00:00:00 --- log: started forth/02.03.04 00:18:45 --- quit: qless ("Download Gaim [http://gaim.sourceforge.net/]") 02:04:27 --- quit: rob_ert (Read error: 113 (No route to host)) 02:57:49 good morning 05:48:03 --- join: rob_ert (~robert@h173n2fls33o898.telia.com) joined #forth 05:49:40 hi 05:49:51 EOS? ;) 05:49:57 Hej onetom :) 05:51:51 how was yr day in school? 05:52:15 Fine... I was thinking of forth :P 05:52:23 :) 05:52:33 I've got a little math question...do you think you can answer it? 05:52:58 what r u workin on nowday anyway? 05:53:08 will c, just ask! 05:54:06 You have a mug of coffee, 80C warm. The room temperature is 20C. Every minute, the temp. of the coffe decreses with 20% of the current temperature. 05:54:15 Find a function for the temp. of the coffee. 05:54:44 current temp of what? 05:54:52 The first you think of, is of course T = 60*0.8^t + 20, where T = temperature, and t is time. 05:54:55 Of the coffee. 05:55:47 But then, of you make it a diff. eq., you get T = 60e^-0.2x + 20, which is almost, but not exactly the same. 05:55:52 Wich one is correct? 05:56:51 later seems better, tho 05:57:01 s/x/t/, hm? 05:57:23 uhm, yes 05:57:31 x should be t 05:57:38 sorry :) 06:00:15 meanwhile i have to work and chat, so sorry 4 the large response time 06:00:52 np 06:00:55 can i get the exact, word-by-word transtription of the problem? 06:01:01 I usually have long respone times too... 06:01:19 oh, wait 06:01:22 Why isn't the one I gave you enough? 06:01:40 probably its enough 06:02:35 hmm whats the relation between room temp and mug temp 06:02:55 i dont really understand whats going on 06:02:58 Hehe 06:03:18 Every minute, the temperature gets 80% of the old temperature. 06:03:24 err 06:03:41 80% of the difference between the mug and room remperatures. 06:03:44 if u say: Every minute, the temp. of the coffe decreses with 20% of its actual temp. 06:03:55 I meant this: 06:04:19 it decreases with 20% of the _difference_ between the mug temp. (80C in the beginning) and room temp. (always 20C) 06:04:22 ahhha!!! getting better :) 06:05:17 then, lets write done some elements of this series 06:05:47 t0: T = 80C 06:06:06 again: 06:06:09 t0: T0 = 80C 06:06:39 t1: T1 = T0 * 80% * (T0 - Troom) 06:07:12 t2: T2 = T1 * 80% * (T1 - Troom) 06:07:16 T' = (T - 20) * -0.2 06:07:26 T' = change speed of temp. 06:07:30 T = temp. of mug 06:08:44 (its not a good idea to use T describing a velocity, anyway...) 06:09:00 ? 06:09:05 T is the temperature 06:09:16 But T' is the derived T 06:09:31 but, hey, write a function what depends on time! 06:09:43 That's what we want to get! 06:09:44 oh, ok. i c now 06:09:47 That's the answer :O 06:09:55 s/:O/:P/ 06:09:58 I also have another question, isn't forth very slow compared to e.g. compiled C? 06:10:29 no, not *very* slow 06:10:59 tho, usually slower 06:11:16 *sob* 06:11:37 a good forth implementations speed is not far from native C codes speed 06:13:06 in yr diff.eq. problem - im afraid - i cant help (at the very moment at least) 06:13:21 Okay, np. 06:13:26 I ask in other channels ;) 06:13:35 i havent dealt w such problems nowday 06:14:05 and it seems ive forgotten it :( 06:14:22 hehe ;) 06:14:44 so, thx 4 the question. it has warned me that its high time to refresh 06:14:54 lol :) 06:15:03 the responsible areas of mind! 06:15:04 :) 06:31:39 --- join: I440r (~mark4@1Cust27.tnt3.bloomington.in.da.uu.net) joined #forth 06:31:42 hi 06:32:48 rewriting my syscall interface - simplifying the issue of adding new syscall handlers at extend time 06:32:56 i.e. making it possible :P 06:38:52 hiii! 06:40:16 what do u mean by "extend time"? 06:40:21 hi :) 06:40:57 ok. well imagine the assemble time thing using nasm as a metacompil of the inner core kernel. 06:41:00 out of kernel code? 06:41:09 ah, ok 06:41:22 extend time is when you have that kernel fload all the extentions and fsave out the extended forth 06:41:29 i havent met this expression before 06:41:49 i dont have a meta compiler yet because i dont have an assembler extention yet 06:42:00 where r those fload and fsave derive from? 06:42:09 which expression. extend ? 