00:00:00 --- log: started forth/05.11.25 00:10:55 --- join: snowrichard (n=richard@adsl-69-155-177-154.dsl.lgvwtx.swbell.net) joined #forth 00:11:29 --- quit: snowrichard (Client Quit) 00:26:04 --- join: snowrichard (n=richard@adsl-69-155-177-154.dsl.lgvwtx.swbell.net) joined #forth 00:27:18 --- quit: snowrichard (Client Quit) 00:38:27 --- quit: homeness (Read error: 110 (Connection timed out)) 01:07:15 --- join: snowrichard (n=richard@adsl-69-155-177-154.dsl.lgvwtx.swbell.net) joined #forth 01:08:12 --- quit: snowrichard (Client Quit) 01:36:46 --- join: snowrichard (n=richard@adsl-69-155-177-154.dsl.lgvwtx.swbell.net) joined #forth 01:37:32 --- quit: snowrichard (Client Quit) 03:18:19 --- quit: Snoopy42 (Read error: 131 (Connection reset by peer)) 03:37:26 --- quit: FiLa (Remote closed the connection) 05:34:20 --- join: PoppaVic (n=pete@0-1pool75-46.nas24.chicago4.il.us.da.qwest.net) joined #forth 05:53:07 --- join: Ray_work (n=vircuser@adsl-66-142-169-152.dsl.rcsntx.swbell.net) joined #forth 05:58:41 --- join: ThinkingInBinary (n=tom@pool-68-163-203-96.bos.east.verizon.net) joined #forth 05:59:08 Quartus: my forth works 05:59:11 Quartus: it's got an interpreter 06:00:14 congrats 06:16:03 PoppaVic: the input buffer system is basically from ANS... i realized I need SOURCE-ID to keep track of what "REFILL" will do, and I need SOURCE (or rather the two pieces of data it outputs) to keep track of what I'm reading... so it worked out. 06:16:28 fun-fun 06:16:38 PoppaVic: yeah 06:16:41 --- nick: ThinkingInBinary -> ashorternick 06:16:51 PoppaVic: have you ever tried irssi? 06:16:56 gods no 06:17:00 PoppaVic: why not? 06:17:14 I want a GUI, and I tolerate xchat 06:17:24 PoppaVic: oh 06:17:34 PoppaVic: yeah I guess irssi gets a 0/10 in the GUI category. 06:27:15 PoppaVic: is it okay for a forth interpreter to explode if you under/over-run either of its stacks? 06:27:41 crash? That's up to you 06:27:47 --- join: snowrichard (n=richard@adsl-69-155-177-154.dsl.lgvwtx.swbell.net) joined #forth 06:27:49 usually, you'd recycle 06:30:56 --- quit: snowrichard (Client Quit) 06:32:20 PoppaVic: like just print an error and restart the interpreter? 06:32:41 after emptying (zeroing the ptr) the stacks, sure 06:33:10 what happens if it occurs as you are defining a new word? 06:33:46 PoppaVic: er, it explodes? 06:33:49 PoppaVic: or maybe 06:33:58 think it thru, look a gforth and such 06:33:59 PoppaVic: the dictionary entry it was writing is still invisible 06:34:32 PoppaVic: ooh, ooh, I know just the phrase for this: "An ambiguous condition exists"! 06:34:39 PoppaVic: ;-) 06:34:57 I'd just make sure the def was lost and the ptr reset in dict 06:38:12 PoppaVic: if I have an interpreter, and I want to load the *rest* of my implementation as text, should I just copy it into a buffer, push the address/length on the stack, and evaluate it? 06:38:31 umm.. 06:38:49 Sounds like an imp-dev decision. I believe FICL does that 06:42:00 --- quit: ashorternick ("Lost terminal") 06:52:05 PoppaVic: Hope your Thanksgiving was good. 06:52:17 that's today for our family ;-) 06:52:24 But, thanks - you as well. 06:52:26 cool :) 06:52:35 Yes, it was a nice, relaxing day. 06:52:45 We went out to dinner, this year. 06:52:49 ahh 06:52:54 We don't live very close to family. 06:53:00 Too damn dangerous to drive around, this week 06:53:06 closest = 9hr drive. 06:53:10 yes. 06:53:53 ouch - 9 hours is extreme 06:54:14 Also, very convenient, if you get my drift... 06:54:26 hmm.. I need a method to ascertain what shared-lib types a platform uses. 06:54:35 Hmm... 06:54:43 np: I get damned tired of sib-squabbling 06:54:58 Yes. 06:55:25 THis month, #3 son is (again) pissed with #2 son.. *sigh* I can't keep track anymore 06:55:25 Would you try to import the lib and "report on failure to load" 06:55:47 Sorry to read that. Happens in the best of families. 06:55:52 well, I'm first thinking of .so (ELF) versus something like .dynlib (Mach-o) 06:56:12 I see. 06:57:03 * Ray_work We have to work today. We won't get many calls or visitors today. I guess Boss just can't seem to bring himself to pay us for a second day off. 06:57:06 hmm 06:58:01 Boss will make a visit, to sign checks, close the cash journal, then he'll go back to his second day off. 06:58:12 bastards 06:58:15 hehe 06:58:24 7 more hours to go. 06:58:32 ok, you on a linux boc? 06:58:35 box? 06:58:36 Ready to talk about nothing for about 7 more hours ? 