00:00:00 --- log: started forth/04.10.06 00:26:18 --- join: wossname (~wossname@rn-v1w5a06.uwaterloo.ca) joined #forth 00:30:21 * Serg_penguin is writing log analysis script in Perl 00:31:11 * colorg doesn't have Perl in his Linux distro 00:31:34 i write on Win32 using old ActivePerl 00:32:05 --- quit: Herkamire ("bed") 00:42:15 kc5tja: You OK dude? 00:42:33 Yeah, why? 00:43:01 Your Forth stamp idea. You were so quiet I decided to check on you :o) 00:43:21 Yeah. 00:43:30 Well, they implemented every idea I had. 00:43:35 AVR microcontrollers, Forth, etc. 00:43:36 * kc5tja sighs 00:43:39 Everything. 00:44:45 All my life, I've been looking for something I can market myself with. I thought I had something with the ForthStamp idea. Pity. 00:44:53 hardware is tough for a startup. 00:45:22 Software is just as hard. 00:48:49 Well, as President, I'm going to kill Microsift, so don't lose hope on software :o) 00:49:44 I really feel thier position is horribly unnatural 00:51:46 Uh oh -- if I suddenly get disconnected, it's because we're having some rolling blackouts. 00:52:23 hehe 00:52:33 we have a diesel generator in the basement for that 00:52:41 fridge_: I'm home. 00:52:48 ahh, thought you were at work 00:52:48 And we don't have basements here in California. 00:52:56 * kc5tja is off tonight and tomorrow. 00:53:08 well, it isn't really a basement, it's a cordoned off section of the lower carpark 00:54:39 no houses in australia seem to have basements 00:54:43 or attics for that matter 00:55:13 i want a tshirt that says forth if honk then 00:55:30 love 01:22:29 * kc5tja just thought of another crazy idea. 01:23:34 Digital communications over an NTSC channel. At 60 frames per second, packing 32 bytes per scanline, you can get 4.0Mbps (approximately). Since amateur radio operators are all about re-using existing equipment ad nauseum, and since amateur TV equipment is already commercially available, this ought to be a salable idea. 01:23:40 * kc5tja will need to do the research. 01:28:16 yeah, that's crazy 01:36:45 * Serg_penguin is EAGER for digimode broadcasts RX'able by comp 01:37:12 the only one is RTTY by very few radio stations 01:44:14 --- join: saon (Ecoder@c-24-129-95-254.se.client2.attbi.com) joined #forth 01:45:17 --- quit: wossname (Read error: 60 (Operation timed out)) 01:45:24 I just walked a guy through is first osimplay instruction 01:45:57 h 02:03:06 colorg: It's not as crazy as you may think. 02:03:27 Amateur radio entheusiasts are world reknowned for their frugality. 02:03:44 An ATV transmitter is basically just an AM radio transmitter with a 6MHz bandwidth (versus 3kHz for voice-grade transmitters). 02:04:07 The only thing the ATV transmitter would require is that the voltages fed to it conform to NTSC specifications. 02:04:41 And a network interface unit for this kind of bandwidth is actually cheaper to build (based on my initial research) than a 9600bps TNC! 02:04:50 Serg_penguin: You don't have PSK31 in your area? 02:05:50 Serg_penguin: RTTY, HF Packet, and PSK31 can all be received and transmitted via the PC sound card. 02:05:55 Honestly, you're totaly over my head. 02:13:14 --- quit: tgunr (Ping timeout: 14400 seconds) 02:15:56 --- join: crc (crc@51-pool1.ras11.nynyc-t.alerondial.net) joined #forth 02:21:44 --- quit: saon ("Leaving") 02:30:35 yes, i heard some of PSK31 in RU ham mags 02:37:16 --- join: wossname (~wossname@rn-v1w5a06.uwaterloo.ca) joined #forth 03:15:57 brb -- getting food 03:15:57 --- quit: madwork (Read error: 232 (Connection reset by peer)) 03:15:57 --- quit: onetom (Read error: 104 (Connection reset by peer)) 03:29:01 that makes me hungry 03:29:50 --- join: onetom (~tom@novtan.bio.u-szeged.hu) joined #forth 03:41:17 --- quit: wossname (Read error: 110 (Connection timed out)) 04:07:23 eforth init code puts the address of COLD in esi and does jump (%esi) 04:08:18 jump *(%esi) actually 04:09:15 Why can't it just jump to CODE directly? 