00:00:00 --- log: started forth/06.09.02 00:15:14 --- join: Cheery (n=Cheery@a81-197-19-23.elisa-laajakaista.fi) joined #forth 01:13:33 --- quit: snowrichard ("Leaving") 01:19:16 --- join: snowrichard (n=richard@12.18.108.181) joined #forth 01:19:39 hello 01:30:11 --- quit: snowrichard ("Leaving") 05:04:05 --- join: PoppaVic (n=pete@0-1pool66-100.nas22.chicago4.il.us.da.qwest.net) joined #forth 05:40:26 gdb is crashing and dumping core on c@ 05:40:39 hehehe 05:40:40 well, to strlen actually 05:40:53 thats evern more amazing 05:41:22 trace into? 05:42:03 i run the bin then it dies again, then i gdb.l and gdb dies 05:42:18 so no, cant. 05:43:06 Oh.. and this is all asm, eh? So: no printf-debugging? 05:43:12 yea 05:43:32 but i added a .globl line to the CODE macro 05:43:32 * PoppaVic chuckles 05:43:45 and use gcc to make a dynamic bin 05:43:53 so libsegfault works 05:44:14 well, that's fine - but you ain't learning why something fails. 05:44:35 gdb dies in strlen. 05:44:47 i think the problem is aligned access 05:44:55 oops, unaligned 05:45:07 well, if you are using C strlen, and Forth strings - it could be that issue 05:45:31 gdb uses strlen 05:45:44 eforth just uses write.2 05:45:49 I have no idea of the proper rules to use for alignments anyway. 05:46:26 it's not that strict actually. 05:46:31 How can gdb use strlen when it can be run on anything compiled from anything? 05:46:41 x86 can do it but aligned access is faster 05:46:44 fewer bus reads 05:46:52 Yes, aware of that 05:48:16 what's the gdb cmd to name a source file? 05:48:31 no idea, I usually call on 'help' 05:48:34 k 05:48:53 gdb is basically useless to me 05:49:02 I think you'd need to rebuild with debugging 05:49:08 ye 05:49:09 a 05:51:46 how do i get linenumber info into the binary? 05:51:52 i remade with -g -ggdb 05:52:01 that should do it all 05:52:06 dint 05:52:24 I don't know your flags, for C I just use -ggdb 05:52:33 that's i i used 05:53:12 doesn't the assembler manpage tell you? 05:53:32 gcc -g -ggdb efort.c -o efort 05:53:45 libsegfault is done by ld.so 05:53:51 oops 05:53:53 efort.S 05:53:54 umm... OK, I thought this was your asm-beast 05:54:05 it's a modified eforth.S 05:54:10 ahh 05:54:12 i rm'd some :'s 05:55:02 turn on abc tv 05:55:11 puppies on a shoulder sling 05:55:14 lol funiest 05:55:35 'doberman in a backpack' 05:55:36 lol 05:58:31 well, *sigh* Eforth doesn't seem too useful to my porpoises, since it seems to be all for the x86 and in asm 05:59:15 not all words are code; and eforth has been ported to many a-cpu 05:59:27 iirc, 32 core words are code 05:59:32 the rest are :'s 06:00:11 wc -l says 59 CODE words 06:00:38 and 3 are macro usage 06:01:48 i have a sneaking suspicion that the way c! is writ is the problem 06:09:34 --- join: madgarden_ (n=madgarde@London-HSE-ppp3546290.sympatico.ca) joined #forth 06:14:37 --- join: nighty_ (n=nighty@CPE00119576a9c5-CM0012c90d36fc.cpe.net.cable.rogers.com) joined #forth 06:21:33 i figered out that eforth filters non-ascii within _type, i killed the filter 06:21:56 ahhh 06:21:58 that should make eforth 8bit clean now except for one thing 06:22:19 input might not be 8bit clean 06:22:26 gotta scope that out 06:22:42 _type had >char in its : 06:27:10 --- quit: madgarden (Read error: 110 (Connection timed out)) 06:32:30 --- quit: madgarden_ (Read error: 110 (Connection timed out)) 07:19:47 --- join: madgarden (n=madgarde@Kitchener-HSE-ppp3576965.sympatico.ca) joined #forth 07:23:26 Quiznos: so, yer fixed now? 07:24:31 I seem to recall a lot of 'type' imps to use 'count' over and over. Simple and cheap, just looked strange. 07:31:10 nah 07:31:37 yea, counted string prefix is a forth staple 07:31:58 and (u a--) is basically what write.2 takes also 07:32:18 lie is simple; life is good 07:33:45 what do they both now look like? 07:34:14 which two? 07:34:23 c@ c! 07:34:27 oh 07:35:29 in pseudo-asm 07:35:40 (thankfully ;-) 07:35:44 heh 07:36:02 dont make me do this pls? 07:36:03 heh 07:36:19 nah, I should you make you write the C ;-) 07:36:40 let's just say that the code is what i call size dependent in the mov's that it uses 07:36:55 that what gonna be pseudo with polishisms 07:37:01 what/was 07:37:04 ewwwwwww 07:37:11 heh 07:37:14 die well, Weedhopper! 