06:58:44 sure - I doubt family will 06:59:16 I have 2 linux boxes at home, but I use a windows box at work and Raystm2 = XP 06:59:31 Here, I can do.. file -L /usr/lib/libc.dylib 07:00:00 so, it looks like I have to write a multiprong-test and do some parsing 07:00:16 Sounds good. 07:00:36 KISS Principle in action, for this configuration 07:00:40 looks like Jasonwoof and tathi got together this holiday and got fovium and mist working. 07:00:48 Cool 07:01:23 I can't wait to try them. 07:02:01 The drive the SDL, so I'll be trying to code some cool 3d stuff as soon as they can get a working situation out. 07:02:14 s/The/+y 07:02:22 the sdl requires them? 07:02:49 No, I beleive it's the other way around. 07:03:13 oh hey, we got a customer, after all , brb 07:03:36 cool 07:04:44 --- join: skylan_ (n=sjh@dialup-216-211-4-10.tbaytel.net) joined #forth 07:15:56 --- join: snowrichard (n=richard@adsl-69-155-177-154.dsl.lgvwtx.swbell.net) joined #forth 07:16:50 --- join: ashorternick (n=tom@2001:5c0:8bc9:1:20e:35ff:fe48:b958) joined #forth 07:17:17 PoppaVic: is there a traditional name for the actual parser loop? (not EVALUATE or QUIT, but the "read word, compile or execute, repeat" loop that they share) 07:17:34 hmm lemme' look.. 07:17:55 PoppaVic: okay, thanks 07:18:03 PoppaVic: i'm gonna call it _INTERPRET if I don't find a better name 07:18:06 --- quit: skylan (Connection timed out) 07:18:19 PoppaVic: btw, thanks for all your help so far 07:18:36 sent in a msg win 07:19:44 I would seriously suggest picking up the Ting books 07:22:07 PoppaVic: what is deferred? 07:22:16 Yeah, deferred words are like 'forwards' and vectors. 07:22:31 think of "function pointer" 07:23:00 PoppaVic: so a deferred word is like saying "PARSER = just go call INTERPRETER instead"? 07:23:18 well... 07:23:20 PoppaVic: so deferred as in you defer the work to another word, not deferred as in you leave it for later? 07:23:22 DEFER parser 07:23:36 : interpreter ...... 07:23:43 ' interpreter IS parser 07:24:15 right, I believe all deferred words initialize to a scream/reset 07:25:06 now, remember: this means other apps or words can "redirect" the deferred words - all the users of the deferred-word would have no clue 07:26:03 help is no prob.. Just gotta' remember I think C more than forth, and my books are all nearly antiquated references ;-) 07:27:20 PoppaVic: oh, it's a way to make two functions that recursively call each other, without redefining one. 07:27:27 not really 07:27:37 it's a way to write a vector or a forward 07:27:57 THe fact it has multiple uses is a golden plus 07:28:24 the drawback would be at least a double-fetch cycle 07:28:40 what kind of vector are we talking about? 07:28:54 i don't get what you mean by a forward. 07:28:55 umm... When I say "vector" I think of a func-ptr 07:28:59 ok 07:29:20 cuz in @#$% a vector is like a variable-length array (actually it's implemented as an array) 07:29:29 but they call it a Vector. 07:29:54 a 'forward' would mean... I need to define umpty words in terms of 'snarf' - but I need all this shit first (or I expect 'snarf' to point elsewhere occasionally) 07:30:06 yeah,I know 07:30:37 PoppaVic: okay 07:30:51 'defer' and 'is' are almost as powerfully-useful as 'does' 07:30:52 PoppaVic: so how would you implement defer? 07:31:08 very carefully - yer using asm, and I don't 07:31:46 basically, it has to act like the deferred word is executable - even though it's a form of variable 07:32:21 PoppaVic: should i implement it by generating a jump instruction and then filling it in later? 07:32:30 so, whatever it does - it's wangling the "registers" 07:32:51 PoppaVic: and how does IS work? 07:32:59 PoppaVic: does defer put an xt on the stack? 07:33:11 no, before you even define 'defer' and 'is', you'dwant to define a scream & die reset for "default" 07:33:18 PoppaVic: and then IS writes the pointer (in asm or bytecode or whatever)? 07:33:22 PoppaVic: ok 07:33:42 PoppaVic: i'm using @#$% exceptions for those, to put it in native code. 07:33:48 well, remember what I showed you: the tick word IS 07:34:05 you need to install and test with gforth, I'm thinkin' 07:34:40 deferred words are very,very powerful 07:34:56 PoppaVic: the thing is i'm not writing most of the lower-level words in forth, but i need to know how they work 07:34:58 ..iirc, there is also - somewhere - a sort of vtable version 07:35:06 PoppaVic: can you modify the destination of deferred words? 