06:42:16 its not and ANS stuff, i guess 06:42:30 fload is the fpc name for "interpret from source file" 06:42:34 yes, the "extend time" expression 06:42:41 he called fsave save-exe 06:42:53 but in linux executables arent called xxxxx.exe 06:43:00 so i called it fsave 06:43:00 forth save 06:43:04 we have forth load 06:43:08 and forth save :) 06:43:12 and turnkey 06:43:14 but its clear now -- ive compiled it in2 me voc : extend-time ... ; ;) 06:43:22 turnkey is almost identical to fsave 06:43:27 but fsave also saves out headers 06:43:35 heh 06:43:50 vocabulary real-life 06:43:57 only forth also real-life also ? 06:44:04 heh 06:44:19 yes :) 06:44:57 so fload is actually the equvivalent of INCLUDE or REQUIRE in gforth eg 06:45:02 right? 06:45:10 yes 06:45:21 i think include could be an aloas for it 06:45:26 but i realy hate making aliases 06:45:41 fpc made so many aloases for all sorts of words 06:45:48 he overused alias 06:45:59 c 06:46:28 alias should only be used to change the CONTEXT of a word - for instance \ and #! are the same in my forth 06:46:43 but #! has special meaning for the shell interpreter 06:47:16 c 06:47:24 fpc did aliases inside the assember for [bx+si] = [si+bx] [bx+di] = [di+bx] 06:47:25 ugh 06:47:37 DUMB 06:48:37 dumb but simple and effective, isnt it? 06:48:55 who would u implement it instead? 06:48:59 how 06:49:29 i think that the compiler should not be creating a shitload of b ullshit aliases. the programmer should train himself to know that the undefined [si+bx] is exaxctly the same as the defined [bx+si] 06:49:41 the compiler should not cater to the lameness of its users 06:49:44 * onetom "im dyslex and im mad, di-ei-em" :) 06:49:47 mine wont :) 06:50:28 i would define [bx+si] and THAT is what the user should be using. 06:50:30 :P 06:50:49 that is - if i was writing the assembler extention. but i dont think i could write one for x86 06:51:51 well... aint u a bit strict w the users? ;) 06:52:22 do u still use your icq acc? 06:52:34 if ur lame - go code c :P 06:52:45 heh 06:52:45 i Am catering to the needs of the users 06:52:57 but forht isnt supposed to have 50 billion definitions for @ and ! each 06:53:08 yes 06:53:08 22111215 06:53:16 but im not on icq rite now 06:55:47 ive written u msgs last day 06:56:09 ive found an icq num in the readme of yr isforth 06:58:09 yes i remember putting that in there now heh 06:58:17 ah, ive chked, its correct, so u should have got them 06:58:34 ill see if i did when i startx next and run licq :) 06:58:51 it would have been my next question :) 06:59:03 what icq client do u use? :) 06:59:07 now i know 06:59:21 but its a terrible 1 06:59:36 is it based on libicq? 06:59:43 i use vicq 07:00:33 yes - whats vicq ? 07:00:43 its a standalone perl script what utilizes libreadline 07:00:57 i think licq is where libicq started heh 07:00:57 not sure 07:01:00 and it has a simple console interface 07:01:13 ugh its perl based 07:01:25 hehe 07:01:37 i dont think i could handle a perl based icq client. 07:01:48 i mean 07:01:48 a console based 07:01:51 oh, im sure u can 07:02:12 the only thing u have to do is 07:02:31 writing msg nick/messagetext 07:02:53 that start further msg w a simple "a" character 07:03:41 shall i who u an example? 07:03:49 that would suck actually, i hated ircii because all 10 channels are in one window and you had to keep /msg #channel message and /msg nick message 07:03:54 probably in flood? 07:04:01 probably on #flood? 07:04:32 SOME console based clients realy suck - other x based ones suck. 07:04:40 no thats ok 07:04:50 ill stick to licq for now. maybe ill write a ficq one day :) 07:04:56 but what if they use tab-completion? 