juggle w object pointers 07:11:56 Etaoin: hold on 07:12:19 onetom: why not? 07:12:20 aum: sprintf("[%02d-%02d-%02d %02d-%02d] asd: %s (%4x, '%c')" 07:12:40 aum: {{ year month day hour min error param char }} formatstring dest sprintf 07:13:04 aum: id like 2 make sg like that 07:13:42 aum: do u know about any implementation of such an sprintf word? 07:13:58 no sorry 07:14:07 Etaoin: because its a bit hard 2 track the code that way 07:14:18 aum: ok, 1 more question 07:14:19 i've implemented my own printf/sprintf/fprintf words 07:14:36 oh.. in netforth? 07:14:42 y 07:15:04 u mean *printf LIKE words 07:15:11 yes 07:15:14 or exactly them 07:15:27 in forth, i gotta put an arg count after them 07:15:46 and do your words use formatstring then? 07:16:00 eg $" %d-%d-%d" year month day 3 printf 07:16:06 yes - formatstring 07:16:13 ah, thats great 07:16:30 : {{ sp @ depth ! ; 07:16:45 : }} sp @ depth @ - ; 07:17:11 i use extensively this construct in my 07:17:20 building controll sw 07:18:10 tho, its not nestable, still very convenient & useful 07:18:26 what's sp again? 07:18:32 aum: ok, thx. i will look them up in netforth 07:18:39 stack pointer 07:18:59 well, i can't afford to allow a word to monopolise the stack 07:19:04 oh, sorry 07:19:09 np 07:19:13 i mean 07:19:30 : {{ depth {depth} ! ; 07:19:44 : }} depth {depth} @ - ; 07:19:48 variable {depth} 07:20:29 & u r right, its a really unforthish solution using the stack to pass massive amount of data 07:21:16 i start forth as "gforth -d 30000", which is well... khm.. :) 07:21:38 but thats an other story 07:21:38 well, forth is a language that totally evades standardisation 07:21:43 thx 4 your help 07:21:47 np 07:21:57 there's an idea if you're interested 07:22:16 if you feel like hacking, you could create words for 'stack frames' 07:22:35 hm... stack frames? 07:22:47 never heard of this term b4 07:22:49 yes - stacks on top of stacks 07:23:06 the term 'stack frame' comes from C and other compiled languages 07:23:07 sounds exciting :) 07:23:24 ah, yes! now i know 07:23:43 one way you could do it is... 07:23:47 some locals implementaion uses stack frames 07:23:52 push 'markers' onto the stack 07:24:21 and have another 'depth' word which simply tells you the number of values pushed since the last frame marker 07:24:31 aha, c 07:24:54 could u advice some url, where i can read more about it? 07:24:55 so create a word 'mark', which pushes a marker onto the stack 07:25:13 onetom: best way to learn is to compile C progs and look at the assembler 07:25:29 here's a quick implementation: 07:25:41 i did it - tho w pascal prg - a lot in the dos era 07:25:58 sp variable pmark 07:27:16 : mark pmark @ sp - dup pmark ! ; 07:27:33 do you know if stack goes up or down in forth? 07:27:43 sure 07:28:01 last night i put together a basic 4th processor 07:28:08 written in pure 4th 07:29:07 http://213.163.52.70/~tom/kernel.fs 07:29:31 i made it 4 educational purposes 07:29:45 and to help me to decide a debate 07:30:11 futhin states there is no need of any support from 07:30:33 the side of the 4th kernel to realize preemtive multitasking 07:31:10 depends on how well your code allows for multitasking 07:31:15 oh, ive just noticed, that was a question :) 07:31:17 whether it hogs processor 07:31:41 usually stacks grow down 07:31:52 right... 07:32:24 do u know what animal grows down? 07:32:29 eg. a duck ;) 07:32:41 : mark pmark @ sp pmark ! ; 07:33:46 i don't know how to write to sp 07:33:55 so you'll have to write your own unmark word 07:33:59 anyway... 07:34:33 : mdepth pmark @ sp - cells ; 07:34:34 writing sp is implementation dependant i guess 07:34:56 i dont know any recomendations about it 07:35:20 so the idea of stack frames is to allow you to deal with the stack in terms of discrete lumps, each lump with an arbitrary number of values 07:36:02 so stack-hogging words like {{ and }} can be changed to deal in stack frames instead - they won't have to disturb values which aren't their business 07:37:06 hmm.. its not quite clear what r u tryin 2 mention... 07:37:29 but i'll analyze it deeper 07:39:57 : mdepth pmark @ sp - cells / ; 07:40:42 here's an analogy for explaining... 