00:00:00 --- log: started forth/04.07.25 00:00:03 ayrnieu: i'm building an extensions framework for ficl forth (ficl.sf.net) 00:00:24 ayrnieu: and the sockets words use standard unix/bsd sockets 00:00:43 albeit dumbed down, so you don't have to spend 400 lines fucking with structures just to do a name lookup 00:04:51 Mixed precision arithmetic rules! 00:05:04 C sucks! 00:05:23 and arbitrary precision rules totally 00:06:16 I just use integers ;-) 00:06:30 And C can be good for some things 00:07:18 well, as the mathematician Kronecker once said, "God made the integers, all else is the work of man' 00:07:53 :-) 00:08:12 * crc likes integer math 00:08:31 * aum likes integer math, especially bigint math 00:08:52 Bigint = 64bit? 00:09:06 no - bigint is n-bit, where n is an integer 00:09:11 bignums, yes. 00:09:21 I just uses 32 bit integers 00:09:24 * aum adds a to-do item to write word prims for bignums 00:09:50 aum - the FSL has a (somewhat poor) bignum implementation. 00:09:51 crc: how long would it take you to generate an RSA keypair, using only 32-bit ints? 00:10:06 I have no idea 00:10:07 FSL? 00:10:22 I don't use RSA, am not familiar with the algo's, etc 00:10:45 crc: suffice to say, RSA is painful and mind-mincing without bignum primitives 00:10:56 Ok 00:11:16 32 bit is not enough. 00:11:25 Why not? 00:11:35 32 bits is 10 digits. 00:11:35 can be cracked in, like, 5 minutes 00:11:54 10 digits is not enough. 00:12:02 decent RSA these days needs more like 8192 bits 00:12:24 another integer challenge - generate a 1000-digit random prime number 00:12:30 I don't write my own encryption tools, so larger integers don't affect me 00:12:31 Sometimes you need 6 more digits for calculations. 00:12:41 And 4 digits is not a precision. 00:12:56 It is a non-precision. 00:13:22 32 bits = 10 digits as you stated :-) 00:13:22 or, 'find the first 10-digit prime number that occurs in the digits of e' 00:13:35 * aum looks for nice, simple, multiplatform bigint libs 00:13:39 crc - you'll want to specify a base. 00:13:46 aum - the FSL has one, then. 00:13:54 what's the 'FSL'? 00:13:57 Bases = 2, 8, 10, 16, 256 00:14:01 Forth Scientific Library. 00:14:10 ayrnieu: multiplatform? 00:14:25 aum - yes. 00:14:50 FSL is not good. 00:15:07 It lacks many of useful (to me) routines. 00:15:34 http://www.taygeta.com/fsl/sciforth.html 00:15:35 Maybe it is good for EEs, but not for scientists. 00:16:36 At least this had been last time I looked at it. 00:22:43 * crc doesn't need big integers for the work I'm doing 00:23:21 Well, I'd be quite satisfied with 64 bits. 00:23:37 It is 20 digits. 00:24:04 I'll add 64 bits to RetroForth once I get a 64 bit computer :-) 00:24:05 This is quite enough. 00:24:31 10 digits is enough for me right now 00:24:33 --- quit: tgunr (Read error: 104 (Connection reset by peer)) 00:24:40 * ASau knows how to double cells. 00:24:49 Double cells? 00:25:01 Make them twice as big? 00:25:11 * ASau is numeric wizard. 00:25:14 Yes. 00:25:52 That would require a massive overhaul of the RetroForth codebase 00:26:06 --- join: tgunr (~davec@vsat-148-63-4-106.c001.g4.mrt.starband.net) joined #forth 00:26:35 Most of the routines assume a 32-bit cell size 00:27:02 I'm back 00:27:06 I missed some fun 00:27:23 : stars 0; 1- star ; 00:27:31 At the least, I'd have to update all of the math words, variables, constants, >number, and CREATE 00:27:32 I need only: 0 0= 0< + XOR OR AND NOT 00:27:42 hi crc :) 00:27:43 And two-complement machine. 