00:00:00 --- log: started forth/04.07.21 00:00:25 ASau: change made. ForthCode, Network, "UDP over SLIP" 00:02:46 --- quit: Quozl ("think of it as evolution in action") 00:16:47 Dobre utro 00:27:52 Guten Morgen! 00:28:20 Danke schön. :) 00:28:34 * Robert continues writing an inline BASIC compiler for his Forth. 00:29:38 I work on GSL. 00:29:59 Random number generators and simulated annealing. 00:30:21 At the end of the week I plan the first release. 00:31:00 Nice, for which Forth (or other language)? 00:31:15 Gforth initially. 00:31:24 I'm on BSD machine now. 00:31:30 Ah, OK. 00:33:31 I've found several bugs in original GSL. 00:35:02 They're so stupid, it makes me think C programmers have no mind at all. 00:35:15 Hehe 00:35:19 * Robert hides.- 00:35:35 They find x modulo y with (x - (x/y)*y) 00:36:36 Heh. 00:36:39 And that's instead of (x%y)! 00:36:55 how else do you calculate mod? 00:37:10 oh, in C it's stupid :) 00:37:38 It looks they suggest finding modulo with 1 division, 1 multiplication and 1 subtraction. 00:37:51 And that's instead of 1 division! 00:38:45 Another example: 00:39:31 "if(a) x+=b+a; else x=b;" 00:39:40 "if(a) x=b+a; else x=b;" 00:40:06 I like this too. 00:40:24 that's rediculous 00:40:36 One has guessed that he can write "a" instead of "a!=0". 00:40:56 that's the same as "x=a+b" 00:41:13 surely that's not what they meant... 00:41:40 But it looks he can't guess more. 00:42:37 More. 00:42:53 x can't be zero. 00:43:12 By default, a=0. 00:43:27 x can't be more than 2^32-1 00:43:33 Code: 00:44:00 "if (a=0) a=1; x = a & 0xff...ff;" 00:44:13 bad bad bad 00:44:27 --- join: arke (arke@melrose-251-251.flexabit.net) joined #forth 00:44:35 Terve, arke! 00:44:37 hii 00:44:39 !!! 00:44:40 :) 00:44:55 ASau: Nice code, all the way. 00:44:57 Hi ASau 00:44:59 er 00:45:00 Hi arke 00:45:06 :) 00:45:13 if(a=0) is stupid 00:45:14 what is up, fellow f0rthz0rs? 00:45:41 common mistake, but it shouldn't make it into any published code 00:46:01 Herkamire, let a=2^32; 00:46:18 ASau: huh? 00:46:25 a<>0 => it passes first operator unchanged. 00:46:39 Herkamire: Reminds me of a guy I told to use -Wall when he did stupid mistakes like that. Next time he came crying I saw that -Wall actually reported the bug, BUT, he was too damn lazy to reas "all that output" (the warnings). 00:46:51 2^32 and (2^32-1) = 0, so x becomes zero. 00:47:08 ASau: what language? 00:47:14 doesn't look like C anymore 00:47:19 What the hell they use bitwise operation for modulo? 00:47:50 Herkamire, C doesn't allow power, I use ^ for brevity. 00:47:59 Sorry, it conflicts with xor. 00:48:01 ok 00:48:18 but 2^23 and 2^32 - 1 are not the same 00:48:33 ASau: bitwise can sometmimes be faster than modulo. Like for x % 4 its faster to do x & 0x3 00:48:51 arke, that's initialization routine. 00:49:09 Why this programmer does care of speed in it? 00:49:39 He makes so stupid mistake, that he better used modulo everywhere in his code. 00:49:42 Besides, ANY compiler optimizes %4 to &3 00:49:51 Before he begins to optimize. 00:50:03 the optimizer should take care of translating "%4" to "&3" 00:50:10 Even I use % 4 these days, for readability. 00:50:21 Agree, it's the problem of optimizer. 00:51:22 should I eat something and work? or should I go to bed? 00:51:41 One defines 0xff...ff constant (damn thing, better calculate it with 1<<32-1). 00:52:19 ASau: what?? 00:52:35 And he uses it for taking modulo. 00:52:41 you want 0x8fffffff 00:52:48 ? 