07:37:28 but i prefer to write forthy asm code that is size independent 07:37:45 ie, for c@, it's: pop adr; pop (adr) 07:37:49 'forthy' is already taken ;-) 07:37:51 through DI 07:37:55 ahh 07:38:09 you want "forthish" or "forth-like" ;-) 07:38:13 so that there's no size specificed except in the opcode 07:38:27 yeah, that'd be the ABI 07:38:29 and in gas the size is always specified, so i have to think twice 07:38:45 in intel syntax, it's easier to see 07:38:50 pop di, pop [di] 07:38:56 I hate that 07:38:59 or pop byte [di] 07:39:09 and that flips one bit in the opcode 07:39:24 one bit difference tween those two insn 07:39:56 imo, the gas developer sinned by not parsing anything ;) 07:39:58 heh 07:40:10 which is why nasm and fasm are so popular on line 07:40:12 linux 07:40:15 lexing or parsing? 07:40:23 well it's both actually 07:40:27 ugh 07:40:40 as/gas registers are required to have a % prefix 07:40:45 Look at forth assemblers. Have even MORE fun. 07:41:13 i actually have my own syntax that's even terser but more intuitive for certain asmisms 07:41:23 it's more forthy in flavor and sight 07:41:49 I want a cpu-less assembler ;-) 07:42:00 L&P just converted the prevailing syntax of the day and dint give any thought to maintaining forth style IMO 07:42:09 then see what Chuck does with his 07:42:31 forth is the asm on some of his chip designs 07:42:45 otoh, I also want real libs for r/w of .o, .a, shlib and plugins ;-) 07:43:01 to do what he does on non-chuck chips i think one would have to really flesh out the macro concept 07:43:18 well that's easy actually on unix-family osen 07:43:35 yep, you'd think so 07:43:36 eforthl is written in gas style, 07:43:51 but by using gcc instead i get a dynamic binary 07:44:05 and all i'd have to do is add code for libdl 07:44:23 and i've already got a free/PD version of libdl that's pre-gnu i think 07:44:28 sure, because only gcc and friends understand what to build or load 07:44:34 basically yea 07:44:45 It's a right clusterfuck 07:44:46 but for me, libdl is much later 07:45:16 bascially, i wanna redo all the good things of the Unix Tao in forth 07:45:40 I use -ldl fine. Which is why I get so goddamned tired of asm and CC variants that can't even use a generic set of ops and terms. 07:45:51 this idear could develope into a find linux distro for us 07:46:00 Quiznos: yeah. 07:46:11 and I live on a powerbook ;-) 07:46:40 i joke in ##Linux and say "i can imagein a two file distro to be installed\n kernel.tgz and forth.tgz" 07:46:40 heh 07:47:40 right now, i'm trying to determine how much of eforthl can be stripped out and put into extend.bat 07:48:00 then eforthl would be easier to remold i think 07:48:42 I don't touch dos or doze, anymore. Shit is either a Makefile, or a sourcefile. 07:49:48 Datafiles are akin to html, or text, or binaries, or libs - and then you need to dick with file-magic. *sigh* Lords, what a cluster 07:50:19 i still think your vision is too wide (macro) 07:50:58 Perhaps it is Someone has to get dirty and be ridiculed. The earth is round and the sun does NOT rotate around the earth. 07:51:37 wow that's one helluva nonsequiter 07:51:38 heh 07:51:44 earth is an egg 07:51:48 not really... think it thru 07:51:59 altho i like to mess with pedantic scientists and say 07:52:06 "the earth is flat and the heavens are round" 07:52:23 one of em seriuosly freaked when i posted that to him 07:52:34 he asked the others in the room, "he's serious?" 07:52:35 lol 07:52:46 they'd seen that line before and knew i was :) 07:52:51 I'm still glaring at a byte-map table also cogitating libevent 07:52:52 and i can prove it 07:57:25 hmm - hold these thoughts. Lemme' recycle ISP 07:57:37 k 07:57:39 --- quit: PoppaVic ("Pulls the pin...") 07:59:12 --- join: PoppaVic (n=pete@0-2pool198-99.nas30.chicago4.il.us.da.qwest.net) joined #forth 08:00:26 ok, my isp is happy again 08:00:41 k 08:01:10 anyway, yeah - I think we are on parallel paths toward the same painful issue. 08:01:43 it's not so fainful for me; i'm not contemplating the cosmic egg ;) LOLOL roflmao 08:01:44 Or rather... "issues" 08:02:03 Yeah, and I want to die knowing I tried to blaze the path. 08:02:31 sorry you're not the first 08:06:04 I was never that worried - after years of pain, it looks like the issue is still more than moot 08:06:31 are you writing specifically or generally of "issew"? 