07:41:02 instead of a huge stack of white sheets of paper, 07:41:25 insert blue sheets in at certain points to divide up the stack into meaningful sections 07:41:46 or another option - run multiple virtual machines 07:42:05 r these blue sheets make up a linked list? 07:42:18 yes 07:42:21 imean, in your previous implementation? 07:42:23 hooray. I have code that works again 07:42:41 Etaoin: update the url then! 07:42:54 let me delete all of my old defintions 07:43:03 Etaoin: imean, the file pointed by the url 07:44:03 http://braznet.com/david/files/mini-oof.fs 07:44:05 it's modified 07:44:11 http://braznet.com/david/files/oovect.fs 07:44:30 aum: aaaaahaaa! managed 2 understand that short example finally. im alwasy a bit slow after meal :))) 07:46:24 Etaoin: ah, u made some dynamic obj creation extension... 07:47:05 I got tired of :noname blah blah ; classname defines methodname 07:47:19 the name was always at the end 07:48:46 Etaoin: ive already was about 2 ask that why had u swapped the method and attrib defs? 07:49:33 onetom: I don't understand. 07:49:41 was that a question? 07:49:42 yeah, that :noname syntax also made me nervous :) 07:50:08 --- quit: aum () 07:50:08 :defmethod uses :noname, though 07:50:18 u have swapped the order of vars & methods. why? 07:50:34 object class 07:50:37 cell var vect-len 07:50:38 cell var vect-first-element 07:50:42 method init 07:50:42 method destroy 07:50:42 method print 07:50:42 method copy 07:50:42 method element@ 07:50:43 method 2element@ 07:50:45 method map 07:50:47 method norm 07:50:51 end-class vector 07:50:58 its more conventional in this order, c? 07:51:04 I don't think the order matters 07:51:09 --- join: futhin (~thin@h24-64-175-123.cg.shawcable.net) joined #forth 07:51:14 hi ;) 07:51:16 sure 07:51:24 sure it doesnt matter 07:51:46 but its a defacto rule 07:52:45 good morning all :) 07:53:01 nah, lets discuss the end of your code... 07:53:11 futhin: what time is that overthere? 07:53:32 8:54 am here 07:53:49 Etaoin: i will give u an example in a moment 07:54:55 vector instantiate constant v1 07:54:55 3 v1 init 07:54:55 v1 copy v1 vect-add-copy 07:54:55 v1 print 07:55:04 futhin: 16:54 here 07:55:27 onetom: wow :P 07:55:37 futhin: tho, ive just waken up 3hrs b4 :) 07:56:09 i normally get up at 7 am, but i stayed up later than usual last night 07:56:22 futhin: why? ;) 07:56:42 futhin: ive read u back, anyway 07:56:44 morning is leet :P 07:57:01 read me back ? 07:57:14 read back what uve written 07:57:56 yeah 07:58:00 so, now i know uve gotten a grap of my 4th-in-4th idea 07:58:15 a grasp.. yup 07:58:19 and also know that u like Etaoins code too :) 07:58:25 hehe :) 07:58:38 :)) grasp, yes - typo 08:00:02 and do u understand what support am i talkin about by the means of 4th kernel? 08:00:15 support ? 08:00:19 s/supp/multitaskin supp/ 08:00:57 the virtual 4th processor 08:01:13 (what is realized by NEXT & RET) 08:01:44 should b written 2 b able to do context switching 08:02:31 or infact, probably the MAIN loop should do the context switching 08:03:11 somehow i doubt you need to create a kernel just to build context switching :) 08:03:54 i know u doubt it :) but then tell me how would u solve this problem 08:04:42 and sure u dont have to bulid an own kernel (fõleg not in forth) 08:04:59 u should modify the forth implementation u use 08:06:06 fõleg=especially 08:07:38 surely a threading program can be written on top of forth, and then you run other programs inside it... 08:08:50 yes, though it doesnt have too much advantage, but... 08:09:06 i will explain that but l8r 08:09:10 heh :) 08:09:23 now, im just happy ive understand what have i done 08:09:55 s/ive/uve/ -- shit :) 08:10:16 and also s/understand/understood 08:10:48 (eeh, watta beginner am i sometimes :) 08:11:02 (or always? :))) 08:11:34 heh :) 08:12:05 BUT: such a 4th-in-4th (4in4 (c) later on) implementaion 08:12:09 is good 4: 08:12:53 - experimenting w various kernel extensions 08:13:37 - helps prototyping of various dictionary structures 08:14:13 onetom: isn't it fun that you just start explaining 4th in 4th to me and then voila, you've coded up one! 08:14:34 - u can easily debug while prototyping your low-level ideas 08:14:53 --- join: I440r (~mark4@1Cust145.tnt2.bloomington.in.da.uu.net) joined #forth 08:15:05 - u can easily extend this 4th sys w a profiler 08:15:17 are there publically available forth profilers? 