00:52:59 Code is like this: 00:53:17 #define X 0xffffffff 00:53:38 Or better, define it as -1 00:53:50 or, define it as -1 00:53:52 DAMMIT 00:54:03 Unless you're not sure of the int length... 00:54:19 If programmer thinks he would have 64 bit CPU, 1<<32-1 is better. 00:54:40 Yes. 00:55:09 But if he has a 16-bit CPU, he will run into trouble. :) 00:55:11 If he suppose, he won't have more than 32 bits, why care? 00:55:55 (BTW, just not far away, there's remark: "this does work on 32 bit CPU.") 00:55:59 what exactly are you trying to do? 00:56:23 arke, it's GNU SL code. 00:56:50 I'm rewriting it, with simplification and debugging, into Forth. 00:56:57 ...in Forth. 00:57:38 SL? 00:57:51 GNU Scientific Library. 00:58:10 ok 00:58:28 I work on random number generators and simulated annealing now. 00:58:43 Hmm... a GNU library contains THAT code? 00:58:59 * Robert quickly loses faith in humanity 00:59:08 Actually, what I'm interested is sim. annealing. But it requires RNG. 00:59:29 Yes, Robert, right. 00:59:36 This is in GNU SL. 01:00:23 BTW, more facts. 01:01:17 After I've seen this (x-(x/y)*y) for (x%y), I had to refer to C std. for clarification of division meaning. 01:01:39 It was nice. 01:01:44 Standard defines: 01:02:29 for all x and non-zero y, x == (x/y) + (x%y). 01:02:50 And sign of (x%y) is implementation defined. 01:03:05 Nice, isn't it? 01:03:32 So x%y isn't really defined in the standard? 01:03:43 I have to care whether division is symmetric or whether it's floored. 01:04:57 In our words, if C's "/" and "%" correspond to SM/MOD or FM/MOD. 01:05:35 Robert, there's a remark in footnote 01:05:54 It looks like "see figure 1". 01:06:18 "See FORTRAN std. ISO/IEC nnnn:nnnn" 01:07:26 "We don't care how your compiler performs division. But if you want more clarity, suppose quotinent is rounded towards zero." 01:07:34 :) 01:07:38 :] 01:07:42 :( 01:08:19 :/ 01:08:41 Hehe. 01:08:44 Robert: you guys have Traumschiff Surprise in your ovie theaters soon? 01:08:48 Enjoy dividing, ASau! ;) 01:08:56 arke: ...what? 01:09:30 ...nevermind 01:10:25 I've got sad idea. 01:10:48 Now I need to check how Gforth performs division. 01:13:12 --- join: qFox (~C00K13S@cp12172-a.roose1.nb.home.nl) joined #forth 01:14:05 Hi qFox 01:14:12 lo 01:15:00 hi qfox 01:15:12 hey arke, back from vacation? 01:15:25 God dag, qFox! 01:15:43 and hello asau, and everybody else :p 01:17:33 qFox: no, not yet 01:17:36 I saw you made some progress 01:17:40 ..while I was sleeping. 01:17:50 yep. its looking smooth now 01:17:58 :) 01:19:16 Hm. Gforth does right. 01:19:32 Congratulations, forthers. 01:19:56 You've got obvious reason why Forth is much better than C. 01:20:00 :) 01:20:11 Because GForth is more standardized than C? 01:20:30 That's not really a proof, but I'll throw it in the face of the next C coder that complains at Forth. 01:20:45 wait, coders complain about forth? 01:20:50 Because you can define your preferred way to divide numbers. 01:21:05 qFox: No, I made that up. 01:21:21 And that utilizes the same notation as usual. 01:21:23 ah, cheat. 01:22:00 ok perhaps my while isnt working perfect just yet :p 01:23:59 Hehe 01:24:24 qFox, you've missed the whole story. 01:24:34 oh oki :) 01:24:46 Read log. 01:27:10 Robert> what do you need ?0branch for? 01:27:37 which branchings, i havent used them but i have defined all the basic branch types, so either i made a booboo, or.. 01:28:22 "if", for example. 