04:09:54 jump *(%esi) is __NEXT 04:10:15 COLD, not CODE 04:10:23 sorry 04:10:35 * crc is tired after a long night at work 04:11:00 what threading model is eforth? 04:11:12 SO, it fetches the address at COLD: and jumps there. That's the address of DOLIT 04:11:19 * crc likes stc :-) 04:11:45 crc it uses ENTER and NEXT, it's not stc, no jsrs 04:12:10 ok 04:13:00 Thanks for listening. I get the indirection to DOLIT now. DOLIT is machine code. 04:13:18 You were a great listener :o) 04:13:24 :-) 04:14:12 I think I'll have an eforth that builds with just a shell in about another day 04:14:31 maybe insertable into Linux kernelspace. 04:14:50 cool 04:15:06 I'd like to get RetroForth running in Linux kernelspace someday... 04:15:59 It'd make a nice debugger 04:16:00 I got my 3-stacker in there. On vt1 04:16:17 I know; I've looked at it in the past 04:16:24 WHen it dies it stays dead though :o) 04:16:26 Quite a nice feat 04:16:33 :o) 04:16:34 Don't kill it ;-) 04:16:47 Doc, it hurts when I do this...... 04:17:05 H3sm isn't robust by any means 04:17:13 Neither is RetroForth 04:17:33 Which is why I'm beginning a major code cleanup/debugging effort 04:18:01 eforth is fairly non-experimental design-wise 04:18:37 I've never looked into eforth's design 04:19:33 Well it's a regular forth with dovar and dolit and next and so on 04:20:21 ok 04:20:24 The threading scheme I used in the C version of H3sm was an accident. 04:20:30 * crc should take a look at eforth again 04:20:44 What was the threading scheme you used? 04:21:23 it was like virtual stc or something. Some of the ugliest C ever written. 04:21:43 sounds ... interesting 04:21:59 I haven't touched C in over a year 04:22:15 (Though I do use libraries written in C sometimes) 04:22:21 I use osimplay for C-type stuff 04:23:02 I gave away my K&R2 when I got osimplay working 04:23:22 I still have a copy that I refer to sometimes ;-) 04:23:31 Helpful when studying other people's code 04:23:37 yeah 04:23:44 * crc is quite happy with Forth & assembly these days 04:25:03 osimplay was as Forthy as I could do a 1-stack language. 04:25:26 I should take a look at osimplay 04:25:46 ftp://linux01.gwdg.de/pub/cLIeNUX/interim 04:25:48 It might be fun to play with 04:26:05 Forreal.tgz has a 1200 byte kernel in osimplay 04:26:29 What can that 1200 byte kernel do? 04:26:53 twiddle 3 VGA cells based on 3 different interrupts :o) 04:26:59 in pmode 04:27:03 ok 04:27:40 i.e. the most miniscule excuse imaginable for the use of the term kernel 04:27:51 I dislike interrupts... 04:28:24 I've heard "kernel" applied to "hello, world!" type things before, so your's is at least more capable than that 04:28:34 yeah, they have to be dealt with ;o) 04:29:17 In RetroForth/Native, I have an empty IDT (48 ints), you can hook Forth words to interrupts 04:29:23 Mine is on the metal, pmode, 100% original, 100% osimplay6 which is 100% bash, and not looking over it's shoulder at unix 04:29:55 but it doesn't have much 04:30:03 no processes, no MM... 04:30:07 RetroForth/Native is a bootable Forth system. No *nix influence that I can see 04:30:30 A handful of drivers 04:30:40 I used ENTH to bang on a floppy, I don't recall if I got Retro booted 04:30:45 Blocks (which are awesome, despite what I440r thinks...) 04:31:01 Retro still has a crappy boot sector 04:31:01 Blocks are basic :o) 04:31:09 Blocks are great 04:31:18 My boot sector is 14 bytes 04:31:27 I *do* have a multiboot version of Retro now 04:31:33 I used to have the osimplay in my .sig 04:32:00 Mine is longer, but it sets up pmode, and the GDT as well as loading the Forth system 04:34:03 What do you mean multiboot? 