00:00:00 --- log: started forth/03.05.09 00:13:49 --- join: gilbertdeb (~gilbert@fl-nked-ubr2-c3a-74.dad.adelphia.net) joined #forth 00:14:09 anyone read paul graham's latest article http://www.paulgraham.com/hp.html ? 00:23:08 not yet 00:23:17 that lisper guy? 00:23:41 hehe 00:24:50 its pretty decent. you might like it. 00:40:26 --- join: ramnull (~nicad@12-241-145-39.client.attbi.com) joined #forth 00:40:39 What's up? 00:40:53 paul graham's latest article http://www.paulgraham.com/hp.html ? 00:41:10 Havent read it yet. I'll check it out. 00:41:46 gilbertdeb: You know much about exokernels? 00:42:06 no thats kc's domain. 00:42:18 Just curious. 00:42:33 I've read a bit about them. 00:42:48 all I ever learnt about exokernels, I learnt and continue to learn from kc. 00:42:51 I think they are a nifty idea! 00:43:14 Time for a netsplit. 00:43:55 I've been thinking of something like a exomicrokernel. I'm calling it a minikernel. 00:44:14 there is no exo in mini :D 00:44:54 Yeah, I know. I just havent thought up a good term for it yet. I think minikernel describes it best. 00:46:14 Basically a kernel that starts execution as an exokernel, and then switches into sysadmin defined modes. etc... 00:46:35 have you seen/used plan9? 00:46:54 I've looked at it briefly. Dont know much about it. 00:47:10 just curious. 00:47:18 they might have some ideas worth looking into. 00:47:31 I was reading earlier about the SqueakNOS (no operating system) 00:47:38 and I rather liked the way it sounded. 00:47:53 Theres lots of good concepts out there. 00:48:01 some person mentioned in #squeak that they had a machine which had nothing but asm beneath their squeak. 00:48:02 no os. 00:48:35 Sounds alot like ColorForth doesnt it? :] 00:48:54 very! 00:49:06 have you ever played with squeak? 00:49:25 Nope. 00:49:54 any smalltalk? 00:50:21 Nope. I'm into J and K though. 00:50:45 Checking out Mercury. 00:50:46 hmmm. 00:51:06 you should give it a look. its as different as forth is different and j/k are different and lisp is different. 00:52:10 I will. 00:52:31 Figures, the thunderstorm doesnt start until AFTER I boot up my computer. 00:53:03 ramnull: the gods of thunder have a better sense of humor than the tvweatherfolk :D 00:53:34 Just got home from work. Had the power shut off cause of the storm was supposed to hit while I was working. 00:54:32 No UPS on this machine. 01:00:22 Working on switching my assembler code to Forth code. 01:01:14 whats the asm about? 01:01:33 gilbertdeb: All the Framebuffer stuff I'm working on. 01:01:52 I was trying to just Assemble it into IsForth. 01:02:00 But that's a PITA. 01:02:58 Probably the only thing that'll be in Assembler are the primitive drawing routines. 01:04:26 Getting and setting modes is the trickiest part of the code, as an error could require a reboot in order to recover the terminal. 01:07:26 I might just finish the primitives and then send the files to I440r, as his terminal routines assume text mode, which could cause problems when switching to graphics mode. 01:08:20 I feel a lot of this is needlessly complicated. 01:08:27 what is causing all this brittleness? 01:08:48 Well, right now... 01:09:26 On linux I actually have to deal with two different drivers. The terminal driver and the Framebuffer driver ... 01:09:53 Each one is written to assume total control of the I/O device. 01:10:12 Any code written for one has to be ported over to the other. 01:11:19 The current IsForth uses the terminal drivers, so not only do I have to switch out of graphics mode, but I have to pass control back to the terminal driver. 01:12:52 When your in graphics mode, in order to have keyboard or mouse or anything, that has to be explicitly coded also. 01:13:46 the gods of OO will turn in their graves at that comment 01:14:50 So right now, I can get into graphics mode, which then segfaults and leaves me without a screen. Of course the terminal driver kicks in, but because the graphics card is set to graphics mode, you cant see anything(but you can type reboot just as well). 01:15:25 can you not make it switch to terminal mode ? 01:16:37 You can, if the code makes it that far. Heh. I have to tell the graphics card to switch back to text mode before handing control back over to the termio driver. 