00:27:51 Hi Herkamire 00:28:01 : stars 0; 1- star stars ; 00:28:01 RF doesn't have XOR OR AND 00:28:07 Labas ritas, Herkamire! 00:28:12 hi ASau 00:28:21 : stars 0; 1 - '* emit stars ; 00:28:36 I also don't have 0= or 0< 00:28:48 crc: yeah :) your code in the scrollback I just read printed n+1 stars 00:28:59 Oops 00:29:12 you had the 0; at the end 00:29:18 Right 00:29:25 so eg 0 stars would print one 00:29:33 I just wrote it and posted without testing ;-) 00:29:40 :) 00:29:41 crc, what do you have then? 00:29:45 we all seem to do that. 00:29:49 my first one was wrong 00:29:56 I love 0; 00:30:03 Hmm, I did the same thing in my implementation of REPEAT 00:30:05 * crc sighs 00:30:20 it's important to note that it does not drop unles TOS is zero 00:30:35 I know. 00:30:35 I missed that 00:30:52 Mine leaves TOS alone unless TOS=0, then it drops it ;-) 00:30:57 right 00:31:03 that is what I find I want most of the time. 00:31:13 I do have another word that drops reguardless, but I don't use it much 00:32:12 I still like it better than "0if ; then" 00:32:35 : star '* emit ; : star-printer: >r : r> 0 ?do ['] star compile, loop postpone ; ; 5 star-printer: **** 100 star-printer: sky-of-stars 00:33:55 ayrnieu: huh? 00:34:20 Herkamire - I create words that print N-stars without looping. 00:34:29 in a silly manner, sure. 00:34:42 looks to me like you redefined r> 00:34:57 no, I didn't do that. 00:35:12 then how do you explain the " : r> " 00:35:21 I explain it in context, of course. 00:35:47 ??? 00:35:47 oh, I get it 00:35:57 I don't get it 00:36:26 crc - do you know how the outer interpreter makes use of STATE ? 00:36:40 I didn't realize you could compile a call to : 00:37:16 ayrnieu: why do you need >r and r> ? 00:37:31 ayrnieu: in RetroForth a : within a definition creates a new entry point into the word 00:37:48 crc: right. that was my assumption the first few times I read it 00:38:01 Herk - because >r and : and r> all happen at runtime, there. 00:38:05 crc: in his forth : is a normal word, which gets called when you call star-printer 00:38:13 Ok 00:38:22 That's odd behaviour in my view 00:38:27 so when you call star-printer: it defines a new word 00:38:30 it's ANSI Forth =) 00:38:39 * crc dislikes ANSI 00:38:57 ayrnieu: I know they happen at run time. why do you need them? does : effect the return stack? 00:39:07 this would work slightly-less-sillyly, anyway: : star-printer create dup c, 0 ?do '* c, loop does> count type ; 00:39:28 herk - no, but it may affect the data-stack in unpredictable ways. 00:39:36 That makes more sense to me :-) 00:39:56 herk - if I wrote in Quartus Forth, say, I wouldn't bother with >r r> because I'd know that : doesn't do anything with the data stack. 00:40:16 ayrnieu: oh, I didn't realize : affected any stack. 00:40:18 ok 00:40:32 herk - in gforth, : test [ .s ] ; =) 00:41:02 gforth puts 3 cells or such on the stack for : and control-flow words. 00:41:04 yikes 00:41:08 4 stack items 00:41:20 : test [ .s ] ; <4> 0 806902608 806902624 0 ok 00:41:47 what are they for you suppose? 00:42:01 I think I used to have an idea, but I don't anymore. 00:43:38 weird wild stuff 00:44:26 I remember doing silly things with that when I implemented RECUR; in gforth. 