04:35:00 GRUB, the GRand Unified Bootloader 04:35:26 It's a standard that allows multiple kernels to be booted by one boot loader 04:35:58 It's bigger than I'd like, but it is very reliable 04:36:42 OK 04:37:05 Is Retro in asm? 04:37:16 Asm + Forth 04:37:29 I've been slowly moving more and more of it into Forth 04:38:03 Only a handful of words (the minimal compiler, interpreter, key, emit, bye) are in assembly 04:38:18 I figure about another day on eforth --> osimplay 04:38:27 Though there is a fair amount of hand-compiled code in the Forth part 04:38:44 * crc will have to try your osimplay'd version of eforth once its done 04:39:48 it's about 100 times slower than gas to assemble, 93 seconds versus .6 seconds for a 14k Forth 04:40:17 but the result is basically bitwise identical 04:40:45 nice 04:40:54 eforth is 14k? 04:40:57 --- join: wossname (~wossname@rn-v1w5a06.uwaterloo.ca) joined #forth 04:41:17 without the 160 Linux syscalls I added. With em it's 19k 04:41:26 That's 9-10k bigger than RetroForth :-) 04:41:32 heh 04:41:37 19k is ~11k bigger 04:42:23 with like a floppy and HD drivers? 04:42:47 shit that can be osimplay in no time 04:42:52 The entire native version (with hard drive, serial port, parallel port, block editor) is about 9k 04:43:08 Only about 2k is written in assembly :-) 04:43:17 I'm converting the wrong Forth :o) 04:43:18 back 04:43:19 It *should* be easy to port to another assembler 04:43:31 It's NASM? 04:43:36 It's in FASM 04:43:57 Converting to NASM is just changing a few macros, and a couple of directives 04:44:20 I have a gasX script for gas-->osimplay, gets about 80% of it 04:44:21 * crc can whip together a NASM version if you want it 04:44:34 cool 04:44:46 I've been thinking of writing a version in gas 04:44:56 and a MASM script I used to bang on colorforth some 04:45:12 where's the latest retro? 04:45:47 "latest retro" -- oxymoron if I've ever seen one. :P 04:45:56 *SMACK* 04:46:10 http://www.retroforth.org/release 04:46:36 Right after eforth :o) 04:46:41 ciao 04:46:44 --- quit: colorg ("ircII EPIC4pre2 cLIeNUX. Can you say that?") 04:47:34 brandnew 04:47:37 you're retro 04:48:44 :-) 04:50:00 I need to get RetroForth/Native to a useable state soon; I'm growing tired of Linux 04:55:42 In Soviet Russia, Linux grows tired of YOU! 04:55:44 ;) 04:55:49 Sorry -- I'm on a food high. 04:56:19 mmm. food 04:56:29 With apologies to Serg_penguin as well; the "In Soviet Russia" bits are parodies of a Russian comedian named Yakov Smirnov (I believe; well, *I* enjoyed his humor at least) 04:56:53 :-) 04:57:20 kc5tja: what version of L4 will you being using with FTS/Forth? 04:58:00 I was planning on using Pistachio(sp?), but Fiasco is a viable alternative too. 04:58:31 http://c2.com/cgi/wiki?InSovietRussia 04:58:39 ok 04:58:50 I've been looking at Pistachio 05:00:21 Why a microkernel as opposed to a monolithic kernel? 05:00:39 In Soviet Russia, Janet's Nipple exposes YOU! 05:00:41 * fridge_ giggle 05:02:37 --- quit: fridge_ ("Leaving") 05:07:55 Better modularity, better maintainability, more flexible and dynamic, easier to test, . . . 05:08:19 Can you directly access the hardware? 05:08:41 If you have suitable permission to do so, yes. But you still run as a ring-3 program. 05:08:51 interesting 05:09:07 L4 is hierarchially arranged too, much like a cache kernel. 05:09:12 nice 05:09:17 There is one master task which owns all of physical RAM. 05:09:33 If you want to manage a portion of physical RAM, you need to allocate memory by asking it for a chunk. 05:09:50 It will (if it thinks it can) apportion a chunk of its resources to your task. 05:10:00 Note: there is a natural law of conservation of resources. 05:10:24 In that respect, resources are never created nor destroyed: they only change hands as far as who manages them. 05:10:28 Likewise with raw I/O. 05:10:58 Although the rigidity isn't quite there for raw I/O (you have one master memory task, and then one "root" task responsible for spawning all other tasks that you need). 05:11:14 It is nonetheless the documented philosophy. 