00:00:00 --- log: started forth/13.01.10 00:23:16 --- quit: kulp (Ping timeout: 264 seconds) 00:24:15 --- join: kulp (~kulp@unaffiliated/kulp) joined #forth 00:24:15 --- mode: ChanServ set +v kulp 00:29:38 --- quit: impomatic (Quit: http://corewar.co.uk) 00:38:07 --- join: epicmonkey (~epicmonke@188.134.41.174) joined #forth 00:38:07 --- mode: ChanServ set +v epicmonkey 02:03:58 --- quit: I440r (Remote host closed the connection) 03:19:51 --- quit: Nisstyre (Ping timeout: 240 seconds) 03:34:32 --- join: Nisstyre (~yours@oftn/member/Nisstyre) joined #forth 03:34:32 --- mode: ChanServ set +v Nisstyre 03:41:19 --- join: MayDaniel (~MayDaniel@unaffiliated/maydaniel) joined #forth 03:41:19 --- mode: ChanServ set +v MayDaniel 03:42:47 --- quit: Nisstyre (Ping timeout: 276 seconds) 03:57:38 --- join: Nisstyre (~yours@oftn/member/Nisstyre) joined #forth 03:57:38 --- mode: ChanServ set +v Nisstyre 04:05:51 --- quit: Nisstyre (Ping timeout: 260 seconds) 04:07:24 --- join: regreg (~regreg@77.81.148.213) joined #forth 04:07:24 --- mode: ChanServ set +v regreg 04:19:23 --- join: Nisstyre (~yours@oftn/member/Nisstyre) joined #forth 04:19:23 --- mode: ChanServ set +v Nisstyre 04:36:11 Can you use Forth as a general purpose interface shell and scripting language on another OS, a la bash? I'm sure the answer is yes, but has it ever been done competently? 04:45:54 --- quit: Nisstyre (Ping timeout: 265 seconds) 04:53:49 --- join: tgunr_ (~davec@cust-66-249-166-11.static.o1.com) joined #forth 04:53:49 --- mode: ChanServ set +v tgunr_ 04:57:55 --- join: Nisstyre (~yours@oftn/member/Nisstyre) joined #forth 04:57:55 --- mode: ChanServ set +v Nisstyre 05:22:04 --- quit: proteusguy (Ping timeout: 264 seconds) 05:46:13 --- join: proteusguy (~proteusgu@ppp-58-8-99-24.revip2.asianet.co.th) joined #forth 05:46:13 --- mode: ChanServ set +v proteusguy 06:14:53 --- quit: proteusguy (Ping timeout: 276 seconds) 06:37:18 --- join: proteusguy (~proteusgu@ppp-58-8-99-24.revip2.asianet.co.th) joined #forth 06:37:18 --- mode: ChanServ set +v proteusguy 06:42:28 --- quit: proteusguy (Ping timeout: 264 seconds) 06:45:39 --- part: msmith1 left #forth 06:50:33 --- join: beretta (~yaaic@cpe-107-8-120-84.columbus.res.rr.com) joined #forth 06:50:33 --- mode: ChanServ set +v beretta 06:58:39 dTal: that's a bit like asking "what's a good editor?" 06:58:52 oh yeah that was my other question 06:58:56 haha 06:59:07 "what's a good distribution?" 06:59:15 that's easy 06:59:17 debian 06:59:18 next 06:59:20 lol 06:59:31 I'm sure for the forth die-hards the Jupiter Ace was the epitome of Forth-as-shell 06:59:38 s/the/some/ 07:02:23 I'm sure it's easy if the entire system is in Forth. The trouble is interfaces. 07:02:59 If I wanted to stop using bash and start using Forth for all the same tasks, what would have to be done? 07:04:56 at the very least some sort of system that made system binaries look like Forth words that played nicely with the stack, in a predictable fashion 07:08:23 dTal: so, here's a thought: fire up a virtual environment (Xen, KVM, VirtualBox, lxc) and load a minimal Debian install in there using debootstrap (or just copy over a working system). 07:08:27 or just use exec() when a word isnt found 07:08:44 Then, make sure a forth is installed (gforth, yforth, and pforth are all in the repos). 07:09:09 Then, create an account with a login shell set to one of those forths, and see what happens? 07:09:48 and have fun with c string conversion 07:10:10 I'd imagine you'd get a forth as a login shell. Which is all dandy until you want to type 'ls'. 07:10:42 I guess the trouble is that being a login shell isn't very Forthy to begin with. 07:11:07 It's more like writing a shell in Forth. 07:12:26 just interface to the kernel and dont try to "bashify" forth 07:13:49 I guess you'd just have to learn to use eg read-dir open-dir etc instead of 'ls' 07:14:20 yup.. 07:15:10 those things are pretty simple to use 07:16:28 --- join: proteusguy (~proteusgu@ppp-58-10-137-209.revip2.asianet.co.th) joined #forth 07:16:28 --- mode: ChanServ set +v proteusguy 07:17:45 i think there is forth code to iface to .so libs anyway so parameter conv shouldnt be too hard 07:18:48 yeah I've been looking at that 07:19:01 don't quite grok it yet but it looks very elegant and neat 07:20:16 --- join: impomatic (~digital_w@46.208.50.63) joined #forth 07:20:16 --- mode: ChanServ set +v impomatic 07:20:20 it's part of what inspired me to ask the question - I never liked that humans were second class citizens when it came to binary libraries 07:20:58 but there's still an issue with running applications 07:21:07 * dzho sees that there are none of the *-dir commands in [yp]forth, but one of them does have a 'system' command that will run 'ls' 07:21:43 i guess you could make pretend that a lib was just another forth vocab 07:22:14 They all expect GNU-style command switches, which you'd have to construct using strings 07:22:17 uber-pain 07:22:33 yeah: s" ls " system 07:22:56 but the whole point of using forth would be for its extensibility, no? 07:23:09 so, you'd just start by writing 'ls' and work out from there? 