08:06:35 dont answer specifically 08:06:55 issew? 08:07:01 oh. 08:07:03 lol 08:07:18 i'll make a geek out of you yet 08:07:54 I want to write a portable binary that can use portable source and combine elements of forth, Makefiles, Shell, and autoshit and.. well, you get the idea. 08:08:02 yea 08:08:11 portable binary for what tho? 08:08:22 portable binary.. use java! 08:08:27 portable only applies to some vm 08:08:32 virl said it first 08:08:35 08:08:37 damn 08:08:41 haha.. 08:08:42 to combine those all, stir well. and generate whatever is needed to do whatever. 08:08:53 use the parrot project then 08:08:56 fuck java.. I never install it deliberately. 08:09:06 heh 08:09:14 same with me and msos 08:09:17 cept dos 08:09:21 dos is good 08:09:25 java is soo cool so object oriented and so damn slow.. then you know why you need new hardware... 08:09:25 cept 4.x 08:09:30 I'd use gforth, except it's had a few recent build-issues. 08:09:31 replace the vm then 08:09:38 then use older 08:09:50 wah.. gforth I wouldn't suggest for that.. 08:09:52 see, a vm should be write once and use everywhere 08:10:02 but if it's continually "being fixed" then move on 08:10:07 cept for add-ons 08:10:11 "new hardware" is not something that some silly, half-assed language should require. 08:10:19 true 08:10:40 xell.. is so nice.. 08:11:16 * virl pushes his vapoware 08:11:22 what vapor? 08:11:31 gime 08:11:40 Quiznos: yeah. One portable GCC-program that can do all the above to the degree we can - but NOT as a fly^H^Hull-blown "shell" or "programming language" - Just the facts, mam. 08:11:50 the stuff I'm working on since a year... 08:12:00 XML-forth 08:12:02 ew 08:12:08 * Quiznos shudeers 08:12:17 xml-forth? have you drunk something? idiot? 08:12:21 lol 08:12:28 Quiznos, it's not xml.. 08:12:32 yea yea sure sure 08:12:35 too late 08:12:41 you've been x'd 08:12:42 heh 08:13:02 virl: you love to vascillate between "ripe turd" and "nearly fun" - try to reach at least a middle-ground. 08:13:09 lol 08:13:12 rofl 08:13:13 ok.. when it will be available everybody will see, that it hasn't todo anything with xml 08:13:52 i believe ya 08:13:56 oh man.. why is PoppaVic so stupid.. god! give him a brain 08:14:10 look, dont get me wrong; xml has a porpose 08:14:16 Whatever. It's just that you were pasting XML and defending it for months 08:14:37 virl: yeah, and I often lose it and namecall folks ;-) 08:14:47 lol 08:14:52 PoppaVic, sry, but haven't got my point and it looks like that you never ever get it. 08:14:53 Quiznos: well, it might - but it's really irksome. 08:15:29 PoppaVic only irkers can unirk 08:15:33 virl: yer likely right. I've yet to see a site with a sample or source - and the source is likely asm anyway 08:15:34 lol 08:22:43 so the question of the moment is, how much code can be removed from eforth.S and still have a functioning kernel that will boot and compile extend.bat? 08:23:44 stop that 'bat' shit! BAD LLAMA! 08:23:49 gaah 08:24:13 lol 08:24:22 it's just an -ism for context 08:24:27 Quiznos: yes, the absolute number of words - and primitives - is a good question. 08:24:35 that's what i'm doin atm 08:24:52 I understand, I've been considering it for days 08:25:25 k 08:26:59 Looking at many different sources, the key seems to be a monster Ass-u-me of "cell" and fixed stacks and mem 08:27:56 somethings must be assumed 08:28:03 cellsize is one of them 08:28:18 otherwise you hit a bee's hive. 08:28:21 it's a piss-poor assumption, unless yer using stdint.h 08:28:23 i'd suggest avoiding that 08:28:28 oh yea, you'ce c 08:28:29 ok 08:28:31 you need an ABI first 08:28:58 seriously, i think you've gone overboard with what you're wanting to do 08:29:10 as i said the other day, you need to rebalance 08:29:18 make a concession or two 08:29:20 well, regardless of perverse desires of asm-variants, you still end up in the same damned boat. 08:29:37 sure but i decided that i'm only writing for x86 08:29:43 that was my first decision. 