05:11:25 It sounds pretty good 05:12:00 L4 totally rules as a microkernel. And it's fast. 05:12:02 How big is L4? 05:12:16 I can't answer that. That depends on what implementation you go with. 05:12:31 The original L4/x86 (which is no longer maintained and has an obsolete API) assembled to 12KB. 05:12:31 ok, I guess that makes sense 05:13:00 12k isn't too bad 05:13:03 The newer versions are written in a strict subset of C++, and offer a new API and new features. I would predict that they compile to somewhere in the 32KB to 64KB range, which isn't TOO bad for something written in C++. 05:13:30 Hmm, I'd need a C++ compiler to build it then 05:13:39 GCC 05:13:59 I don't have GCC installed right now 05:14:02 Ahh 05:14:15 All the open source L4 variants, I believe, rely on GCC's C++ compiler. 05:14:33 ok, I'll have to dig out the CD with GCC on it then 05:14:44 And, you know, I've actually been thinking, maybe I should just not persue Forth for Linux, and just concentrate instead on FTS/Forth for L4 directly. 05:14:48 I mean, Linux already has gforth. 05:15:05 And a whole bunch of other Forths too :-) 05:15:10 And GForth is the closest thing to a standard Forth for Linux 05:15:24 Yeah, but you gotta admit, for a Forth, gforth is *popular* for Linux. 05:15:27 * crc doesn't use GForth anymore 05:15:29 True 05:15:38 I like GForth all around. 05:15:50 I just wanted something that I can build self-contained applications with. 05:15:59 But I've been thinking that that isn't quite the right way to go about things. 05:16:19 Maybe I should just abandon the Linux port of FTS/Forth, and just start developing Dolphin right away. 05:16:39 Dolphin being the old OS you were working on years ago? 05:16:45 yep 05:17:02 I remember reading about it... 05:17:10 never tried it though 05:17:30 Use L4, make Forth the native language environment that runs under L4, and then code the rest of Dolphin on an as-needed basis all in Forth. 05:17:40 cool 05:17:42 Dolphin never existed in any state where you could try it. 05:18:05 I guess I didn't miss anything then. 05:18:36 I've been working on RetroForth/Native a lot, but it's rather buggy at the moment 05:18:47 If you could imagine a version of L4 that didn't support the MMU, that's basically what Dolphin evolved into. A flat, single-address-space microkernel. 05:18:52 It worked *damn* well. 05:19:04 single address space? 05:19:17 One address space. 05:19:26 No processes, only threads. 05:19:37 ok 05:19:47 I dislike address spaces 05:20:01 I don't dislike them. 05:20:07 They're necessary in many cases. 05:20:28 such as? 05:21:05 you have a RAM disk filesystem you want to use, on a 32-bit architecture like x86. 05:21:13 You allocate 3GB of space to the RAM disk. 05:21:20 That's very nearly the entire address space for a single program. 05:21:37 In a single-address-space system, you can't use any other processes while this filesystem is running. 05:21:46 In a multi-address-space system, it's a trivial matter. 05:21:55 ok, good point 05:22:19 I *prefer* single address space then 05:22:31 Oh, well, I don't think there is anyone here who would disagree. 05:22:39 SAS OSes are definitely preferred all around. 05:22:43 :-) 05:22:56 But MAS overcomes the native CPU's address space limitations. 05:23:16 I was toying with the idea of using multiple address spaces in a future version of Dolphin to hold untrusted software. 05:23:27 All trusted software (controllable by the user) would all run in the same address space. 05:23:33 Hence, trusted software would run faster. 05:24:52 ok 05:25:08 --- quit: Serg_penguin () 05:25:21 Do you have any idea how a microkernel-based OS would run on older hardware? 05:25:46 (e.g., a 386 or 486 with 1-4MB ram and a small hard drive) 05:25:50 Well, Dolphin achieved over 1000 task switches per second on a 386SX-16 with 4MB of RAM (of which it used only 1.5MB). 