07:23:45 It's a pain writing a Forth wrapper for every application you install. 07:24:01 well, see, then maybe 'ls' is not where you'd start 07:24:03 yup id stay away from text params..thats not forrh 07:24:35 ls shouldnt be too horrible 07:24:46 well, there are different ways to do this, and yes I think it'd be fun to have at hand a forth atop just the linux kernel with nothing else 07:25:15 but if one is starting with a Debian system and whittling downward, I think you'd have to accomodate text params for a while 07:25:44 so, maybe the second thing you'd make would be something more like a way of passing arguments down 07:25:48 ARGV or something 07:26:03 accessing the dirent structure will be the fun part 07:26:30 I've sometimes wondered if it's possible to automatically parse manpages to get syntax 07:26:52 They usually have that line in them that's supposed to tell you rigorously what the allowed syntax is 07:27:09 the semantic man-db! 07:27:10 some basic forth to c and c to forth words are in order 07:27:13 of course they're all hugely inconsistant 07:27:31 yeah 07:27:39 like everything else that "just grew" 07:27:51 but on the opposite end you have info pages 07:28:30 when what we're really looking for is something like BNF 07:28:50 i would make your own cp mv ls etc... 07:30:03 most of that stuff is built into bash, you might as well build it into forth 07:30:12 good point 07:30:46 and learn a bunch in the process 07:30:54 (actually they're all system binaries, but I still agree) 07:32:54 then you wouldnt have to worry about crappy text parsing args 07:40:41 --- quit: Nisstyre (Ping timeout: 276 seconds) 07:55:20 --- join: Nisstyre (~yours@oftn/member/Nisstyre) joined #forth 07:55:21 --- mode: ChanServ set +v Nisstyre 07:59:52 --- quit: proteusguy (Ping timeout: 264 seconds) 08:58:00 --- quit: impomatic (Quit: impomatic) 08:58:26 --- join: kumul (~kumul@173.215.194.228) joined #forth 08:58:26 --- mode: ChanServ set +v kumul 09:03:31 --- quit: beretta (Ping timeout: 276 seconds) 09:04:15 I thought someone had written a forth "shell", but I would bet you could use FICL and create what you are looking for 09:16:33 goodness this channel has become the most active of the ones i've been in 09:17:06 there was something called fth, it wanted to be for the shell and derived from ficl 09:17:18 *its purpose was* 09:17:48 it would be quite strange doing shellscript programs in forth, although possible 09:17:59 --- join: anannie (~chatzilla@unaffiliated/anannie) joined #forth 09:17:59 --- mode: ChanServ set +v anannie 09:18:35 weird how many forths have pwd and ls and it works, i dont know the purpose of doing ls on forth 09:19:09 Folks decided loading & running a program was too slow. 09:19:13 hello anannie 09:19:25 hello kumul 09:22:05 that doesnt make sense to me PoppaVic, running a program? what do you mean? 09:22:29 you mean forking? 09:22:35 it is quite slow actually 09:23:12 what do you mean it makes no sense?? In earlier systems, you simply loaded a new .com file and ran it, reloading the prior program on completion. Fork/spawn remains viable. 09:23:37 running expr n + 5 in a while loop on a shell is extremely slow 09:23:45 Only a complete dumbass would try to build a "shell" - any application - replicating every damned program already available on a system. 09:27:40 by shell i mean unix shell, bourne shell. you cant load a .com file, was talking about unix specifically 09:29:03 Yes. I was aware of that. I mentioned .com from the historical cp/m - which would still apply in assorted places. meanwhile, any unix has so many available support programs that replicating their functionality into one massive "shell" is just plain stupid. 09:31:30 who was suggesting that? 09:32:06 Thus, I'd fork for ls, and a number of other things. 09:32:58 although, I always felt forth had issues with string-handling. 