00:45:02 groups.google.com/groups?group=comp.lang.forth would have it, maybe =) 00:45:21 eh, I'm not that curious 00:48:02 seems so complex 00:48:38 in herkforth the editor seems a bit complex, but the compiler is dead simple 00:51:06 the compiler, ELF header, and init stuff, fits in less than 1K 00:51:25 on PPC (which has much larger code) 00:54:21 nifty. A native PPC forth, or hosted? I have an iMac right here =) 00:54:50 PPC Linux at the moment 00:54:59 It's a nice Forth :-) 00:55:15 well, I have qemu, if not PPC Linux. 00:56:04 that's 147 RISC instructions for the compiler 00:57:35 I really should finish the port to OS X. people keep saying they have a mac 00:58:08 I'm waiting for the native version so I can run it under PearPC :-) 00:58:21 :) 00:59:22 * ayrnieu realizes that he has stayed up *way* past his bedtime. 00:59:37 I gotta get some sleep too 01:00:24 crc: keep up the good work on retroforth. I'm looking forward to a more complete native version so I can play with it under bochs :-) 01:00:47 I will. I'm slowly making progress on the new kernel 01:01:16 maybe I should skip the Mac OS X version, and go straight to getting it to run under PearPC 01:01:21 The multitasking parts are making the existing drivers for HDD, etc break 01:01:31 oh dear 01:01:53 that sounds worse than tracking cache flushing issues 01:01:57 It is 01:02:09 Everything worked great in single tasking mode 01:02:25 I think I might go for cooperative multitasking instead of preemptive 01:02:55 It'd be easier than rewriting all of the drivers 01:03:38 :) 01:03:45 I'm definately going to do cooperative 01:03:48 at least at first 01:03:59 we'll see if I have reason to do more 01:04:08 I have lots of ideas 01:04:24 Me too 01:05:10 I had hoped to have it working by now, but time/bugs got in the way... 01:05:14 * crc sighs 01:05:32 ok, I'm off 01:05:36 ok 01:05:38 snack, then bed 01:05:39 Good night Herkamire 01:05:49 crc: nice to see you again. are you normally on at this hour? 01:05:55 very late here 01:06:02 Only on weekends (~4am here) 01:06:11 oh. 01:06:14 here to 01:06:23 where do you live? 01:06:26 I work overnights, normally on 7am-10am, then 8pm-9pm 01:06:31 FL? 01:06:34 Just outside Philadelphia 01:06:37 PA 01:06:39 neat 01:07:05 tathi (my cousin) (and his family) lives I think 45 minutes from Philadelphia 01:07:13 Cool 01:07:18 I go down there at least every year 01:07:21 near allentown 01:07:25 ok 01:07:39 I'm ~10 minutes from Philadelphia 01:11:43 ~50 miles from Allentown 01:12:16 cool 01:12:18 --- quit: Herkamire ("bed") 01:25:25 --- join: I440r (~mark4@216-110-82-59.gen.twtelecom.net) joined #forth 01:26:06 Hello I440r 01:26:13 hi 01:26:25 time for zzz methinks 01:26:45 Hehe, good night then :) 01:26:59 ;-) 01:27:01 :) 01:55:33 --- quit: tgunr (Read error: 104 (Connection reset by peer)) 01:57:21 --- join: tgunr (~davec@vsat-148-63-4-106.c001.g4.mrt.starband.net) joined #forth 02:04:37 --- join: qFox (C00K13S@cp12172-a.roose1.nb.home.nl) joined #forth 02:05:49 Hi, foxie. 02:05:53 yo 02:14:53 My fellow Forth Crusaider. 02:15:00 How are things going?= 02:16:37 busy 02:16:41 with the move and all 02:16:42 :( 02:16:46 so i'm gone now. 02:16:49 OK, see you 02:31:20 --- quit: aum () 02:33:06 What do you think about 4000-line files? 02:33:19 Long. 02:33:33 I have a 12,000,000 line file I use for performace + bug testing :-) 02:33:51 My IRC logs are something like that ;) 02:34:15 12,000,000 line IRC logs? -- cool! 02:34:35 Nah, not quite THAT much :) 02:34:48 * Robert counts 02:35:11 Just use 'wc -l' to count :-) 02:35:19 3387012 total 02:35:28 crc: What did you think, counting by hand? 02:35:35 Yup 02:35:38 Why not? 02:35:44 But I'm _almost_ at 12M. ;) 02:35:50 Just... 9M left. 02:36:16 You're not even a third of the way there yet 02:36:30 Bah. 02:37:07 robert@litium:~$ du -h irclogs/ 02:37:07 177M irclogs/ 02:37:13 Yay. :) 02:37:17 177million lines? 