05:26:02 wow 05:26:44 L4 is very Dolphin-like in overall capability, except that it also supports the CPU's MMU. I don't know how much of a hit the MMU support incurs, but the synchronous IPC model it has ought to make for some nice performance numbers. 05:26:55 ok 05:27:09 But then again, does L4 even *support* the older CPUs? I'm not sure that it does. 05:27:16 hmm... 05:27:21 I know the L3 supported the 486, but L3 isn't anywhere to be found. 05:27:26 You may have to hand-hack L4 to support the 386. 05:27:48 I can do that :-) 05:28:06 * crc loves his 386 05:28:21 My laptop is a 486 05:28:43 My laptop is a 600MHz Celeron 05:29:20 486 at 33MHz, 8MB of RAM, monochrome 640x480 VGA screen only, floppy drive, with no internal networking. 05:29:25 Though I do most of the testing of RetroForth/Native on the 386 and a Pentium 05:29:42 Although if I get it up and running again, I'm going to build my own UWB-like device for it. FCC be damned. :) 05:29:42 I remember laptops like that... 05:29:46 :-) 05:30:15 * crc has an old 8086 laptop... I'll be running the 16-bit RetroForth/Native on it soon 05:30:45 Haha 05:30:49 What use do you have in mind for UWB? 05:31:03 Just some homebrew RF-emitting devices that have really low power. 05:31:15 About the same amount of power as a cordless drill puts out. :) 05:31:29 ok 05:31:59 Plus it'll be fun to play with the concept. 05:32:03 yeah 05:32:18 I've played around with low-power rf stuff in the past... 05:32:41 To distinguish one UWB radio from another, instead of tuning by frequency, they "tune" by correlation sequences. Most UWB devices have 1024-bit correlation sequences. 05:33:02 (This means that a single 1 or 0 to transmit literally takes 1024 bits on the air) 05:33:43 * kc5tja will use only 8-bit correlation sequences. So I fully expect my device to get a lot of RF interference from other devices. But hey, that's the life of an FCC Part 15 device. 05:33:54 yup 05:34:19 1024 bits is excessive for hobby stuff I think 05:34:21 --- join: madgarden_ (~madgarden@Kitchener-HSE-ppp3576798.sympatico.ca) joined #forth 05:34:53 Yeah. Having 1024 bits of correlation is equivalent to 30dB of processing gain (e.g., it mimicks the behavior of an antenna having 30dB of gain). 05:35:17 I suspect 8 bits will have something closer to 13dB. 05:35:30 Still quite respectable, all things considered. 05:36:32 I saw your radio tuner code; it's quite nice 05:36:37 --- join: rob_ert (~robert@c-bf5a71d5.17-1-64736c10.cust.bredbandsbolaget.se) joined #forth 05:36:38 I figure I can use the ATmega8 microcontrollers for these too. 05:36:47 (re: old computers ;) 05:36:49 Thanks. 05:37:20 --- quit: onetom (sendak.freenode.net irc.freenode.net) 05:37:20 --- quit: swsch_ (sendak.freenode.net irc.freenode.net) 05:37:22 --- quit: warpzero (sendak.freenode.net irc.freenode.net) 05:37:22 --- quit: madgarden (sendak.freenode.net irc.freenode.net) 05:37:40 * rob_ert is using Minix, on a 6MB 386sx25 05:37:46 --- join: onetom (~tom@novtan.bio.u-szeged.hu) joined #forth 05:37:46 --- join: swsch_ (~stefan@p5091F975.dip.t-dialin.net) joined #forth 05:37:46 --- join: warpzero (~warpzero@dsl.103.mt.onewest.net) joined #forth 05:37:52 --- quit: warpzero (Excess Flood) 05:38:09 you have 6MB on a 386? you're lucky. 05:38:18 I only have 2MB on mine :-( 05:38:28 --- join: warpzero (~warpzero@dsl.103.mt.onewest.net) joined #forth 05:38:31 * kc5tja notes that Minix is a microkernel-based Unix clone. 05:38:39 (Plenty big enough for RetroForth/Native though) 05:39:21 crc: Yeah, I know.. this came with 2MB, but I found some old memory stick for it. 05:39:28 It's an IBM PS/1, btw. 05:39:32 cool 05:39:43 Ahh, a tank. 05:39:50 Hmm, I seem to have an original IBM PC floating around here somewhere... 