09:34:21 you mean that there arent any? 09:47:58 Any Macintosh programmers online to help me test a Forth mdimporter 09:48:29 --- join: proteusguy (~proteusgu@ppp-58-8-99-24.revip2.asianet.co.th) joined #forth 09:48:29 --- mode: ChanServ set +v proteusguy 10:01:55 --- quit: karswell_ (Remote host closed the connection) 10:02:15 --- quit: kulp (*.net *.split) 10:03:14 --- join: kulp (~kulp@unaffiliated/kulp) joined #forth 10:03:14 --- mode: ChanServ set +v kulp 10:06:00 --- join: ASau (~user@176.0.64.166) joined #forth 10:06:00 --- mode: ChanServ set +v ASau 10:11:50 --- join: beretta (~beretta@cpe-107-8-120-84.columbus.res.rr.com) joined #forth 10:11:50 --- mode: ChanServ set +v beretta 10:12:14 --- join: karswell_ (~coat@93-97-29-243.zone5.bethere.co.uk) joined #forth 10:12:14 --- mode: ChanServ set +v karswell_ 10:40:27 --- quit: karswell_ (Remote host closed the connection) 10:50:45 --- join: karswell_ (~coat@93-97-29-243.zone5.bethere.co.uk) joined #forth 10:50:45 --- mode: ChanServ set +v karswell_ 11:21:49 kulp: did you get a chance to diddle that file? 11:22:22 --- join: MayDaniel_ (~MayDaniel@unaffiliated/maydaniel) joined #forth 11:22:22 --- mode: ChanServ set +v MayDaniel_ 11:22:30 PoppaVic: no i cleaned my room instead :P 11:22:33 now it is very clean 11:22:37 and i am no smarter 11:22:43 hahaha 11:22:48 i heard you fixed it, did you update it too ? 11:23:14 I did paste an updated link or three, yeah 11:23:39 oh ok i see http://ideone.com/kVQqdt now 11:24:03 yah 11:24:33 --- quit: MayDaniel (Ping timeout: 248 seconds) 11:25:04 --- join: I440r (~zhiming@203.sub-70-194-130.myvzw.com) joined #forth 11:25:04 --- mode: ChanServ set +o I440r 11:26:57 --- quit: karswell_ (Read error: Connection reset by peer) 11:37:14 --- join: karswell_ (~coat@93-97-29-243.zone5.bethere.co.uk) joined #forth 11:37:14 --- mode: ChanServ set +v karswell_ 11:59:59 --- quit: beretta (Ping timeout: 240 seconds) 12:13:19 --- join: RodgerTheGreat (~rodger@71-13-215-242.dhcp.mrqt.mi.charter.com) joined #forth 12:13:19 --- mode: ChanServ set +v RodgerTheGreat 12:13:26 howdy folks 12:13:30 howdy 12:13:33 anannie: hey 12:13:42 did you get E12 working? 12:13:42 how ya doing RodgerTheGreat? 12:13:56 not great, but surviving 12:13:59 Haven't had the chance to look at it, but I'll do it right now 12:14:03 I'm sorry to hear that 12:14:09 --- quit: I440r (Ping timeout: 248 seconds) 12:15:03 Would you like to talk about it? 12:15:17 --- join: I440r (~zhiming@203.sub-70-194-130.myvzw.com) joined #forth 12:15:17 --- mode: ChanServ set +o I440r 12:15:26 um 12:16:20 you could always try sticking to the topic - just to be different, of course ;-) 12:16:58 * kulp puts a damper on PoppaVic 12:29:44 meh, dont do that, or else no one will speak 12:31:31 lol 12:31:56 the topic of #forth is similar to the forth programming language in general... "User Defined" 12:33:14 It's better to talk about something and let the community evolve than just let it whither in silence 12:34:29 --- quit: karswell_ (Read error: Connection reset by peer) 12:35:26 plus im sick of going into #whatever-cannel and being pounced on by the topic ghestapo 12:35:36 so i refused to implement that policy in here :P~ 12:37:54 --- part: PoppaVic left #forth 12:44:43 --- join: karswell_ (~coat@93-97-29-243.zone5.bethere.co.uk) joined #forth 12:44:44 --- mode: ChanServ set +v karswell_ 12:49:00 anannie: if you're still having trouble with E12, one suggestion I'd give is that finding the prime factorization of the triangle numbers is a fair bit more work than is actually necessary 12:49:20 it isn't an *incorrect* approach, but there are efficient solutions which are much simpler 12:49:24 Actually I'm figuring out how to short circuit things 12:49:29 and well I haven't found a way yet 12:50:16 I solved it earlier today and my approach was to start with the dead-simple brute force approach, strip it down to the bones and then think about ways to shave off unnecessary work 12:50:45 there are some obvious optimizations which aren't sufficient but which lead in the right direction 12:51:52 try thinking about it starting from the basis of knowing that it is not necessary to use primality to solve the problem efficiently 12:52:08 --- quit: regreg (Remote host closed the connection) 12:52:22 anannie: um, is that helpful at all? 12:54:13 --- join: rabenauge_ (~rabenauge@88.130.160.127) joined #forth 12:54:13 --- mode: ChanServ set +v rabenauge_ 12:54:25 --- quit: rabenauge_ (Remote host closed the connection) 12:54:41 --- quit: anannie (Ping timeout: 248 seconds) 12:54:45 d'aw 12:54:59 well shucks 12:55:47 --- quit: karswell_ (Remote host closed the connection) 12:56:29 I440r: so what are you up to 12:56:45 --- join: rabenauge (~sag@88.130.160.127) joined #forth 12:56:45 --- mode: ChanServ set +v rabenauge 12:57:00 well im adding pulldown menus to isforths text user interface. 