01:29:00 oh, ok then what do you need 0branch for? :) 01:29:23 ?0branch you mean? 01:29:41 well, no, if you use ?0branch for if, what would you use 0branch for 01:30:04 There's no 0branch, is it? 01:30:21 guess i'll rename it then 01:30:41 I thought you were refering to F 01:31:07 in general 01:32:56 and my while is working properly after all. yay. 01:33:52 Uhm, you should know by now that the words in F aren't very ... standardized. ;) 01:34:03 good point 01:34:28 now i have if, i can finally implement a signed . 01:36:11 Hehe 01:39:29 heh 01:44:21 : . dup $8fff > if '- emit s>u then u. ; 01:44:24 (probably wrong) 01:44:35 * Robert continues his "inline BASIC" feature. 01:46:05 In F it would be : s. dup 80000000 and if 2d emit neg then . ; 01:46:13 Yes, s. is signed and . is unsigned. :P 01:46:31 not anymore 01:46:34 ;) 01:46:39 * Robert cries 01:46:42 Traitor. 01:46:51 phoey 01:47:02 i've turned your F into a sweet looking forth 01:47:03 :p 01:47:28 arke: s>u ?? 01:47:59 why not "-1 *" 01:48:06 or is that what it does? 01:48:43 Herkamire, why not MINUS then? 01:49:02 there's probabyl a word for it. 01:49:05 Or how's it in ANTI? 01:49:15 INVERT ? 01:49:25 --- quit: arke ("Lost terminal") 01:49:34 grrr. I can't think of the word for it 01:49:44 negate 01:49:52 ding! 01:50:10 Dobryjj den', fridge! 01:53:13 : s. dup 2147483648 and if -1 * 45 (emit) . else . then ; 01:59:09 Borosh --- Niederreiter RNG works! 01:59:15 --- join: crc (crc@0-1pool88-47.nas48.philadelphia1.pa.us.da.qwest.net) joined #forth 01:59:26 Dobryjj den', crc! 01:59:35 Good morning ASau 02:01:00 --- join: arke (arke@melrose-251-251.flexabit.net) joined #forth 02:01:45 Good morning arke 02:01:52 gr33ts 02:02:55 I'm preparing to start work on a new native version of RetroForth 02:04:24 This should be much more interesting than the RetroForth/Native 6.x is 02:06:21 qFox: what's (emit) ? 02:06:40 qFox: this is silly: " . else . then" just do "then ." 02:06:51 core word 02:07:01 using F's emit for now 02:07:03 qFox: how is it different from emit? 02:07:19 nothing. it just indicates for me that it is a code word 02:07:27 ok 02:07:39 i'm building this forth on Robert's forth 02:07:55 I see 02:08:00 so some of my "code" words, are simply words from his forth, at least for now 02:08:14 ahh 02:08:24 output stuff will be changed later, but some words wont. like core math words etc 02:09:22 and you're right about the then . 02:11:28 crc, read logs of one or two hours age. 02:12:03 I wonder, how you find that. 02:12:10 crc that's great :) 02:12:16 It is interesting. 02:12:33 crc cool 02:12:38 crc: I'll play with it under bochs when you get it to an interesting stage :) 02:31:22 --- quit: arke ("Lost terminal") 02:32:38 --- quit: crc (Client Quit) 02:34:05 lol. i should be more carefull when changing a word :p 02:34:13 i wanted to make . signed 02:34:21 so i hooked . to s. 02:34:23 but s. calls . 02:34:31 so it ended up in an endless loop :p 02:34:41 (the code word is (.)) 02:46:16 heh 02:46:23 --- quit: Herkamire ("bed") 02:52:15 How is it better: "random", "rand" or "rnd"? 02:52:29 word that produces a number? 02:52:34 Yes. 02:52:35 * Robert likes "rnd" 02:52:38 Short. ;) 02:53:00 random or rnd. i would go with random, since rnd could be round 02:53:14 and rand is Cish :p 02:54:22 Heh. 02:54:54 hm. what to do next 02:59:51 Code some BASIC? 03:00:48 die. 03:03:18 OK, brb 03:04:29 Two screens. 1 for description, 1 for code. 03:04:37 How do you like that? 03:04:53 And 1 additional for adapted code. 03:05:02 Not bad. :) 04:26:19 --- quit: kuvos (Ping timeout: 14400 seconds) 04:30:08 --- join: crc (crc@0-1pool176-14.nas6.philadelphia1.pa.us.da.qwest.net) joined #forth 04:30:19 Back :-) 04:32:18 ASau: What were you saying about reading the logs? 04:34:25 --- join: kuvos (C00K13S@cp12172-a.roose1.nb.home.nl) joined #forth 04:35:41 Hello kuvos 04:36:25 thats just me :p 04:36:39 Guten Abend, kuvos! 