07:05:03 :))) 07:05:45 but w vicq u only have to append a letter "a" in front of the messages 07:05:50 look: 07:05:51 vICQ(online)/103483577>msg lac/azaz? 07:05:51 [2002.03.04 12:28:24] >> lac: azaz? 07:05:51 vICQ(online)/103483577> 07:05:51 [2002.03.04 12:28:25] filler, ado, meg egyebek, de majus korul all. lesz egy olyan 200-250 rugo uj gepre 07:05:51 vICQ(online)/103483577>a ugh, hat annyi azert nem kell ra semmikepp 07:05:53 [2002.03.04 12:29:08] >> lac: ugh, hat annyi azert nem kell ra semmikepp 07:05:55 vICQ(online)/103483577> 07:05:57 [2002.03.04 12:29:46] tehat akkortajt lenne aktualis egy P4 mondjuk :) Addig nem kell kartya 07:06:25 hehe im running a : foo begin again ; in a console and its using 98% of my cpy on my laptop - this is keeping my room warm :) 07:06:26 certainly it is coloured 07:06:38 :))) 07:06:55 heh 07:07:37 i mean the example ive pasted is coloured in "real life" 07:07:54 so its much more readable 07:08:10 yes - i got that 07:08:27 that what does "heh" mean 4 u? 07:08:49 in hungarian it means "whaaat?" 07:10:14 its "hehehehehe" but im too lazy to type all that :) 07:10:51 qless & segher also helped me a lot this morning 07:11:25 in the mulit* 4th area 07:11:39 they mentioned me bigforth 07:12:23 (can remember why did i use it, but probably it didnt work that time why ive tried it) 07:12:51 and qless has found a simple x86 eforth variant w multitasking extension 07:13:02 but multo 07:13:36 but the odds-and-ends of multiuserness still covers itself into dense fog 07:14:59 but ive managed to explain yesterday concisely why do i need multiuserness 07:15:23 imagnie a forth running 07:15:49 why do you need a multi USER forth ? 07:16:03 just run linux. have forth run in multiple consoles 07:16:06 u should be able to allow webusers to read and write variables in it 07:16:35 a forth web interface ? neat idea :) 07:16:38 or more generally how would u allow the to execute arbitrary forth code? 07:17:44 thanks, for the praise :) 07:18:39 u r developing the forth bot, aint u? 07:18:54 its a bit similar case, i think. 07:18:59 yes 07:19:18 thats on hold till i get this syscall mess straightned out 07:19:22 i think we could merge the two projects... 07:19:36 the way it is in the files you have is a bit kludgy 07:20:02 emmmm, do i have such files? 07:20:34 ive only dled isforth-1.00b.tgz 07:20:36 brb 07:22:09 thats whats out there right now - im working on rewriting the syscall stuff right now 07:22:22 syscall.1 just became VERY simple heh 07:22:22 very little in there now 07:26:52 let me warn u, i dont know much about the syscall mechanismz of *nix OSes 07:27:18 i did os-close programming under dos last time 07:27:34 but it was maaaany-many yrs b4 07:28:23 but i saw - in your syscall.1 - there r various syscall words depending on the number of args 07:29:36 but linux kernel hacker guide should explain this topic quite well, i guess 07:29:49 * onetom better not bother u w it 07:30:06 syscalls.1 is what gets included. it is a copy of either linux.1 or fbsd.1 depending on what you are compiling for 07:30:15 syscall.1 has one function for every syscall thats used in the kernel 07:30:36 now i have one syscall function for each type of syscall. syscall type based on how many parameters it uses 07:30:51 so sys_read is now just a 07:31:04 call do_syscall 07:31:04 dd 3 07:31:04 erm or is it dd 4 07:31:17 each syscall is a constant syscall number that uses do_syscall 07:31:50 do_syscall will branch to the correct syscall type based on how many parameters the syscall uses 07:31:50 i have a table of 256 bytes 07:32:12 each byte says how many parameters the syscall of that number (offset in table) uses 07:32:24 im adding a creating word 07:32:34 you would add a new syscall as follows 07:32:38 c, thats clean 07:32:46 #parameters syscall_number syscall name 07:33:08 but this means there r 255 syscalls could exist @ max ? 