12:57:10 i have the menu creation done, am working on the menu display code now 12:57:13 still working along on that 12:57:15 then i write the menu control stuff 12:57:19 i never rush code 12:57:27 have you found any more bugs in people's terminal emulation? 12:57:49 oh no man, I understand- code is as much about the process as the result 12:57:53 i stopped looking for them beause there are so many of them its too difficult to keep up with them all 12:58:23 RodgerTheGreat, mostly its the result thats important, most coders FORGET but its "the end users" that are THE point to code 12:59:11 yeah but if you want it to come out right and you don't have someone screaming at you to get it done faster you should take your time and start over when the time is right 12:59:45 an argument can also be made for code-as-art, in which code CAN be the point of code 13:00:00 screaming to "git-r-dun" doesnt work with me 13:00:05 you want it right? or "right now" 13:00:07 pick one 13:00:13 good, cheap, fast. pick 2 13:00:15 you'd make a terrible PHP programmer. :) 13:00:33 php makes php coders terrible php programmers 13:06:05 --- join: karswell_ (~coat@93-97-29-243.zone5.bethere.co.uk) joined #forth 13:06:05 --- mode: ChanServ set +v karswell_ 13:08:23 RodgerTheGreat, code IS art. but its not art in the same way as "rent a jet for the day. pay pilot to sit in aircraft and rev the jets so i can throw paint into the jetwash and spalsh it all over the canvas and sell the 3 paintings i did for 6 million each" 13:08:56 not Hirst. Check. 13:08:59 which is what an "artits" did at the airport my father worked in when i was a kid 13:09:30 and then you get an artist that doesnt want to paint at all. so he takes an empty canvas and hangs it on the wall 13:09:34 --- dire straits -- 13:10:34 probably not a Rothko fan either, hunh? 13:14:03 who? 13:14:22 http://en.wikipedia.org/wiki/Mark_Rothko 13:22:15 --- join: aesping|work (~rixard@host-95-199-222-139.mobileonline.telia.com) joined #forth 13:22:19 --- mode: ChanServ set +v aesping|work 13:46:28 --- quit: kumul (Read error: Connection reset by peer) 13:48:57 --- quit: tgunr_ (Read error: No route to host) 13:50:03 --- join: tgunr_ (~davec@cust-66-249-166-11.static.o1.com) joined #forth 13:50:03 --- mode: ChanServ set +v tgunr_ 14:20:47 --- quit: aesping|work (Quit: Colloquy for iPhone - http://colloquy.mobi) 14:32:11 --- join: anannie (~chatzilla@unaffiliated/anannie) joined #forth 14:32:11 --- mode: ChanServ set +v anannie 14:32:22 RodgerTheGreat: I solved it :D 14:32:30 rockin'! 14:32:49 share your solution! 14:33:06 It runs in .5s!!! 14:33:37 it's probably actually faster than that 14:33:58 half a second is within the jitter offered by many system clocks depending on how you timed it 14:34:03 yeah but ruby is quite slow, so you'll have to forgive me 14:37:30 I wish to see your code 14:37:39 --- quit: MayDaniel_ (Read error: Connection reset by peer) 14:39:39 https://gist.github.com/4506430 14:39:42 It's horrible RodgerTheGreat 14:39:49 I haven't worked on it 14:39:53 it's just functional 14:39:56 yow 14:40:00 that is fairly complicated 14:40:36 here is what I was saying before you lost your connection: http://hastebin.com/raw/sofegacenu.txt 14:42:21 I just ported my implementation to logo and I am figuring out how many aeons it will take to run 14:42:32 for 50 factors it takes about 30 seconds 14:42:47 this is a logo written in forth running on a vm written in java 14:43:20 anannie: still with us? 14:43:23 Yup 14:43:53 You don't need primality to solve this? 14:43:57 O.o 14:44:29 At the end of the day you will have to run a check whether it's up to root(num) to find the factor pairs and so on 14:44:48 yes 14:45:41 I want to tell you the simple way to do this so badly but I know you'll have more fun figuring it out 14:45:52 you're super close 14:46:21 These words will probably come back to haunt me, but prod me in the right direction... What am I looking for? 14:46:51 well, remember that you just care about *factors* 14:47:10 you're doing extra work to get a prime factorization and then work back again to factors 14:47:11 so recursive decomposition? 