05:39:54 Running httpd and ftpd now :) 05:40:11 I wonder how hard it would be to cram RetroForth onto that beast 05:40:28 It has all of 64k RAM installed 05:40:48 Man, please tell me it ran DOS 1.0. :) 05:40:50 * crc would need to get RetroForth onto a 5.25" disk though. That could be a problem 05:40:54 --- join: fridge_ (~fridge@dsl-220-253-75-187.NSW.netspace.net.au) joined #forth 05:41:05 Actually I ran DOS 2.11 on it for a while 05:41:41 It used to have more RAM, but that was removed for something 05:41:45 Aack -- I bet that consumed a fair chunk of RAM. 05:41:51 It did 05:42:26 I used to play with the crummy BASIC interpreter that was in the ROM too 05:42:41 :) 05:42:59 Where every time you put a character to the screen, the display would blank. 05:43:08 (well, that was a BIOS issue, but still) 05:43:12 Yeah 05:43:24 There's a nice VGA card in it now :-) 05:43:37 And despite the blazingly fast 4MHz CPU, it was slower than a C64 drawing to the screen because of the obnoxious number of CPU cycles the INT 10h API consumed. 05:43:52 I can fit the 16-bit RetroForth into the 64k of ram, and still have enough space to write simple programs :-) 05:44:14 crc: Remember that PygmyForth only used 64KB of program space tops too. 05:44:21 Really? 05:44:23 You could readily write very nice programs in it. 05:44:30 Yes. It's a .COM file! 05:44:45 You *can* use >64k in a .COM file. 05:44:57 oh dos 05:45:04 crc: *only* if you deliberately touch the segment registers. Which Pygmy doesn't do. 05:45:07 Guaranteed. 05:45:13 I didn't know that 05:45:36 --- quit: Raystm2 ("User pushed the X - because it's Xtra, baby") 05:46:01 warpzero: Not just DOS -- DOS 2.11 and earlier. :) 05:47:32 man am i glad i was too young to program back then 05:47:36 I should try to get a DOS 1.0 disk 05:47:46 warpzero: why? 05:48:07 sounds scary 05:48:37 it was fun :-) 05:50:12 i mean DOS is like 05:50:19 barely an OS right 05:50:25 I liked DOS 05:50:51 warpzero: what makes it "barely an OS"? 05:51:05 well no liek 05:51:15 i mean nothing was protected 05:51:17 was it 05:52:18 Nope 05:52:24 Why should it be protected? 05:53:06 well i mean it was basically a bunch of hooks and a command line 05:53:10 So? 05:54:42 It was a nice OS to program for 05:55:00 I miss using it 05:58:00 --- join: Raystm2 (Rastm2@AC8C91DC.ipt.aol.com) joined #forth 06:13:41 --- quit: rob_ert () 06:19:30 --- join: madwork_ (~madgarden@derby.metrics.com) joined #forth 06:20:24 --- quit: crc ("Time for bed... Goodnight!") 06:25:05 DOS is nice to code for :) 06:25:48 is it ? 06:26:39 denial of service? arke is teh scirpt kidie!!1 06:26:52 heh. 06:27:02 xor ax, ax 06:27:06 mov bx, 0x13 06:27:08 int 0x10 06:27:10 BOOM 06:27:20 instant 320x200x256 video mode 06:27:31 he he 06:30:21 and easy to plot pixels too 06:30:51 0xa000:(320*y+x) 06:30:58 --- quit: Raystm2 (Read error: 104 (Connection reset by peer)) 06:31:35 * kc5tja downloaded the L4Ka::Pistachio microkernel sources. Studying them now. 06:31:46 --- join: Raystm2 (Rastm2@ACD08C94.ipt.aol.com) joined #forth 06:32:28 As advertised, it is written in C++. I have yet to see one template though. They're basically using the C++ compiler as a glorified macro engine though. Lots of classes contain precisely *one* field of some integer type, and a bunch of methods that are inlined (should expand to a small handful of CPU instructions each). 06:32:33 All in all, I like it. 06:32:44 And it's a heck of a lot easier to read than the Linux sources, that's for sure. 06:33:52 Indeed, I'd wager that *most* of the code backing L4 is implemented in the headers. :) 06:37:28 :) 06:39:40 WOW. I had no idea that you can use a #define macro in a #include statement. 06:39:46 #include INC_API(tcb.h) 06:40:07 INC_API(x) converts the string x into "x86/x" for x86 machines, "ppc/x" for PowerPC, etc. 06:41:37 new to me too :) thanx kc5tja 06:42:56 Raystm2: INC_API() is a macro defined within L4. You won't be able to use it elsewhere. (Unless you define your own of course) 06:44:51 I was commenting about the usage of #defing in an #include. I've never done that. seems like you could provide porting diffs in #include -- no ? 