01:17:30 See, the termio driver knows nothing about graphics. As far as it's concerned ALL is text mode. 01:18:23 I can understand that. 01:19:03 It was designed for passing info between bunches of different kinds of terminals. Not for accelerated graphics cards. 01:19:48 That's what the Framebuffer is for. 01:20:34 I'm sure it's something simple I'm overlooking. I'm still a greenhorn as far as low level coding is concerned. 01:20:48 I can get it to work in Assembler though. 01:20:57 I'm a greenhorn period :D 01:20:58 shhh 01:21:37 Heh. 01:22:47 I have a routine called "fb_error" that's suppose to do that automatically if something goes wrong, so it must be dying between the switch to graphics and the jump to fb_error. 01:24:07 But man, this thing is gonna be so cool when I get it working right. 01:24:32 I can see demo coders lining up by the dozens to mess with it. 01:24:53 you mean 4thers? 01:25:22 Maybe 4thers, Forthers. But no, demo coders. Ever heard of the demo scene? 01:26:42 ramnull: what are you working on? 01:27:04 Klaw: A Framebuffer extension to IsForth. 01:27:49 Klaw: If I can get mode switching to work, the rest will be easy. 01:28:32 After that, it's all drawing and font routines. 01:29:35 I feel like a tart because I'm having so much trouble with it. 01:29:57 --- join: karingo (karingo@188.portland-11-12rs.or.dial-access.att.net) joined #forth 01:29:58 ramnull: take a break and read the paul graham article. 01:30:03 you'll love it. 01:30:05 * ramnull nods 01:30:10 infact, you might even stop feeling like a tart. 01:30:44 Klaw: You do any graphics programming? 01:30:48 none :P 01:31:02 i'm entirely webbased 01:31:54 * ramnull is lurking and reading the article 01:31:57 4 years of fulltime PHP -- all as sole programmer 01:32:05 until this last year, now i've got 8 guys i'm trying to manage 01:32:39 ye gads klaw! who is trying to apply the peter principle on you? 01:34:12 the heirarchy ends with me 01:34:21 :) 01:34:31 i did the CRM/buzzword feh for three crap startups 01:34:42 but the CRM was mine the whole time, i was basically a service provider for them 01:34:58 are the startups still alive? 01:35:07 they weren't the good kind of startups 01:35:10 i've some shame :P 01:35:21 i partnered with someone this last time, about a year and 3 months later i've finally hired help 01:35:53 the first startups were telemarketing rooms, i made the data entry stuffs (just forms via internet explorer) 01:36:15 the very first, i was under 18, bad situation. i did many failed projects in all manner of web stuffs. 01:36:19 back in the internet boom. 01:36:33 anyway, point is that my experience is very specific to the internet. 01:36:35 and your clients didn't know any better :D 01:36:44 know any better about what? 01:37:01 whatever on earth it is they were startingup to begin with. 01:37:04 re: boom 01:37:07 oh, they were just con artists 01:37:16 the first guy was a loser, otherwise he'd have made it in the boom 01:37:40 i created some decent stuff.. I know an "online office" is usually a trite little app, but I did the full suite and had it available as a $10 / month service.. 01:37:45 way back when. 01:37:46 what solutoin was the first guy offering? 01:38:01 well, let me test your knowledge of scams :P 01:38:11 he was offering a home business opportunity 01:38:21 aka, an overpriced merchant account and a templatized website 01:38:29 then later dropshipping on a small catalog 01:38:54 something like amway ? 01:39:06 i've actually never researched amway.. 01:39:09 it's not MLM.. 01:39:22 They buy leads (or generate), have a salesmen pitch the people into $99 a month for 4 years 01:39:40 after 2 months or so the people can't back out, the leasing company then sticks with it, or sells the paper to a collections agent (aka satan) 01:39:50 a little about amway here -> http://www.skepdic.com/ 01:39:56 Satan then sticks someone like hitler to squeeze the lemon 01:40:17 so $2k for a merchant account (cost of goods, $50) and a template website 01:40:25 template websites of course were the easiest thing i made 01:40:46 and the hardest? 