02:37:28 No, megabytes. 02:37:45 I've reduced 7 to 4 that GSL part. 02:38:01 Ahh 02:38:04 It's not all done. 02:38:13 My test program is 150MB 02:38:23 ASau: 4,000 lines of Forth? 02:38:26 No. 02:38:36 There's plenty of C inside. 02:38:53 Ok 02:39:11 I walk along this file and translate C to Forth. 02:39:24 I find bugs and correct them. 02:39:29 Etc. 02:39:38 Any strange things in the C? 02:40:09 Common mistake is of this kind: 02:40:28 x should not be zero 02:40:43 if(s==0) s=1; 02:41:06 s &= 0xff...ff; /* 32 bits */ 02:41:11 x = s; 02:42:10 * crc hates code like that 02:42:27 Another thing is multiplication in finite ring. 02:42:42 It's done this way: 02:43:25 x = a*(x - (x/q)) - r*(x/q); 02:43:33 if (x<0) x+=m; 02:44:10 Yuck 02:44:13 Where a, m, q, r are hard coded through "#define". 02:44:24 That's ugly code 02:44:56 Yes, it's much much simpler in Forth. 02:45:11 Unless you don't know what's it for. 02:45:24 The C code is commented right? 02:45:30 No. 02:45:45 WTF???? 02:45:50 There's not even hint of what it is. 02:46:03 That's wrong 02:46:18 There's source to make a mistake in q and r values. 02:46:53 a, m, q and r are closely tight. 02:47:07 q = m div a, r = m mod a. 02:47:58 And those two lines of C (which actually are 4 lines in the source --- they are written so!) 02:48:07 mean actually this: 02:48:17 x = (a*x) % m; 02:49:40 Problem is: C does not have */MOD. 02:50:10 Neither does RetroForth 02:50:17 Though it could be added easily enough... 02:51:23 I can do it by hands, provided minimal set mentioned above. 02:51:58 Right 02:52:47 There're more hard to translate sources. 02:53:07 48 bits arithmetics. 02:53:20 In Forth I can write it easily. 02:53:51 But reverse engineering from C is not as easy. 02:56:12 I understand 02:56:25 Reverse engineering is often hard anyway 03:22:03 --- join: crc_ (crc@0-1pool176-9.nas6.philadelphia1.pa.us.da.qwest.net) joined #forth 03:22:03 --- quit: crc (Read error: 104 (Connection reset by peer)) 03:22:36 --- nick: crc_ -> crc 04:00:29 --- join: aum (~aum@port-204-54-210.fastadsl.net.nz) joined #forth 04:04:20 hello aum 04:04:29 hi crc 04:04:42 do you, or does anyone, know of some event frameworks for forth? 04:04:56 I don't 04:05:16 Though the GameForth experiments will have one... 04:06:26 event frameworks are highly implementation specific 04:06:47 BigForth or Win32Forth might have some sort of event framework 04:06:49 that's the problem - i'm trying to conceive of a general platform-independent implementation 04:07:28 bigforth pisses me off somewhat - not too well done on windows, and badly underdocumented 04:07:42 win32forth is, as its name implies, windoze only 04:08:00 True 04:08:18 i guess the simplest (but not the best) kind of event framework would be something like: 04:08:21 GameForth will handle mouse and keyboard events under Windows and Linux once done 04:08:42 registering event handlers with the framework 04:09:06 passing in two xts - one which checks for an event, and one which handles an event 04:09:24 no 04:09:27 that won't work 04:09:31 hmm 04:09:35 ahh yes 04:09:59 the client can register zero or more 'event sources', and zero or more 'event handlers' 04:10:05 I have an event loop (written in assembly) that sets up some constants for type/ID of event and then passes control