07:33:29 the word syscall would create a word called name that uses #parameters and has syscall_number compiled into its body 07:33:55 how many syscalls are there right now ??? hehe 07:34:03 dont know 07:34:08 i could make the table 16 bit eventually :P 07:34:19 very easy to do heh 07:34:27 but 65 thousand different syscalls would be a monster 07:34:36 ill stop using linux when that happens heh 07:34:41 but it eats up 64k then 07:35:27 but dont u either know the number of syscalls? 07:36:25 nope 07:36:34 the linux kernel isnt going to support every single syscall 07:36:55 if the user wants a syscall thats got no function in the kernel he just adds it 07:37:08 if the kernel didnt support sys_write he could do.... 07:37:31 4 3 syscall write 07:37:45 that would create a syscal word called write 07:37:55 that was syscall 4 and uses 3 parameters 07:38:05 x y z write 07:38:05 would then work 07:38:15 ive already understood it 07:38:37 not this 1 is the point 07:38:45 what if i have 2 write: 07:38:48 it has the disadvantage that all the syscalls are scattered arround the sources but keeps the syscall interface alot simpler 07:39:16 3 &1200 syscall something? 07:39:46 then u do: 07:39:52 &1200 ? 07:40:00 syscall-table 1200 + ! 07:40:06 1200 in decimal 07:40:15 dont u know this notation? 07:40:34 $xx hexa, &dddd decimal 07:40:40 0ooo octal 07:40:50 #bbbbbbbb binary 07:41:02 just try it! it works in gforth 07:42:01 so, what i want to say is: 07:42:05 my forth uses $ for hex % for binary \ for octal 'x' for char 07:42:15 check the syscall numbering conventions 07:42:16 and no prefix for current base 07:42:25 check the current number of syscalls 07:42:30 if you do that its a bug in your code basically heh 07:42:49 and just after knowing it, start using such a small table 07:43:59 bug? it just makes it not portable 07:44:52 its already not portable 07:45:06 i dont WANT isforth portable 07:45:36 im not trying to be "just like all the other forths" 07:45:57 but, come off it! ive tried me make it clear what does the 1200 mean - failed, ok. 07:46:05 for instance. im NOT ans compliant 07:46:05 i refuse to be ans compliant 07:46:05 i hate the ans standard - or parts of it 07:46:06 ans forth is a cammel 07:46:18 a cammel is a horse designed by a committee 07:46:43 what you are saying is that what if - in the future there are 4000 different syscalls 07:46:56 well 07:47:00 @least more than 255 07:47:10 when that happens it would be a simple matter to update isforth to account for that 07:47:10 but, for gods sake 07:47:27 HOW MANY syscalls exists NOW? 07:47:29 for now - allowing only 256 differnt syscalls in isforth isnt realy an issue 07:47:31 do u know it? 07:47:38 ive told me u dont 07:47:48 hang on ill go look 07:48:11 highest syscall is 255 07:48:25 but when u say "256 differnt syscalls" u r wrong, i think 07:48:25 from unistd.h 07:48:59 because this limit is not the number of the USED syscalls, but 07:49:08 syscall 0 is not used 07:49:32 the limit of the globally defined syscall constants 07:49:37 linus is a c coder. he doesnt understand zero based :P 07:49:37 heh 07:49:49 if there are more ill just make my table 16 bit and use 512 bytes instead of 256 07:50:10 oh, maaan, how hard can u get the point... 07:50:33 syscalls r identified by a number 07:50:49 dood - right NOW - its not an issue heh 07:51:02 im writing this so that CHANGING it will be easy 07:51:12 ok, got it 07:51:17 yes. and right now there are only 255 of them defined 07:51:17 syscal zero is unused 07:51:30 if he adds ONE MORE syscall - its a very small change i have to make in isforth 07:51:56 but u should have already checked it b4 writing the your code 07:52:28 i didnt need to - i know there are less than 256 - i just didnt know exactly how many 07:52:54 and my code is easy to change - thats why im writing it this way - to make FUTURE enhancements easier 07:53:08 im totaly rewriting isforths syscall interface to MAKE it easy to update 07:53:08 before - it wasnt 07:53:50 oh, then why didnt u say it at the very beginning!? 