08:15:37 futhin: ~:))) 08:15:39 hola i440r, how's the pr0n bot doing? 08:16:06 heh 08:16:09 Etaoin: but u can write your owns easily in 4th 08:16:10 no bot sex for you! 08:17:08 lies! you'll give me full unrestricted access to coding on the bot, and the first thing i'll implement is : SEX ; 08:17:10 - and what could the best of all: u can have a standard forth implementation 08:17:30 I just uploaded a slightly updated version if anybody cares 08:17:36 what could serve as a common denominator 08:17:50 that being good depends on WHAT standard you are talking about 08:17:51 and now, I'm off for a while 08:17:56 ans is NOT good imho 08:17:57 --- quit: Etaoin ("raise KeyError, "I'll bet that's the last time you lock them in the car..."") 08:17:58 when discussing low-level realated 4th problems 08:20:57 I440r: aha, u r right. we could implement the various well known "standards" 08:21:24 i just wrote an itterative fibonaccie function in isforth 08:21:45 this way we gain HIGHLY portable reference source codes 08:21:45 it takes 6 minutes to calculate the 40th fibonacci number using the recursive method 08:22:10 it takes 1 minute 12 seconds to do 9 million itterations of the same using my itterative method 08:22:10 ??? 6mins? 08:22:27 yes 08:22:27 the 40th? 08:22:36 the recursive methoed is very slow on isforth 08:22:39 40 millionth? 08:22:42 no 08:22:44 40th 08:22:49 errr 08:22:51 40 fib 08:22:54 takes 6 minutes 08:23:07 9000000 0 do 40 fib2 loop 08:23:12 takes 1 minute 12 sseconds 08:23:40 who could it b sooo slow? 08:23:46 : fib2 ( n1 --- fib ) 0 1 rot 0 do tuck + loop nip ; 08:24:05 i dunno 08:24:16 show me the recursive 1 08:24:40 : fib dup 1 > if dup 1- fib swap 2- fib + then ; 08:25:12 cripple the recursive ability! prevent morons from using recursive! bwahahaha 08:25:32 lolll 08:25:34 futhin i dont need to cripple it - any recursive function is already crippled 08:25:39 yup 08:25:48 but it's much more obvious in this case ;) 08:25:49 but. i allow anything in isforth 08:26:21 i wrote an itterative region flood fill routine 15 years ago for mark smiley (of fpc fame) 08:26:23 that's weird though.. 6 minutes just to do _one_ iteration of the recursive fib !? 08:26:56 yes but it nests 41 times TWICE in order to calculate the result 08:27:03 its nesting and unnesting thats slow in isforth 08:27:07 I440r: but there is nothing wrong w the recursive version... it calculates a lot more 08:27:35 onetom my itterative methid will calculate exactly the same result in a billionth of the time :P 08:27:39 given the same input 08:28:41 sure but have u counted how man times has fib called? 08:29:07 onetom on the recursive method ? 08:29:12 sure 08:29:16 i440r: the itterative region flood fill routine? how much faster than the recursive version? 08:29:27 let me add a counter - off> counter and incr> counter every time its called 08:29:31 futhin ALOT 08:30:04 ok, meanwhile i'll try to give u and apporx about the result 08:30:08 * onetom thingik.... 08:30:14 * onetom thinkin.... 08:31:14 an=a(n-1)+a(n-2)... 08:32:35 * futhin is afk for a bit 08:32:39 im running 40 fib now on recursive method 08:33:03 u could give it a smaller number... 08:33:27 eg. just try it w 3 fib 4 fib 5 fib ... 08:33:28 wanted to knnow how many times it nested for 40 tho 08:33:44 k, but it lasts f.kin long 08:33:54 hehe 08:34:01 only SIX minutes :P 08:34:17 put it into background & 08:34:28 start another isforth ;) 08:34:56 heh 08:35:05 isforth will eventually fork processes 08:35:15 may even do threading (in the linux sense) 08:36:07 u c, i u run such example code like this fib word 08:36:13 on top of a 4in4 08:36:35 that u can platform independent results about 08:36:58 how fast is it, regardless of the speed of the machine u 08:37:06 run it on 08:37:21 (3 more mins.... ;) 08:37:42 no i messed up 08:37:56 inside fib i put off> counter instead of incr> counter heh 08:38:01 had to stop it and restart :P 08:38:11 6 fib itterates 15 times 08:38:12 noooo 08:38:15 erm 40 times 08:38:18 wait! 08:38:24 5 fib itterates 15 times 08:38:34 5 fib = 40 netst 08:38:37 aha 08:38:39 argh im going insane 08:38:44 5 fib itterates 15 times 08:38:52 6 fib itterates 40 times 08:38:56 & what about 7 8 9 fbi fbi fbi? 08:38:58 40 fib does... 08:39:05 dunno its doing 40 now :) 08:39:10 imean fib, certainly 08:39:19 but its funnier in this form ;) 08:39:41 ah, how greedy u r :) 08:42:47 how many times do you think it will nest for the recursive 40 fib ? 