08:29:51 you havent decided that. 08:30:02 then that is yer one and only platform and abi, and me washes me hands ;-) 08:30:04 not even c imp's are portable across venders 08:30:31 GNU has its own variant of c called gcc 08:30:51 the kicker is, yer not even tight on x86 - not when you keep talknig of .bat files ;-) 08:31:03 that wasnt a literal ref 08:31:08 only a context reminder 08:31:09 ahhh 08:31:18 for recall of extending a kernal.com 08:31:28 we all know what i was referring to 08:31:30 with that ref 08:31:34 that's the point 08:32:23 and as i440r said here (or wrote in his package docs) his naming of his kernel to `kernel.com' was in due respect to previous forthers on dos. 08:32:27 a Kudos. 08:32:30 and a thanks. 08:32:56 see, there's a whole thread of thought that is carried by one reference 08:33:13 sure, and I knew .com intimately 08:33:21 exactly, we all do. 08:33:27 us meaning old-enoughs 08:33:40 no, many folks just "assume", but yeah - old-farts would ;-) 08:33:45 yep 08:34:35 i think i had a grand total of only 2 or 3 years on win31 before getting to linux. i have no real clue how to use a winos 08:35:02 first intro to linux was ~'94, first install was in '95; slackware 3.1 08:35:13 well, ok... Let's pause for a sec.. What sorta' plain-text file of notes on "wtf is our min. req. words (prim/colon) do we need?" 08:35:18 i never really got into the whole moose thing 08:35:24 - do you have? 08:35:32 have what? 08:35:46 see above - I lost the thread of thought for a mo 08:35:50 ok 08:35:58 ok 08:36:12 atm i'm just going back and forth in the .S file looking at deps 08:36:23 Yeah... been there myself 08:36:29 ie, words and see are gone 08:36:37 hi is gone but ver is there. 08:36:41 right. Core words 08:36:46 ver is the var for version number 08:37:00 dump and dm+ are gone 08:37:26 words that just twiddle the engine and run local-code (primitives) and (default) secondaries (aka "colonwords") 08:37:32 i pondered rm'ing the definers, : ; var create for a nonasecond :) need those 08:37:38 yea 08:37:44 ? is gone 08:37:56 right, right... Just pause, dude. 08:37:59 i also started renaming certain word names 08:38:00 ok 08:38:47 coffee 08:38:51 THe kick seems to be 100% words that support a default "colondef" interface to the engine; an ultimate interface between that engine and machine. 08:39:08 [and] an 08:42:25 At no time is there a requirement for secondaries to directly-access CPU registers and whatnot (like OS). And, at no time is there a need to add bizarre primitives that let some lunatic access anything outside the engine. This becomes the baseline. 08:42:26 well, there is some fluff 08:42:45 a syscall is needed for linux 08:42:55 or a full assembler in kernel.s 08:43:29 well, writing an "emulator of primitives and secondaries" is sounding better and better. 08:43:54 an emulator? in c you mean? 08:44:21 If your X86 thing requires special asm code, you know how to build and link it - think names and args and stack/ 08:45:15 ok, so, you're saying that the syscal i need would be (and is) the defining word for all syscall words 08:45:28 names, args, stacks, xt's, value's - the real asm underlies the engine; the 2ndries ride above 08:45:37 right, and a primitive 08:45:58 since after the actual int 0x80 the (syscall) word returns the return code that kernel reutrns 08:46:11 it might 08:46:18 here would would have to 08:46:21 as i "see" it 08:46:34 or it passes/fails.. or you code a "throw" and "catch", etc 08:47:05 to me, catch/throw are hll, and not needed in kernel 08:47:09 remember: the primitives are supposed to return 'codes' that matter to the engine and the secondaries 08:47:32 right and (syscall) returns a negated error value when indicated 08:47:44 So, you can't rely on low-level info meaning the same all over. It's all interpreted to one or another extent. 08:48:03 but i've read that errno is actually used in some man2 functions, so i have to figer that out 08:48:17 Imagine writing for X86 linux OR dos OR doze 08:48:31 right, but me is the final arbiter of interpretation 08:48:37 i give context to the code 08:48:52 yer actually targetting a complete cpu/os mess - when you say "x86" 08:49:09 writing for those various os's requires only separation of os-specific code 08:49:14 however 08:49:47 between bare-dos and linux, it would be more than possible to code a truly portable kernel.com i think 08:50:04 not knowing windows, i cant be sure that i could write such a kernel 08:50:15 but the only real diff tween dos and linux is the pmode 08:50:25 and opcodes can be chosen to be pmode-agnostic 08:50:40 which is what i was tellin you about size-dependency earlier 08:52:33 actually, re your [targetting mess] i can see a clear partitioning between writing for dos and linux simultaenously; the only issue is linux binary format and binfmt_flat answers that being equivalent to a dos .com binary image 08:52:55 Sorry, I was cycling laundry - lemme' catch up... 08:52:57 but then there's the whole segment-register thing 08:52:59 ok 08:54:09 I think our entire issue is centered around the actual MC that has to deal with the OS - and the semantics/syntax. 