14:47:20 mmno 14:47:30 think dumb search 14:47:46 first, what is the brute force way? 14:48:04 and why does the work to do it that way grow too fast? 14:48:22 okay so for each number that's divisible by it up to root(num) add two? 14:48:30 yeah 14:48:35 exactly 14:48:46 that's actually more inefficient 14:49:06 it's still very fast and has dramatically fewer moving parts 14:49:26 if you think about it, for larger numbers you have to process *every* number all over again until that point 14:49:39 Whereas in my situation the prime list grows side by side 14:49:51 and the number carries forward 14:49:56 the number of factors you must search grows logarithmically instead of geometrically due to the square root upper limit 14:50:09 so you only have a limited number of numbers per number to search through 14:50:33 so it has a O(logn) runtime? 14:50:41 also don't ignore the overhead of having data structures in your approach whereas if the only state is a couple ints in a loop everything is in registers 14:51:16 I may be wrong but I think it is something like nlgn 14:51:49 yes, but it's the classic memoization versus run-time calculation question 14:51:56 absolutely 14:52:05 I don't know much about comp sci 14:52:23 you know what the word "memoization" means, which makes you better than some CS majors I know 14:52:26 I never went to college, but I would like to learn how to predict whether the algo is nlogn or so on 14:52:43 meh reading up crap on the internet doesn't mean jack shit 14:53:51 a quick and dirty approach to algorithmic analysis is to plot runtime versus n. If the line is straight, linear. If the line is straight on a log plot, exponential 14:54:41 I think there are actually some test frameworks in ruby which allow you to assert performance for a procedure is, say, roughly linear 14:55:48 lies damned lies and statistics ;) 14:56:14 anannie: if you want to read crap on a dead tree instead of the internet, the Cormen text is the gold standard: http://www.amazon.com/Introduction-Algorithms-Thomas-H-Cormen/dp/0262033844/ref=sr_1_2?ie=UTF8&qid=1357858518&sr=8-2&keywords=Algorithms 14:56:19 :) 14:56:47 i trust dead tree format more... but not entirely :) 14:56:53 I prefer dead tree as well 14:57:05 plus trees ARE a renewable resource 14:57:09 yep 14:57:18 there are more trees in the USA today than there were the day the pilgrims landed 14:57:32 nice thing about digital docs are they're easier to search 14:57:44 nice thing about paper is you can have a bunch open on your desk at once 14:57:57 you just need MORE MONITORS!!!!! 14:57:58 lol 14:58:01 here is my quick-and-dirty logo solution to E12: http://hastebin.com/hoqacanawe.logo 14:58:08 Why isn't protist around when you need him? He nerd sniped me and then disappeared, I'm so totally gonna rub my solution in his face (after I get it working within .09s( 14:58:10 I440r: I'm a graduate student. :'( 14:58:13 :p 14:58:30 and im an unemployed broke guy 14:58:43 I440r: That can change rather quickly 14:58:50 by our powers combined... uh... 14:59:45 Okay, so let's whittle this down to something that runs within .1s 15:00:11 you have used the range for searching a prime up to sqrt(n), but isn't it just (sqrt(n)-1) the maximum range to test? 15:00:42 anannie, if it doesnt ill still be ok :) 15:01:11 well something can still be composite if it is n*n so you have to check sqrt(n), right ? 15:01:11 I440r: You're the heir to a great fortune or something? 15:01:15 * kulp says, not having looked at any code 15:01:43 anannie, isforth compiles 228831 bytes of source code in .3 of a second ! 15:01:57 Whoa that's awesome I440r 15:02:12 anannie, no... im a christian. "who soever cannot give up everything they own cannot be my deciple" plus... i have friends i can rely on heh 15:02:26 kulp: Eh? 15:02:35 anannie: i was responding to rabenauge 15:02:44 probably foolishly, since i lack context 15:02:46 kulp: The problem statement is for triangular numbers, so that doesn't quite parse 15:02:49 but finding work not not finding work... thers a highre authority deciding and controling that :) 15:02:52 kulp: Ah. 15:03:05 anyway it parses fine, it just doesn't make sense ;) 15:03:12 lol 15:03:27 no syntax error? it just doesnt run as expected? 15:03:28 * kulp EXCUSES HIMSELF EXCUSILY 15:05:42 I440r: let's just hope it isn't undefined behavior 15:06:29 there was a joke about some programming language where each day the users would come in and type their name and try to figure out what atiaon their name performed on that day... i forget what language it was in reference to tho 15:06:42 what action 15:06:47 man i need to L2 type 15:06:54 I thought that was TECO 15:07:08 or vi I guess 15:07:23 no it was in reference to an actual programming language. 