08:42:59 dunno :((( 08:42:59 closest guess ?? :) 08:43:19 ~10^7 08:43:27 or even much more 08:43:27 20 * 15 08:43:40 what? 08:43:58 onetom do u code x86 asm ? 08:44:12 for many yrs i dont 08:44:32 i did it last time ~10yrs b4 08:44:44 have u looked at isforths kernel source ? 08:44:59 not much, but 08:45:11 certainly i ran through it 08:45:21 looking for multitasking support ;) 08:45:29 it will be there 08:45:38 all i have to do is add a fork syscall 08:45:45 and it will be multitasking 08:45:49 and ive also chked yr syscall stuff, remember? 08:46:32 but when im talkin bout mt (short 4 multitaskin) 08:46:56 ok 08:46:58 i think of the ability to run 4th words in paralel 08:46:59 i got a result 08:47:06 nah?! 08:47:07 331160281 08:47:17 it recursed that many times 08:47:20 thats alot! 08:47:27 ~10^9 08:47:41 thats three hundred and thirty one MILLION ++ times! 08:47:42 i wasnt too far away :) 08:48:23 so, your is forth is absolutely not that baaad 08:48:35 just your fib implementation :P 08:48:48 other forths run the recursive method in a minute or less 08:48:53 anyway, how much memory has u allocated 4 it? 08:48:54 run it on gforth 08:49:02 isforth? 08:49:07 or the stack 08:49:13 didnt the os swap a lot? 08:49:45 it doesnt make much sense to run it on gforth, coz 08:49:56 i have a different speed machine... 08:50:09 how fast is your box ? 08:50:11 thats what i was talkin bout... 08:50:26 cant remember, hold on checkin 08:50:37 tom@ex:~> cat /proc/cpuinfo 08:50:44 model name : Celeron (Mendocino) 08:50:45 stepping : 5 08:50:45 cpu MHz : 501.145 08:50:45 cache size : 128 KB 08:50:46 bogomips : 999.42 08:51:13 mines a 550 k6/3 08:51:28 bogomips? 08:51:37 1097.72 :) 08:51:41 cat /proc/cpuinfo 08:51:46 quite close enough 08:51:52 cache 128k 08:51:54 then show me the code 08:52:12 for the recursive ? 08:52:13 hey, dont u have a mouse 2 paste? 08:52:17 sure 08:52:57 : fib1 dup 1 > if dup 1- recurse swap 2- recurse + then ; 08:53:31 : fib2 0 1 rot 1 do tuck + loop nip ; 08:54:11 tom@ex:~/prj/forth/isforth> time gforth -e ': fib1 dup 1 > if dup 1- recurse swap 2 - recurse + then ; 40 fib1 bye' 08:54:26 its already runnin 08:55:44 didnt you need to escale the > ? 08:55:58 ...fib1 dup 1 \> if.... 08:56:01 ? 09:05:28 how long ? 09:07:50 real 1m25.688s 09:07:50 user 1m21.050s 09:07:50 sys 0m0.080s 09:07:55 real 1m25.688s 09:07:55 user 1m21.050s 09:07:55 sys 0m0.080s 09:08:03 real 1m25.688s 09:08:03 user 1m21.050s 09:08:03 sys 0m0.080s 09:08:25 oops, sorry 4 the tripple paste 09:08:29 heh 09:08:47 erm i just ran it your way 09:08:51 from time 09:09:02 and it finished in 8 mins 11 secs real 09:09:06 7 mins 44 user 09:09:11 what does that mean ? 09:09:15 but, wait, let me check wheather it was the fast or the nonfast gforth 09:09:52 what means what? 09:10:15 it took longer according to time ./isforth - : blah blah ; 40 blah 09:10:15 and what command line produced this result? 09:10:22 imean, paste it! 09:10:35 i cnat paste in x but i can show you 09:10:56 ouch, how come u cant paste? 09:11:24 time ./isforth - : fib dup 1 \> if dup 1- fib swap 2- fib + then \; 40 fib bye 09:11:43 because im on my laptop and cut and paste with the touchpad are fucked in the head 09:15:02 dont u use screen? 09:15:17 anyway, what was yr prev question? 09:15:23 [06:10] what does that mean ? 09:15:26 brb 09:15:43 why does it take LONGER according to "time" heh 09:19:29 longer than what? 09:21:27 than NOT using time 09:21:36 isforth has its own timers 09:21:43 time-reset foo .elapsed 09:24:47 gimme an up2date isforth version, k? 09:25:24 what is the substantial src of isforth, anyway? 09:26:06 the what ? 09:26:49 eeer... 09:27:07