04:37:49 Two nicks?> 04:38:20 crc, conversation has began at 11:30, now it's 3:40. 04:38:23 * crc is sorry for suddenly leaving -- family emergency 04:38:28 yep 04:39:06 Why? 04:39:40 ? 04:40:18 i dunno. that computer is grepping words from a few irc chans on freenode for a random ai chatter bot project 04:40:41 :-) 04:40:50 Ah! 04:41:30 So who is the real man of you? 04:46:32 that would be moi 04:46:49 that computer is my server 04:46:55 for internet and other stuff 04:49:59 * crc is cleaning up the hdd driver right now 04:55:17 Looking back at this code, I'm suprised it actually works as well as it does 04:58:08 All PIO, no real error checking... 05:00:30 98 lines of assembly + Forth code 05:00:51 I still need to add a fake LBA mode to it though 05:02:57 This will be an interesting challenge to work into the new kernel... 05:03:33 Oh well, time for bed. 'night all! 05:03:59 --- quit: crc ("http://www.retroforth.org/dev/beta_releases/retroforth-7.beta4.tar.gz (now works with BeOS!)") 07:10:02 --- join: Topaz (~top@spc1-horn1-6-0-cust117.cosh.broadband.ntl.com) joined #forth 07:15:58 Dobryjj vecher, Topaz. 07:19:32 er, hello to you too :P 07:25:48 --- join: warpzero (~warpzero@mi245.dn183.umontana.edu) joined #forth 08:28:37 --- join: arke (arke@melrose-251-251.flexabit.net) joined #forth 08:28:53 hi 08:28:56 Hi 08:29:54 uum 08:29:58 Robert 08:30:13 dunno if you care 08:30:19 but i did some experiments 08:30:25 and the Windows 95 memory manager kicks ass 08:30:39 but the Windows 98 and up one sucks 08:32:04 Uhm... 08:32:25 Too bad everything else in Win95 sucks. 08:32:33 what!? 08:32:33 What brought you to this conclusion? 08:32:40 explorer.exe in win95 is way better 08:32:48 way faster too, because it doesnt include ie 08:32:54 oh 08:33:12 i ran programs and the system monitor and saw what it did 08:34:46 and now i must go 08:35:18 --- quit: arke ("btw, Win95 is MUCH better than Win98! Mail me for info why (ask in your chanel for address)") 08:36:17 Heh. 08:53:09 --- quit: warpzero (Remote closed the connection) 08:58:54 --- join: warpzero (~warpzero@mi245.dn183.umontana.edu) joined #forth 09:08:27 --- quit: warpzero (Remote closed the connection) 09:15:27 --- join: warpzero (~warpzero@mi245.dn183.umontana.edu) joined #forth 09:30:24 --- join: Serg[GPRS] (~z@193.201.231.126) joined #forth 09:33:26 Privet, Serg[GPRS]! 09:33:36 Kak dela? 09:34:38 bad as volcano a..hole 09:34:56 Esli khochesh' ubojjnuju veshh', mozhesh' gljanut' logi za segodnja s 11:30. 09:35:34 tomorrow from work - GPRS is too xpensive 09:35:45 so what is the thing ? 09:36:07 Well, read and you'll realize. 09:36:18 It's from my current project. 09:36:34 heh.. log URL? how big is it ? 09:37:07 Today it was quite. 09:37:14 Somewhere at tunes org. 09:37:34 ok, i'll look 09:38:11 Google for "dobryjj vecher". 