15:07:40 well even perl isn't that goofy 15:08:00 I guess you could try it in K or similar but most letters are just used for variables 15:08:24 algol or something... some ancient thing 15:08:51 could you give an example for a composite of nÂ² where n is greater 3 that had to be testet for when testing for a prime? 15:09:43 I440r: algol doesn't look that different from pascal really 15:10:04 i wasnt sayint it was in reference to algol... just something OLD like algol lol 15:10:38 SNOBOL man, now there was a horror 15:11:00 one of my professors gave me a manual to read and I think it took years off my life 15:11:23 it's basically tcl from hell 15:13:42 rabenauge: Not a single one if you do the 6k +/- 1 15:17:00 I am not sure, but I am curious if there is any nÂ² (n>3) that is an odd nÂ² product (2 factors composite) that one had to test for when looking for primes. so perhaps it is already enough to look up to sqrt(n)-1 ? but that's just an idea... 15:18:40 I440r: following my previous analogy maybe I should make a language which is somewhere between tcl and snobol 15:18:46 call it "grope" or something 15:18:49 maybe all remaining odd non-primes fail long time before getting close to sqrt(n)? 15:18:51 --- quit: karswell_ (Remote host closed the connection) 15:22:08 :) 15:22:22 as retribution for them ever creating either of those 2? 15:22:33 RodgerTheGreat: I'm thinking about cheating a bit. 15:22:44 do tell 15:22:59 RodgerTheGreat: I think that I can get the run time down to .09s if I store the prime list to disc for later retrieval 15:23:09 erm 15:23:14 how big is that list? 15:23:31 hrm lemme tell you in a bit 15:23:51 I strongly suspect that it will take longer to load it from disk than compute 15:24:04 but I do invite you to try it, profile it and prove me wrong 15:24:49 maybe not if she has an SSD.. but mayte that died with the mbp :( 15:25:29 even an SSD has a lot more latency than ram and in turn cache 15:25:52 if the prime list is a couple dozen kilobytes it should all fit easily in cache 15:26:25 anannie: also you're storing the primes as an actual list of numbers rather than a bitmap, right? 15:27:05 as a bitmap you would turn a pixle on only for pixles whose x*y position was prime? 15:27:20 I mean bitmap as in bit vector 15:27:22 that would be a HUGE image 15:27:30 not like a png or something 15:27:35 lol kk 15:28:05 and you can get some savings by remembering that EVERY multiple of 2 except for 2 is composite 15:28:13 so halve your indices 15:28:28 I don't know if you're at all familiar with farbrausch and the demoscene, but I remember reading them talk about how their procedural texture generators were expected to be quite slow as they are computationally intensive, but in practice they could unfold them into ram faster than they could have loaded from disk 15:28:50 Wow 15:29:05 those are the guys who wrote KKrieger 15:29:08 --- join: karswell_ (~coat@93-97-29-243.zone5.bethere.co.uk) joined #forth 15:29:08 --- mode: ChanServ set +v karswell_ 15:29:15 http://en.wikipedia.org/wiki/.kkrieger 15:29:22 any of 5 as well, RodgerTheGreat. 15:29:28 You're probably right, I weighed that against my head and I'm generating just 1476 primes... I can generate 10k primes within .3s so this isn't the bottle neck 15:29:37 --- join: ncv (~quassel@89.35.216.197) joined #forth 15:29:37 --- quit: ncv (Changing host) 15:29:37 --- join: ncv (~quassel@unaffiliated/neceve) joined #forth 15:29:37 --- mode: ChanServ set +v ncv 15:29:57 I wanted to try doing it though, but opening the file and then doing it is probably going to take longer 15:30:03 anannie: it was a good idea, but you're not computing on a sufficiently large n for it to be helpful 15:30:45 RodgerTheGreat: I wanted to implement Gosper's Hash life btw, but I don't know enough and I don't think I'm smart enough to implement it 15:30:59 if you really want to know how much it speeds up your program to do that you could print out an array literal for the 1476 primes and paste that into your program 15:31:06 oh pshaw 15:31:10 can't be that bad 15:31:28 what is the range of numbers anyways, anannie? for a small range you could really fill a bitmap with the multiples of numbers and your gaps are primes afterwards. that can be very fast as well... 15:31:55 rabenauge: No range, it's an open ended search for a solution 15:32:13 RodgerTheGreat: That algorithm is pretty damn complex 15:32:14 sees 15:32:32 anannie: wasn't there just an HN article about that? 15:32:46 RodgerTheGreat: Oh? 15:33:04 yes, it linked to this: http://www.drdobbs.com/jvm/an-algorithm-for-compressing-space-and-t/184406478 15:33:29 if a number cannot be evenly divided by any known prime number lower than itself... is that number a prime? 