01:41:09 hehe, depends on the growing complication 01:41:44 on the scam front, at the end, i made an online system to allow people to manage a full product catalog, apply various filters to product images, and use WYSIWYG for the page editing (internet explorer has had the ability for a long time) 01:42:01 that was somewhat difficult, but nothing too mind bending 01:42:23 how is it you are still able to survive? 01:42:30 where are the clients coming from? 01:42:37 i walked away from that project 01:42:51 satan demanded my soul, so i walked out 01:43:05 now I manage leads, from generation to distribution 01:43:13 yeah, he can be a little sonofab* sometimes that satan. 01:43:17 businesses will be buying leads for quite some time 01:43:36 leads? 01:43:52 and that's been my most time consuming project. taking arbitrary data, storing it, making it searchable by any criteria (and at the hands of a non technical operator) 01:44:02 shipping it back out is also trivial 01:44:04 leads.. 01:44:10 glengarry glenross? 01:44:13 boiler room? 01:44:13 ah 01:44:16 remember leads :) 01:44:29 both. yes. now. :) 01:45:06 all types of leads, mortgage, insurance debt consolidation, merchant accounts, etc etc. 01:45:19 so.. marketing :P 01:45:56 i unify the sources, allow a lead to be transfered from a prospect to an office ready to sell them something, in under 5 minutes.. 01:46:00 and take a cut 01:46:16 it just so happens that the costs are negligable and the machine takes care of the work. the margins are wonderful 01:46:28 and i have always had a perverse attraction to marketing 01:46:31 aren't you worried about competition? 01:46:50 http://labs.google.com/glossary?q=Qualified%20Leads 01:46:54 i would be 01:46:58 except that i'm just that good 01:47:09 haven't you noticed competition in software design to be rather abysmal 01:47:27 obviously if i went on vacation for 2 years i'd lose ground 01:47:46 at present i'm far ahead of everyone we've encountered, and i've done it single handedly 01:48:23 most geeks are really bad at the business aspects of business. 01:48:39 many of them never see the light of day, and managers can't distinguish the valuable ones 01:48:46 the others are weighed down by beaureocracy 01:49:06 and many more have succumed to tunnel vision 01:49:43 it's the four solid years of programming (and little else) that led me to Forth -- i've yet to come across anyone else as interested 01:49:53 anyway, it's just a big mess out there 01:50:07 anyone else as interested in what as who? 01:50:16 any web programmers 01:50:21 ah. but why forth? 01:50:34 forth isn't doing anything php is currently doing! 01:50:44 hah 01:50:56 to me, php is just functions tailored for web work 01:51:13 --- join: PoppaVic (~pfv@s63.waters.gtlakes.com) joined #forth 01:51:14 other than that, it's a bloated interpreter that runs on top of a bloated os to perform trivial tasks 01:51:32 you can pull up a "why forth?" essay anywhere on the net to see the reasons I like it 01:51:34 --- quit: karingo () 01:51:46 how did you hear of forth? 01:51:50 randomly? 01:51:54 yes 01:52:00 its domain seems to be embedded systems programming I think. 01:52:02 chuck moore interview on slashdot i believe 01:52:07 ah okay. 01:52:10 that is its sole domain 01:52:51 mornin' 01:52:58 I know that i can take a typical company with 10 high powered servers and stick all their webservices easily on a single machine, provided their applications can be properly adapted 01:53:01 hi PoppaVic. 01:53:24 hmmm I'm beginning to see where you're going klaw. 01:53:40 95% of the code I do is just the same thing over again tweaked a little. most of the services out there are bogus and will be obsoleted, I hope within the next 2 years 01:53:43 hi gilbertdeb... I gotta' get up at 0500 to catch conversations now? ;-) 01:53:46 though you know how these things can turn out 01:54:04 PoppaVic: heheh. 01:54:08 aside from that.. all of the hype that people have found.. p2p networks, distributed computing, have all been easily available for a decade 01:55:02 so you have intentions to stick forth in there somewhere? 01:55:08 What I walk into, gilbertdeb? 01:55:25 well i'd like to bootstrap forth to a database to start with, because I don't like the existing procedural languages.. 01:55:37 PoppaVic: klaw is a webprogrammer and he sees forth as a better tool. 01:55:42 you can fit a webserver into something the size of a quarter.. 