to a Forth word named 'event' 04:10:08 : event:key EVENT_ID 04:10:08 events can be identified by a number 04:10:08 dup KEY_RIGHT = if ;; then 04:10:08 dup KEY_LEFT = if ;; then 04:10:08 dup KEY_UP = if ;; then 04:10:08 dup KEY_DOWN = if ;; then 04:10:09 drop ; 04:10:11 : event:mouse ; 04:10:13 : event EVENT_TYPE 04:10:15 dup KEY_EVENT = if drop event:key ;; then 04:10:17 dup MOUSE_EVENT = if drop event:mouse ;; then 04:10:19 drop ; 04:10:32 It actually works pretty good for my current needs 04:11:11 what does ';;' do? 04:11:26 ;; exits a definition without stopping the compile process 04:11:31 Like 'exit' 04:11:53 so it's the same as 'exit'? 04:12:20 Yup 04:12:58 hmm, maybe something like: 04:13:18 client can register (1) event sources, (2) event receivers 04:13:44 event receivers can optionally pass back another event, thus acting as filters 04:15:03 could even simplify to just a single entity - event receiver 04:15:31 You still have to have hooks in the implementation to produce the events 04:15:48 client supplies the event sources 04:16:09 an event source could be done as an 'event receiver' which is bound to 'idle' events 04:16:15 That gets confusing from my point of view 04:16:42 ok - let's say we're after two different event types, events from the GUI, and events from a socket 04:16:59 ok 04:17:01 we'd register two event handlers - both bound to 'idle' event 04:17:12 the first 'idle' handler polls the GUI 04:17:20 the second 'idle' handler polls the socket 04:18:03 You still need hooks into the GUI and socket code, so it's still bound to a specific implementation 04:18:07 if the first handler gets a GUI event, it can then translate the event into another event type, and return another event - eg 'gui-dialog-button1' 04:18:33 meanwhile, the client registers more handlers, listening for these translated events 04:19:12 the only pain is that it involves a poll cycle, which is never good 04:20:20 I still fail to see how it would be independent of the Forth implementation 04:21:05 you're right - there are implementation-dependent details, such as how the handler would physically poll external things like guis and sockets 04:21:21 but it can be made to work in a platform-independent manner 04:21:53 :-) 04:22:06 I'll stick with my current approach for now 04:23:02 are you planning any user-interface words for your retroforth? eg, XFree, or WinGUI, or a 40x16 LCD panel? 04:24:16 Um, RetroForth itself uses a text console 04:24:49 There are plans for a graphical interface using libSDL under the various "hosted" ports 04:25:06 And a native VGA/VESA interface for the native version 04:25:37 (Releases up to 6.2x included an early version of the SDL bindings) 04:26:01 The idea is that RetroForth is a nucleus; you extend it to meet your needs 04:26:13 Dynamic linking of libraries, etc will be possible 04:26:22 At least for some ports (Linux) 04:26:59 Example is GameForth: it takes the RetroForth core, and builds up a set of words specifically for programming games 04:28:19 Those using RetroForth seem to like this approach, so I see no reason to change it 04:28:28 The main uses at the moment are: 04:28:34 1) Simple shell scripting 04:28:52 2) Web development (using Forth as a markup language) 04:29:09 3) As a basis for larger, application-specific Forth systems 04:29:18 4) As a scripting language for applications 04:29:40 5) As a simple standalone system for control of robotic equipment 04:37:41 I would like to have a nice 64x17 LCD one day though... 