09:41:09 i bought a matchbox cam for 677 rub, 640x480, 26 shots 09:41:21 667 rub ;) 09:50:55 * Serg[GPRS] listens for PainKiller muzik from June GameMania magazine 09:55:27 * Serg[GPRS] suffers from lack of sleep 09:55:41 gonna shoot off some heads in FarCry and then Zzzz 09:55:46 bye! 09:55:49 --- quit: Serg[GPRS] () 10:06:52 question. does parse by default to " 10:09:15 i have a word now that parses ( c -- c-adr u ) 10:09:27 where c is the char to use as delimiter 10:42:12 Yeah, that's what I use 11:18:48 --- join: ASau_ (~root@217.16.31.100) joined #forth 11:18:48 --- quit: ASau (Read error: 54 (Connection reset by peer)) 11:33:22 --- join: ASau (~root@217.16.31.100) joined #forth 11:33:22 --- quit: ASau_ (Read error: 104 (Connection reset by peer)) 11:49:08 --- join: ASau_ (~root@217.16.31.100) joined #forth 11:49:08 --- quit: ASau (Read error: 54 (Connection reset by peer)) 12:21:56 --- join: tucknip (~Huck@170.141.109.33) joined #forth 12:30:28 --- quit: ASau_ (Read error: 60 (Operation timed out)) 12:41:23 --- join: ASau (~root@217.16.31.100) joined #forth 12:42:53 --- part: tucknip left #forth 13:26:28 --- join: Herkamire (~jason@h000094d30ba2.ne.client2.attbi.com) joined #forth 13:30:41 --- join: Sonarman (~matt@adsl-67-113-235-152.dsl.snfc21.pacbell.net) joined #forth 13:30:59 mur: are you here? 13:31:45 yeps 13:33:31 i remember yesterday you were asking about a forth for macintosh with gui support, and someone suggested mops. here's a nice example of mops code: http://groups.google.com/groups?dq=&hl=en&lr=&ie=UTF-8&group=comp.lang.forth&selm=BD23B049.1E93%25dhoffman%40journey.com 13:36:55 --- quit: Sonarman ("Lost terminal") 14:46:54 --- join: tathi (~josh@pcp02123722pcs.milfrd01.pa.comcast.net) joined #forth 14:54:03 --- quit: ASau (Read error: 104 (Connection reset by peer)) 14:56:32 --- join: ASau (~root@217.16.31.100) joined #forth 15:01:32 --- join: ayrnieu (~julian@205.241.56.30) joined #forth 15:51:54 --- join: blockhead (default@dialin-36-tnt.nyc.bestweb.net) joined #forth 15:55:28 hi blockhead 15:57:46 hey slava; wahat's up 15:58:01 i'm stuck on a little problem 15:58:12 me too :D 15:58:20 or is that "to"? 15:58:39 --- join: doublec (~doublec@coretech.co.nz) joined #forth 15:58:50 hi doublec 15:58:51 --- quit: skylan_ (Read error: 104 (Connection reset by peer)) 15:59:15 hi slava 15:59:29 doublec, i'm a bit stuck on the | parsing word 15:59:55 what does '|' do? 16:00:32 [ 1 | [ 2 | [ 3 ] ] ] == [ 1 2 3 ] 16:00:54 ok. 16:01:06 So a light weight syntax for consing lists. 16:01:19 well its really a way of having a non-list cdr 16:01:20 [ 1 | 2 ] 16:01:36 yes 16:02:26 anyway, i'll add 'catch' to httpd first. 16:02:33 isn't there ',' for that? 16:03:13 , is the old syntax 16:03:46 '|' does look better. 16:03:59 | also follows prolog syntax, there. 16:04:13 Should I add the 'contrib' directory to cvs and add the continuation responder? 16:04:56 if so, where in cvs? On the same level as 'library' or inside 'library'? Or one level up? 16:06:10 same level as 'library'. 16:06:27 ayrnieu, in practice I only use conses which are not proper lists in a handful of places. 16:06:33 ok, I'll add it. 16:07:12 thanks 16:08:13 there's still lots of good stuff I plan to add to it. It's useful even as it is but it's definitely going to get much better. 16:08:20 great! 16:08:53 --- quit: ASau (Read error: 110 (Connection timed out)) 16:09:02 for now, the httpd will still die on an error, but pending sockets will close. 16:09:42 That's good. 16:10:02 later, if the error is in the responder itself and not i/o, it will serve an error page. 16:10:07 i'm not sure how to realize this yet though. 16:10:18 A top level catch around the httpd caller could restart it though? 16:13:18 --- quit: warpzero ("Tried to warn you about Chino and Daddy Gee, but I can't seem to get to you through the U.S. Mail.") 16:13:41 doublec, you could have a 'catch' in your responder 16:13:44 it doesn't have to rethrow 16:13:52 yes, good point. 