08:54:17 `mc'? 08:54:20 mach code 08:54:31 machine code, yeah - NOT "asm" 08:54:40 os is just another interface; it aint magic :) 08:54:46 the opcodes and registers, 08:54:50 it just uses a different interface 08:54:57 yea 08:55:00 yes, the OS can be simple. 08:55:46 on linux, int 0x80 provides a more regular interface to services; more so than dos even, given that everything goes thro 80h 08:55:53 Anyway we dice it all up, it seems to boil down to a BASIC set of engine, primitives and the acceptable-ABI 08:56:18 i think tween our respectives projects, it'll be easier for me 08:56:54 Quite possible, but I'm not a fool: I can pitch in and think as well as contrib idea, at the least. 08:57:13 absotively 08:57:16 ..basically, you are writing a form of "shell" 08:57:22 and so much more 08:57:50 the entire UNDERLYING "engine" becomes part of the issue, and this all strikes me as "interpreters" 08:58:02 my main complaint against c and everything built within it is the very-low ratio of reuse 08:58:07 memory-elements, sizes, and interpreters 08:58:49 Quiznos: I dunno, I reuse often.. There are many generations of legacy and haqueers and asswipe-profs. 08:59:05 sure 08:59:38 Everyone seems to want to write THE ULTIMATE compact/fast program - in C - and ignore reuse while also ignoring asm 08:59:39 but the latin source for "universe", "university" all come from "one truth" 08:59:45 yep 09:00:23 in this regard, reuse, lisp is truly closer to "holy grail" what with the macro concept it espouses 09:00:35 Personally, I am willing to trade a few clocks for reuse. A few bytes for reuse.. BUT, they even bitch about "compile" versus "assemble" 09:00:36 i want that for forth 09:01:03 sadly, I agree - most of Z's lisper args make me want to scream, but I see his points often. 09:01:12 well, i'm aware of the current netnews thread regarding synonym v. alias and other wods 09:01:29 words; but i think the problem is the ANS that's causing the issew being discussed there. 09:01:47 yea 09:01:51 Yet, lisp strikes me as a cross between a shell, interp, compiler and assembler - all bypassing existing tools. 09:02:01 yep; it is 09:02:13 ANS is cute - it's never been more than cute. 09:02:36 but those things are concretely part and parcel of lisp, and have been for decades 09:02:42 yes 09:02:58 ..which is why I don't bother to learn lisp or use it 09:03:10 i have an orignal copy of the book Lisp 1.5 and even then it had a lap in it 09:03:17 I figure we have tools for reasons - bypassing them is a mess. 09:03:27 i dont use lisp either but i am learning it so i can port certain concepts 09:03:32 same with functional programming 09:03:42 the forthy `joy' language is interesting 09:03:42 yeah, they have some intriguing ideas 09:03:45 yep 09:03:55 lambda's are really interesting 09:03:58 consing 09:04:07 that makes me quiver with possibilities 09:04:28 I never really understand when they throw greek around. 09:04:30 consing would be built ontop of here or H as some forths use it 09:04:35 lol 09:04:40 blame Church 09:04:42 :) 09:04:54 it's just the settled language they use in math 09:05:07 No, I blame haqueers and computer "scientists" 09:05:14 well yea, them too 09:05:20 but that shouldnt be an issew 09:05:32 it's "settled language" for easier porting of concepts. 09:05:34 and that's a good thing 09:06:09 the entire problem still circles the drain around .o static/shared/plugin 09:06:32 only because that's the platform you've chosen to being with. 09:06:36 ie, foundation. 09:06:51 sure, and the assemblers and linkers all use 'em 09:06:53 i'm saying that that foundation maybe too high 09:06:59 for your project 09:07:11