01:55:44 ahh. 01:55:46 ... for php stuff. 01:56:13 there's no reason to have a big OS and all of its utilities to serve html and process data streams 01:56:16 but Klaw what about scalability and such? 01:56:25 server-side... I see. 01:56:46 scalability of what? 01:56:58 'morning 01:57:04 hi Speuler. 01:57:08 lo 01:57:10 clusters are all based upon minimalism for the sake of efficiency.. 01:57:10 Klaw: of servers and such. 01:57:29 * Speuler rubs eyes 01:57:43 are you people still awake, or already ? 01:57:51 are you talking raw processing power or a farm of servers (akin to google) 01:58:02 I couldn't sleep, so.. I got up & serious again ;-) 01:58:06 * gilbertdeb 's been up since 2am. 01:58:15 Klaw: farmwise 01:58:32 there's no difficulty in creating a cluster 01:58:43 as for processing power we'll depend on Saint Gordon Moore for that. 01:58:50 Klaw: forth cluster ? 01:58:54 Klaw: even with forth? 01:59:01 well look at the x25 01:59:08 that's an example of a cluster on a chip 01:59:14 stack is not ideal to distribute code across several machines 01:59:23 distributing code is a piece of cake 01:59:29 one machine sends a stream to the other machine 01:59:37 there's nothing more to it 02:00:08 Klaw: next machine waits until results have been produced ? 02:00:31 Speuler: that all depends on the application, but i find it hard to believe you would thing clustering to be nontrivial 02:00:43 take a farm of webservers.. 02:00:58 Klaw: ok. that would do 02:01:03 you have one machine that handles requests from a client, routes it to an available box, that box processes and sends the results back to the client.. 02:01:32 distributed computing shows a large scale effort with a variety of platforms 02:02:01 anyway, it's not really an issue imho 02:02:04 IB 02:02:33 but having several machines work on code which works on one set of intermediate data seems difficult to me 02:02:58 give an example.. 02:03:34 it's the same as having more than one processor at your disposal -- programming for the x25 would be the same 02:03:51 ramnull: did you manage to read through it? 02:04:03 delegate the work, if it can't be split into pieces, then you're limited in the amount of delegation you can do. 02:04:13 that's obvious 02:04:28 * gilbertdeb throws in "use xml for data" 02:04:37 eww 02:04:40 * gilbertdeb crawls back to his corner. 02:04:41 gilbertdeb: Yeah. Good article. Some good points. 02:04:49 on the internet, XML was the right way to send data 02:05:03 Klaw: was??? 02:05:09 it is still in its infancy! 02:05:09 you have so many species of computer, something plain and easy to process in many languages makes a lot of sense 02:05:18 XML is just 02:05:27 Klaw: create foo 10000 allot : bar ( x -- ) 10000 0 do dup i * foo i + c! loop drop ; 02:05:33 gilbertdeb: I'm not sure I'd want him in charge of writing a flight control system, but for most other computer fields his reasoning stands. 02:05:55 Speuler: why would you need to distribute that, it's so small. 02:06:02 anyway, you know what threading is 02:06:13 Klaw: create foo 100000000 allot : bar ( x -- ) 100000000 0 do dup i * foo i + c! loop drop ; 02:07:30 i haven't taken many mathematics courses 02:07:39 problem is see is, logic flow and data flow are often matched in forth, with the result that one operation needs the results of the previous one 02:07:48 ah, tunnel vision 02:08:00 or "pipelining",anyway 02:08:14 Speuler: i think you're looking far too closely at this.. 02:08:42 first of all, internet systems generally assume there will be many clients requesting things. so it's already been broken down and easily delegated 02:08:42 ramnull: why not flight control systems? 02:08:58 Klaw: distributing code on a say, forth multitasker level would probably be easy 02:10:51 ramnull: I have a question on that.. my ethics professor said that there is a known bug in autopilot that with some astronomical chance the plane could nosedive 02:10:51 Klaw: but majority of programs just use a single thread 02:10:51 which programs are you thinking of 02:10:51 the majority of programs i deal with are multithreaded 02:10:51 however the majority of small programs i write are of course single threaded, but can be run with different parameters simultaneuosly.. 