16:17:00 doublec, there is a very nice factor mode in jedit cvs 16:17:11 i should update the one on the web page 16:20:11 cool, I'll try it out. 16:20:27 ok i've uploaded a new factor.xml to factor.sf.net/factor.xml 16:21:34 slava, I've added contrib and cont-responder to cvs. 16:21:41 thanks. 16:23:50 doublec, eventually there will be lisp-style restarts 16:24:10 I like that idea. The lisp restart/condition system is very powerful. 16:26:50 doublec, update cvs. 16:26:50 --- quit: tathi ("leaving") 16:26:54 httpd now uses catch 16:27:03 cool, thanks. 16:31:46 slava, if I use the new httpd and 'stop' or terminate the web browser in the middle of a request I get a 'broken pipe' java exception. 16:31:48 --- quit: qFox ("meh ircii's netsplit detection appearantly sux ^^") 16:31:54 slava, I can't then restart the server: 16:32:01 10] USE: httpd 16:32:01 11] 8888 httpd 16:32:01 ERROR: java.net.BindException: Address already in use 16:33:11 do you have the latest httpd? if i do the same thing here, i get the error too, but the socket closes 16:33:15 try "httpd" see 16:33:55 yes, it has the 'catch' in it. I'll see if I can reproduce it. 16:35:35 slava, display something long like: http://localhost:8888/inspect/global%27vocabularies%27cont%2dhtml 16:35:49 When requesting that page hit refresh a few times in succession fast. and it occurs 16:35:54 ok 16:36:27 or request it and stop the browser before it completes. 16:40:34 doublec, i can't reproduce that. 16:40:45 doublec, if i make it throw an exception, the httpd stops and can be restarted 16:41:39 slava, is it the broken pipe exception? 16:41:50 I'll see if I can produce a test case that does it all the time. 16:41:57 Calling responder inspect 16:41:58 ERROR: java.net.SocketException: Broken pipe 16:41:58 Stacks have been reset. 16:41:58 :s :r :n :c show stacks at time of error. 16:41:58 :j shows Java stack trace. 16:42:00 3] (http 16:42:14 : (http 8888 httpd ; 16:42:16 hmm, strange. I'll track it down. Must be something on my end. 16:43:41 try a responder that does "Foo" throw 16:44:56 aha, found it 16:45:09 slava, Run the httpd word in a thread and that's when it happens: 16:45:13 [ 8888 httpd ] in-thread 16:46:30 A httpd server that restarts : x [ 8888 httpd ] [ x ] catch ; 16:46:32 easy :) 16:46:48 of course it never quits either. 16:46:50 threads? grr. ok i'll add this to the to do list ;) 16:47:05 threads are a pita aren't they. 16:47:06 : x [ 8888 httpd ] [ [ x ] unless ] catch ; 16:47:17 ahh, nice. 16:48:06 I might throw a demo server up on my web server for people to look at to see Factor and the continuation based web server. What do you think? 16:48:25 There's no way they can execute code right? 16:49:02 i can't guarantee there's no security holes 16:49:17 does your todo check for a username of ../../ before creating the file? 16:49:19 etc. 16:49:19 One way to find out. There's nothing on that server for anyway to do anything with anyway. 16:49:40 I won't have the 'todo' example on there just yet. 16:49:49 I'm confident the other ones are fine. 16:53:03 the other problem is DoS attacks. 16:53:08 somebody can connect and never close the socket etc 16:54:12 --- nick: ayrnieu -> nleavrh 16:56:02 that's true. 16:57:05 --- nick: nleavrh -> ayrnieu 16:57:15 doublec, you might've noticed a new { ... } printed form for vectors. 16:57:19 i'll add the reader syntax for this shortly. 16:57:57 yes. That stack prints out like that now too I see. 17:01:00 --- quit: Topaz ("Leaving") 17:31:52 --- join: crc (crc@0-1pool89-27.nas48.philadelphia1.pa.us.da.qwest.net) joined #forth 17:33:19