04:56:21 --- quit: I440r (Read error: 110 (Connection timed out)) 05:00:10 Well, I'm tired of this by now. 05:00:46 Work-in-progress release is at http://retroforth.org/dev/asau/RNG/25JUL04.TGZ 05:01:12 You may pick any random number generator of working 5 or 6. 05:01:31 (Maybe even more, but I'm not sure.) 05:03:17 index.html will appear later. 05:03:34 Looks good to me :-) 05:05:39 --- quit: tgunr (Excess Flood) 05:07:27 --- join: tgunr (~davec@vsat-148-63-4-106.c001.g4.mrt.starband.net) joined #forth 05:38:54 Well, I have to go... I'll be back tonight! 05:39:00 --- part: crc left #forth 05:45:42 --- join: lalalim_ (~lalalim@pD95EA18A.dip.t-dialin.net) joined #forth 05:46:14 Guten Tag, lalalim_! 06:06:38 --- quit: lalalim (Read error: 110 (Connection timed out)) 06:51:10 --- join: tathi (~josh@pcp02123722pcs.milfrd01.pa.comcast.net) joined #forth 07:18:08 What diff representation is better for patches? 07:18:15 Unified? 07:18:17 Context? 07:20:26 I like unified... 07:21:53 If you just use -u, that gives a unified diff with three lines of context IIRC... 07:22:24 I think, -U 5 or even 7 is better. 07:23:14 Well, I've never had a problem with patching things with 3 lines of context. 07:23:26 But the patch might be more readable with 5 or 7, I suppose. 07:24:23 But then, I think a lot of GNU projects also like to use -p, which gives the C function that the changes are in. 07:24:35 So I suppose that helps diff find the right place as well. 07:27:34 Hm. It seems that unified patch is easier to read. 07:28:11 And I don't count readability as the least matter. 07:29:21 yeah, I find unified patches _much_ easier to read. 07:29:51 With normal diffs, I can never remember which lines are from which file. 07:34:05 Well, first patch released. 07:34:14 What are you patching? 07:34:28 http://retroforth.org/dev/asau/RNG/ 07:34:43 Special thanks to crc for hosting. 07:34:55 ah. 08:18:19 --- quit: tathi ("leaving") 09:07:35 --- join: Frek (anvil@h87n2fls31o815.telia.com) joined #forth 09:08:41 Guten Abend, Frek! 09:09:05 Guten Tag Asau 09:09:49 How's your code? 09:10:08 it's fine I guess; I don't do much coding right now though 09:10:19 how about yours ? 09:10:26 What are you coding in Forth? 09:10:43 I'm done for today. 09:11:10 I've just did some OF programming as far; otherwise I've been investigating using Forth as a plugin language for applications 09:11:55 I'm rewriting part of GNU SL for my needs. 09:12:00 I'm mostly monitoring what happening in this channel; there's lots of interesting discussions here that's not always directly forth related. 09:12:21 i see 09:13:44 Probably, I should send corrections to GNU SL, but I have no willing doing this. 09:51:46 Frek, are you here? 09:57:08 ASau: yes 09:57:29 What's you domain of interests? 09:57:38 EE? 09:57:44 EE ? 09:57:57 Electronic Engineering? 09:58:23 no it's not my main interest although I did some electronic engineering in the past 09:58:51 Are you programmer/ 09:58:56 I've been 09:59:23 although programming is my main interest when it comes about computers I guess 09:59:39 Why embedded? 09:59:46 Or why OF? 10:00:11 well basically it started of with that I've did some work on powerpc linux, and powermacs has OF 10:00:49 however as I said; I don't do much programming/coding at all now days 10:01:09 And what's your occupation now? 10:01:18 currently unemployed basically 10:01:44 I do some part time work as pc support service stuff 10:01:53