15:33:32 http://www.drdobbs.com/jvm/an-algorithm-for-compressing-space-and-t/184406478 15:33:35 Ah yes that one 15:33:58 I haven't read it in detail but it doesn't *look* that bad 15:34:14 I'll read it tomorrow when I wake up 15:34:23 I hadn't come across that before 15:34:26 what time zone are you in? 15:34:32 /me tucks anannie in.. repeatedly! 15:34:37 I can't tell you that! 15:34:49 * RodgerTheGreat sighs 15:34:57 she is in cambridge england... we have been over this! 15:34:58 lol 15:35:04 it's not like I'm trying to squeeze information out of you or something 15:35:53 RodgerTheGreat: I know you aren't, but I'm worried that daddy dearest might have intelligence contacts in which case I'm screwed 15:36:09 they tapped my friend's phones and monitored their emails 15:36:21 it was some pretty heavy crazy shit 15:36:30 how about you tell us when you're safe? 15:36:36 I shall. 15:36:43 :) 15:37:18 did you try statistics about number of composites in non-primes so far? a statistical approach might tighten search space perhaps? 15:39:07 rabenauge: I'm afraid that my approach doesn't work that way.... What I do is I basically calculate the factors of each number to quickly figure out the factorisation of the triangular number 15:39:09 I think that it is really worth trying a data structure-free approach- allocation/reallocation/garbage collection and multiple layers of dynamic dispatch can really add up over using ints 15:40:14 in my experience with java, using longs as bit-vectors passed through recursive procedures is *dramatically* faster than doing the same thing with hashsets 15:40:22 Oh I just had a thought, over here multiple arrays are probably being generated in different memory locations... I wonder if there was a way to make the test in place and if it would make it faster 15:40:40 could give better locality maybe 15:41:07 I wish I'd gone to college as I want to learn all of these beautiful things that you're talking about RodgerTheGreat, 15:41:24 anannie, your only 20. its not too late! 15:41:37 well I'd love to teach you anything I can 15:41:40 my uncle went to uni at 50 for a degree in something or other i forget 15:42:10 hehe..."for a degree of something"...nice phrase! 15:42:24 RodgerTheGreat: I'd love to learn everything I can. 15:42:31 well he was airforce inteligence prior to that and then a DEA agent... 15:42:41 I440r: Oh I intend to get an education, but I'll have to do it on my own I'm afraid 15:42:42 anannie: I like you. :) 15:42:42 i think he wanted a career change :) 15:43:11 thats what i did but sometimes i see the weakness in my education 15:43:12 I would love to get a degree in something :) But I just did computing science -.- 15:43:13 or lack thereof 15:43:41 I440r: I intend to be more rigorous 15:43:49 i could never suffer getting a degree in computer science. my mind would outright reject 99% of what they tried to cram into it 15:44:22 it is good to temper academia with real-world experience 15:44:38 is that even possible :) 15:44:44 brb, gotta cook something :) 15:44:54 It is 15:44:56 it is indeed 15:45:03 lots of people go into CS thinking they're going to learn to make videogames or websites or something and are shocked to discover it's essentially applied mathematics 15:45:12 well, almost the same for me, but I took the easy way of studies in operating systems architecure and go away with it :) 15:45:30 but at the same time I am often shocked at how many CS majors don't learn to program very well 15:45:41 coding is your superpower, dudes! 15:45:54 computer scientists are mathematicians with power tools 15:45:57 RodgerTheGreat: Exactly and there isn't a degree for what I want to do 15:46:06 well what do you want to do? 15:46:45 I'm going to build the first great robotics company 15:46:58 I'm going to help make that revolution happen. 