02:10:51 gilbertdeb: Well, he seems to tilt towards the a more freewheeling philosophy, which creates an environment where bugs are more likely. 02:10:52 ramnull: i wouldn't want to program flight control systems 02:10:56 that's not my idea of a good time 02:10:58 Klaw: there's no equivalent of threads in forth, i guess forth tasks are the closest thing to threads 02:10:59 ramnull: I see. because he hates static languages right? ;) 02:11:14 gilbertdeb: Nope. 02:11:21 Speuler: running the same program twice 02:11:35 lets say 1000 people want to search a database 02:11:43 all within the same few seconds 02:11:51 that can be split up at least a thousand ways 02:12:23 we're obviously talking from seperate ends of the spectrum 02:12:55 gilbertdeb: When the risks go up(i.e. lives are on the line) the margin for error tends to drop. 02:13:29 gilbertdeb: i.e. The code hits a bug and people die. 02:13:37 ramnull: i don't want to be the twat responsible for a bomb missing the mark by a hundred meters 02:13:41 that's not my gig 02:14:16 Speuler: back to you.. you're asking me.. how do you take a number to the 10000th power and return the result using more than one machine 02:14:20 Klaw: your overseer will report 'colateral damage'. don't worry too much about it. 02:14:36 Klaw: not quite 02:14:48 i was just multiplying index with top of stack 02:14:52 Klaw: you have to have a parallelizable algorithm for that to work. 02:14:57 and initialize that array 02:15:05 gilbertdeb: But you can get away with stuff like that when it comes to most other types of computer systems. 02:15:17 the search for prime numbers, decryption, encryption.. 02:15:19 gilbertdeb: So mostly I agree with him. 02:15:31 all of this can be distributed and has been to various systems around the internet.. that's all i was getting at 02:15:41 --- join: karingo (karingo@188.portland-11-12rs.or.dial-access.att.net) joined #forth 02:15:55 anyway, i'm not quite sure what we're debating/discussing at the moment 02:16:00 Klaw: i was thinking of compiler intelligence, aiding distribution 02:16:11 oh.. well you can easily create words for this.. 02:16:18 individual machines can ask for more units of work 02:16:22 as they do with SETI, etc 02:16:32 those are of course overly complicated exampels 02:16:50 Klaw: which would probably be easier if the dependancy on input parameters for a word was less localized 02:16:55 but as i said, take moores new chip.. think of how you'd program for it.. it's the same idea. 02:17:07 x25? 02:17:20 gilbertdeb: yes.. 02:17:36 it's not a monolithic piece 02:17:51 is it yet available for purchase? 02:17:58 i don't think it exists in a physical form 02:18:21 bbl. Gotta go hack. 02:18:23 --- quit: ramnull ("This isn't Happy Hour!") 02:21:05 would it be correct to state that mimd would be more useful on a thread/task level, and simd on a primitive level ? 02:21:56 terminology that i have to use google for.. 02:22:08 mimd as in, a spawned web server, and simd, as in , say, and associative array 02:22:17 yes, i'd imagine so 02:22:24 Klaw: simd: single instruction,multiple data 02:22:34 mimd: multiple intruction, multiple data 02:23:19 google glossary allows me to appear more informed than I am 02:23:45 Klaw: dict is also quite good for acronyms 02:23:53 anyway, this is all heavy computer science philosophy 02:24:20 From The Free On-line Dictionary of Computing (12 Sep 2002) [foldoc]: 02:24:20 MIMD 02:24:20 02:24:20 {Multiple Instruction/Multiple Data} 02:24:21 but i can't imagine there is any debate in it. if you can delegate, and you have somewhere to delegate to, it's faster than sticking the task next in line 02:24:33 While single-processor vector computers are able to operate in MIMD mode because of overlapped functional units, MIMD terminology is used more generally to refer to multiprocessor machines. 02:25:29 matrix operations would tend to be simd 02:25:40 --- quit: karingo () 02:25:40 web servers mimd 02:25:59 alright.. 02:26:34 you'd use something like simd for running the same routines on different data 02:26:42 like distributed cracking, seti, etc.. 02:26:44 right 02:26:51 you'd use mimd with specialization 02:27:20 or, not synchronized processing of the same code base 02:27:58 allowing for different control flow paths 02:29:08 one program running multiple times, vs multiple programs running multiple times.. 