07:54:11 i mean : " i know there are less than 256 - i just didnt know exactly how many " 07:54:36 ive told u i dont know how many syscalls r there 07:54:45 thats what i was saying heh 07:54:45 oh 07:54:54 heh 07:55:07 oh, me gawd.. :) 07:55:44 we can agree, then 07:55:54 thats right, what uve done 07:56:07 :) 07:56:10 thats right that uve done 07:56:25 what i am doing - its not quite done yet heh 07:56:31 :) 07:57:05 ok, do it and if it works, we should start discussing this multi* topic over 07:57:09 again 07:58:44 i also have a - probably - completely different question 07:59:40 is there a standard way of using printf from forth? 08:00:03 k 08:00:07 or @least how can i implement sg like 08:00:17 vsprintf (or what)? 08:00:30 isforth will use sycalls to do multi tasking 08:00:52 printf is a c thing - i wouldnt want to write a printf for forth myself 08:01:05 ah, thats the other thing id like to know. how? 08:01:14 and i would definatly not want to write a vsprintf for forth heh 08:01:32 why? 08:01:47 imagine a logging problem 08:02:00 u have 2 construct log messages 08:02:24 than u have 2 send the constructed string to various loggers 08:02:41 eg. send it to syslog/klog 08:02:59 into a console 08:03:29 or into a pipe (like | mail -s 'log message' sb@sw.com) 08:03:35 how would u do that? 08:04:19 a log mgy would look like: [date time] subsystem: msg (parameters) 08:04:48 well - sysklogd says what logs should go to what files etc 08:04:50 s/mgy/msg/ 08:05:05 thers a standard way of writing to a log file 08:05:15 im not 100% sure how it works 08:05:27 but i know i can redirect certain types of messages to different places 08:05:28 oh, man, the forget syslog 08:05:39 using sysklogd.conf or something 08:05:42 just consider u have to write the logmsg 08:05:49 onto the screen 08:05:52 i forget heh 08:05:52 into a file 08:05:59 and u should email it away 08:07:52 2 do this, its comfortable to prepare the string in memory first 08:08:11 then output it into the various destinations 08:08:51 <# # # #> :) 08:09:01 for preparing such complex strings, the format-string way is very convenient 08:09:16 then copy the formatted number into a buffer 08:10:04 i would just prepare all the info in a buffer. i would not want to have to deal with " xxxxx %s yyyyy %n", string, number 08:10:04 aha, suure :) try to code the following example in forth: 08:10:30 i would just copy xxxxx into a buffer. append string to said buffer. append yyyyy and thenh <# # # # #> cmove 08:11:48 printf("[%02d-%02d-%02d %02d-%02d] asd: %s (%4x, '%c')", year,month,day,hour,min,error,param,char) 08:13:03 i would imagine it in forth like this: 08:13:50 {{ year month day hour min error param char }} dest sprintf 08:13:58 i mean: 08:14:12 {{ year month day hour min error param char }} formatstring dest sprintf 08:14:17 where 08:14:32 : formatstring s" [%02d-%02d-%02d %02d-%02d] asd: %s (%4x, '%c')" ; 08:14:33 char param error min hour day month year <# # # # #> type <# # # #> type etc etc etc 08:15:13 but, maaaan, type doesnt convert into memory... 08:15:23 but just onto the screen 08:15:31 you wouldnt have to type it. the returned value from #> is the address of a string. you could send that to a buffer 08:16:24 you could defer emit to a word that STORES to a buffer 08:16:35 that way youru "type" word would NOT emit each char 08:16:35 it would be stored in memory 08:16:49 but this way its quite a pain modifing the format of log msgs 08:17:25 ahhha, thats what i wanted to hear 08:17:31 think forth. not c 08:17:32 "defer emit" 08:17:59 what is the standard way doing it? 08:18:22 in gforth, eg, u have 2 defer both emit & type 08:19:16 emit and type are alwayd defered words. it makes sense 08:19:30 sometimes emit means send to console 08:19:30 other times it means send to printer 08:19:32 "think forth. not c" -- how would u internationalize such a program? 08:19:41 other times it means send to file 08:19:41