06:45:23 #define even 06:49:07 I've never done it before either. 06:49:42 --- quit: Raystm2 (Read error: 54 (Connection reset by peer)) 06:53:37 neat :) 06:53:54 I should d/l L4 and take a look at it. 06:54:11 Alan wants me to, and kc5tja is recommending it 06:54:11 :) 06:54:46 adios, school! :) 06:58:26 Well, some functions are necessarily ugly. The system IPC() function is several hundred lines long in and of itself. 06:58:32 But then, it has to be, because it needs to be very fast. 06:58:38 But most functions are relatively clean and simple. 07:06:41 WOW. L4 still only has 7 kernel entry points. I'm rather impressed. 07:07:02 --- join: Raystm2 (Rastm2@ACA04A32.ipt.aol.com) joined #forth 07:15:14 --- quit: wossname (Read error: 60 (Operation timed out)) 07:31:04 OK, I stand corrected about my size estimates for the microkernel. L4Ka::Pistachio weighs in at around asdfasdfjasl;djfas;jfdasl;f;jdf;jdfjd;fasjd;fasjd;asj;dlasf 07:31:12 oops 07:31:19 Damn CTRL-S and CTRL-Q. :( 07:31:27 Anyway, it weighs in at around 135KB. 07:31:46 hehe 07:32:00 So it's about 10x as big as the original L4/x86. 07:32:18 I suppose that is more or less to be expected, though -- new API and support for a lot of things the original L4/x86 didn't originally support. 07:32:27 Plus, of course, GCC doing loop unrolling to optimize for speed. 08:11:39 What's the L4 y'all are talking about? 08:11:43 * swsch_ scratches head 08:14:58 Microkernel? 08:15:14 I remember another coder speaking about it long ago, and that he admired it a lot. 08:17:45 Yes, L4 is a microkernel. 08:17:55 * kc5tja has been thinking of basing FTS/Forth on it, as you might recall. 08:21:42 One thing is for sure, despit the size of the kernel sources, it sure builds amazingly fast. 08:22:36 Does it have a homepage? 08:22:45 * Robert feels like he should check it out 08:25:09 http://www.cse.unsw.edu.au/~disy/L4/ is a good trampoline site. 08:25:48 http://l4ka.org/projects/pistachio/ is the site I grabbed my copy of L4 from. 08:25:55 Thanks 08:27:09 Just remember that L4 is like Unix now: there is no "the" L4 -- you have different flavors of them now. But they're all sufficiently close to each other; for example, the x86 version of L4Ka still has 7 entry points. :D 08:27:15 Just like the original L4/x86. 08:27:30 Isn't there an L4 standard? 08:28:05 An API standard, yes. 08:28:16 No ABI standard though (just like Linux and BSD, for example). 08:28:22 ABI? 08:28:33 Application Binary Interface 08:28:44 E.g., what registers parameters get stuffed in, how system calls are made, etc. 08:30:24 If I wanted to, I could take the API documentation, and make a version for the 65816. :) 08:54:23 --- join: tgunr (~davec@vsat-148-63-4-107.c001.g4.mrt.starband.net) joined #forth 08:55:02 OK, I'm sending buffer data via tcp. Which stack syntax makes more sense... ( "URL" addr len ) or ( addr len "URL" )? 08:55:15 http, rather. 08:55:18 --- nick: madwork_ -> madwork 09:00:49 I don't understand what you are trying to communicate with the stack syntaxes. 09:01:15 Does "URL" refer to a single stack item, or something that is parsed from the input stream, or...? 09:01:28 Yes, it's a single stack item. 09:02:21 I'm leaning towards the 2nd syntax, as I can flow easily into a connection open followed by the send. 09:02:43 Anyway, brb. 09:02:52 Generally speaking, you want the parameter that changes the least often to go towards the top of the stack. 09:03:00 I don't know if that helps you at all. 09:12:02 --- join: arke_sch00l (apache@11.198.216.81.dre.siw.siwnet.net) joined #forth 09:12:55 Going to bed. 09:13:01 OK, good night 09:14:16 good night kc5tja 10:09:07 --- quit: swsch_ ("Leaving") 10:13:59 --- quit: arke_sch00l ("CGI:IRC (EOF)") 10:20:09 --- join: wossname (~wossname@rn-v1w5a06.uwaterloo.ca) joined #forth 10:33:59