02:29:13 is that an adequate analogy 02:29:27 keep in mind my idea of programs is different than yours.. you're at the level of hardware 02:29:28 i guess so 02:29:46 well, something to add to my vocabulary :) 02:29:53 no not quite 02:30:04 one program, running once 02:30:11 at least, appearing to run once 02:30:29 but working on a wider data set 02:30:30 sisd 02:30:40 http://carbon.cudenver.edu/csprojects/csc5809F99/definitions.html 02:30:51 that's nice and simple 02:31:22 with SIMD you can chunk the data up, send it off to clones of a program and they come back with results. there can be a bottleneck if all results are needed to continue.. 02:31:36 SISD.. you have one piece of data, one entry point.. can't delgate it 02:31:55 am i close? 02:32:24 i'm not sure about your sisd statement 02:32:50 that's where compiler intelligence would come in 02:33:07 i love the internet. beats the hell out of any compsci class out there 02:33:15 free lessons on IRC :) 02:33:36 but you don't know whom to believe :) 02:33:56 just like the real world 02:33:59 and any compsci class 02:34:07 well, neithe you don't know whether you can trust in what your teacher says 02:34:12 ;-) 02:34:51 both school as well as irc is opiniated 02:35:05 people are opinionated 02:35:21 unfortunately, science still suffers. 02:35:23 but that's what you got your mind for 02:35:43 i try to imagine what my classes (i'm in orange county, california) would be like in the deep south. 02:35:49 like bob jones university 02:36:21 that's where people run around, stating that they are not crooks ? 02:36:32 *grin* 02:36:42 that's where people shit a brick when someone with dark skin touches someone with light skin 02:37:00 "nooo touchy" 02:37:08 bob jones, isn't that a christian missionary school? 02:37:10 that kind of thing reaches deep into their scientific classes 02:37:50 like the scientific proof that black people are inferior.. versus the real science, where race is scientifically non-existant 02:38:02 icky stuff 02:38:16 Klaw: are you referring to the bell curve book? 02:38:19 riight 02:38:41 i worry that too many of our leaders were tought "science" by pseudoscience 02:38:48 the bell curve measures iq, and iq measures ONLY IQ. 02:39:08 I mean it uses iq measuers for its data. 02:39:30 gilbertdeb: it's an example, but just an example. 02:40:10 take a look at this short piece sometime -> http://www.skepdic.com/iqrace.html 02:40:15 many still deny that differences in IQ, crime rate, etc are a result of economic conditions and instead point at inferior genetics and culture 02:40:28 Klaw: it is much easier that way. 02:41:21 it would be easier to ask for economic class than "race" 02:42:12 generally meaningless, but you have to see that 'economic class' is more abstract than race, which is immediately visible to some extent. 02:42:26 that's not true.. 02:42:34 i strongly disagree 02:42:57 obviously someone making $1k /month in the US is worse off than $1k USD in india 02:43:08 but that's something you can scientifically measure 02:43:19 what are you going to do with race.. compare skin tone and facial proportions 02:43:31 "Are you on wellfare" 02:43:33 "Are you black" 02:43:36 take a pick there. 02:44:56 economics is a science 02:45:00 bs. 02:45:06 expand 02:45:06 economics is a pseudoscience. 02:45:09 hah 02:45:35 it relies a lot on statistics for its 'scientific validity'. 02:45:36 Economics is chaos... created by the irrationality of man. It will make the world end and stuff 02:45:43 you're both generalizing 02:45:50 Klaw: I'm not at all. 02:45:55 well i'm going to have to disagree 02:46:07 Klaw: before the 1900's, there was no such thing as economic science. 02:46:14 nor was there computer science 02:46:16 all that existed was politicaleconomy. 02:46:18 nor much in genetics 02:46:21 TINSTAACS 02:46:31 there is no such thing as computer science yet. 02:46:44 gilbertdeb: i can agree a little more on that 02:46:46 TINSTAAES 02:46:52 but economics has some very strong foundations 02:46:58 Klaw: like what? 02:47:00 statistics? 02:47:09