15:47:02 that's ambitious 15:47:10 maybe you want a degree in mechatronics 15:47:14 well no one said that I lacked ambition 15:47:21 which is one of the most awesome sounding degree names btw 15:47:22 Meh I've checked them out 15:47:29 there's a reason why MIT doesn't offer one 15:48:28 Here's what I'm going to do instead; 15:48:31 * Mathematics; An understanding of multi-variable calculus, vector calculus, differential equations, partial differential equations, linear algebra, higher probability and statistics, linear algebra, undergrad level discrete mathematics mixed with some grad level stuff, Basic to intermediate level Analysis, Applied Mathematics - Principles and helpful techniques 15:48:33 * Computing; I'm going to acquire a functional understanding of algorithms and analysis of algorithms, Control theory, and Signal processing. I really want to focus on how to create feedback loops with encoded logic that can respond to dynamic environments with ease. 15:48:34 * Electronics; An understanding of micro-controllers and circuits - enough to build a micromouse on my own by combining everything together, analysis of circuits, MIT's signals course. 15:48:36 * Business : At least 1 nano-"startup" that solves a hard problem by applying what I have learnt and brings in non-trivial revenue greater than $5000/month. 15:48:37 * Mechanical Engineering; Brushing up on essential physics within Giancoli along with modern physics at the end, 20 hardware robots that do something interesting. 15:49:22 did you read valentin braitenberg "cybernetic vehicles"? he did some very basic research about autonomous robotics in the 80th/90th. nice stuff! 15:49:41 I haven't rabenauge Thanks for the pointer though! 15:49:42 on the computing side I recommend learning about automata and formal models of computation 15:50:16 Well RodgerTheGreat I planned to understand Rivest's algorithms and a lot of the good stuff in there has automata and it follows from there 15:50:26 ok cool 15:50:26 http://en.wikipedia.org/wiki/Valentino_Braitenberg 15:50:55 --- join: kumul (~kumul@173.215.194.228) joined #forth 15:50:55 --- mode: ChanServ set +v kumul 15:50:57 I have to read this, thank you rabenauge! 15:51:15 rabenauge: ah, Vehicles is great 15:51:23 you're welcome 15:51:27 It's all in the air right now, RodgerTheGreat. I have to first improve my problem solving techniques and get into coding, hence the entire project euler thing 15:51:27 reminds me of a lot of Mark Tilden's work 15:51:34 BEAM? 15:51:37 yeah 15:51:43 tilden is one of my heroes 15:51:47 That flashed through my mind as well 15:53:09 I remember seeing "Walkman Daggerwrist" on a PBS special as a kid and it was enchanting 15:53:48 the idea of making simple robots out of essentially junk but having them capable of adapting to damage and changing environments was wonderful 15:58:31 RodgerTheGreat: I had a similar moment as a child and it's what prompted me to pick up robotics 16:01:58 heh RodgerTheGreat my implementation is actually slower than what I had before! 16:02:14 It's probably because I'm modifying the array in place 16:02:42 are you changing the size? 16:03:15 --- join: ncv_ (~quassel@89.35.216.197) joined #forth 16:03:15 --- mode: ChanServ set +v ncv_ 16:03:17 Yes I'm appending it to the array 16:03:39 This is weird, but this code actually takes longer to execute 16:03:45 I wonder what I'm doing wrong 16:04:21 if you're changing the size of the array, that means that somewhere under the hood you are allocating new arrays and copying the data from one to another 16:04:55 Ah 16:05:03 scripting languages like ruby often hide this fact but the performance impact shows through 16:05:07 Can I simply not add the number to the next memory location directly 16:05:26 can you create an array of the appropriate size up front? 16:05:31 I believe, robotics suffer the most from their weight (resulting in energy consumption). So a good aproach might be also research in "printable ceramics" (3d-laser-printing/prototyping). ceramics are light weight and combined with alloy/titanium molecules very sturdy (metall-ceramics). 16:05:38 This could also be an interesting field of research along with robotics, anannie. 16:05:47 if you just write to cells in the array rather than append to the array there is no reallocation necessary 16:06:47 rabenauge: The problem is that they are also quite expensive 16:06:54 nope RodgerTheGreat 16:07:11 Wait, I definitely can, hrm. 16:07:30 if I was more familiar with ruby I could be of more help 16:07:43 well depending on their components. some are not too expensive. but might need research and love to besome valuable 16:07:46 * rabenauge goes to bed for now. getting very late over here in europe ... already past midnight 16:08:00 sees people! 16:08:36