07:35:12 sure 07:35:22 ' now IS was 07:35:51 so that will make "now" call "was"? 07:36:15 there is even a way to get the value FROM the deferred-word, chuck it on the stack - run yer new shit - and restore it. 07:36:32 well, yeah: 'was' would execute 'now' 07:37:14 wait a sec 07:37:31 PoppaVic: i should really give dict entries a "type" field, so i can have code words, lit words, deferred words, native single-bytecode words, etc... 07:38:06 those two should help 07:38:41 I've always felt the "dictionary" was always sorta' busted 07:39:05 cfa: code-field-address; cfa, is a store 07:39:33 absolutely, gforth is a handy experimental platform 07:42:15 PoppaVic: whose dictionary was sorta busted? 07:42:21 all of em 07:42:43 ..either we are not using them well, or they are simplistic - I can't really tell. 07:43:14 ..it's a semi-personal issue - nm 07:43:53 PoppaVic: someone should make a dict with multiple code fields... one for execute, one (optional) for compile, maybe others 07:44:16 fields are easy.. Most folks worry more about ANS and asm 07:44:36 oh, and "optimized" and "speed" 07:45:22 PoppaVic: meh, i have an easy way to make it fast, i just write most words in @#$%. 07:45:31 Personally, I want a forthish system that allows for an interactive forth (shell) and a "compiler" 07:45:54 PoppaVic: what's wrong with colon definitions? 07:46:09 nothing, for interactive development and one-offs 07:46:20 ..I want to integrate my forth with C 07:46:27 PoppaVic: oh i see 07:46:31 PoppaVic: cool 07:46:56 if you can get it from forth to C, and thus to asm - you can create objects and plugsins, and libs 07:46:56 PoppaVic: see bye 07:47:16 see bye 07:47:16 : bye 07:47:16 199448 catch drop bye ; ok 07:47:28 PoppaVic: yeah, it's oddly recursive. 07:47:37 it'd be an earlier word 07:47:47 PoppaVic: mine used a different number before catxh 07:47:52 s/xh/ch/; 07:47:56 sure - different platforms 07:47:56 --- quit: snowrichard ("Leaving") 07:48:04 PoppaVic: ppc? sparc? 07:48:10 powerbook g4 07:48:16 PoppaVic: ah, shiny. 07:48:42 all I can say positive about it is: 1) it works; 2) I don't often dick with admin 07:48:43 PoppaVic: that explains your dislike of terminals and thus irssi, and your mere "tolerance" of the not-Aquafied GUI of xchat 07:48:49 PoppaVic: try Conversation 07:49:01 I hated irssi on my linux box as well 07:49:18 I tried aqua: I hated it 07:49:38 PoppaVic: how do you not use aqua on a mac? 07:49:47 the aqua-xchat 07:49:54 PoppaVic: oh ok 07:50:05 I don't like all the stupid twists and changes from x11 07:50:10 can't you convert a mac over to using x11 07:50:26 saon: X11 runs *under* aqua 07:50:37 I'm running the Z11 stuff now: xchat under x11, everything else is macosx/aqua 07:50:45 x11, I mean 07:50:50 saon: maybe you can totally convert it, but apple's X11.app is an aqua "client" 07:51:08 ah, okay. i don't actually have a mac, i was just curious 07:51:13 PoppaVic: hehe, Z11 is the one we all wanted, with widget-level network communication ;-) 07:51:31 prob with mac/aqua is the damned fools chose obj-c 07:52:06 PoppaVic: yeah i heard someone programming it once, he complained the method names are often huuuuuuuuuuuuuuuuge, like 80 char lines are an impossible dream 07:52:27 PoppaVic: But apparently Apple's APIs are quite rich. 07:52:29 well, that and - I swear to Bog - it looks the preprocessor has a preprocessor 07:52:44 reading objc is godawful 07:52:54 PoppaVic: Yeah, one preprocesses and one inserts the "Steve's Ego" section into the code ;-) 07:53:26 I never understood why the fuck C volken couldn't just write a NEW cpp - and still create C 07:54:31 PoppaVic: is lastxt a variable? 07:54:44 Personally, I've raged against the tide for a decade: My feeling is the asswipe-pundits that write the "standards" should be forced to provide a lex/yacc source for the cpp and c languages 07:54:52 PoppaVic: ;-) 07:54:59 ashorternick: if it isn't, it accesses one 07:55:09 PoppaVic: standards are just pathetic attempts at justifying non-standard behavior. 07:55:19 PoppaVic: ANS "QUIT", for example. dumb name. 07:55:34 well, the C std commies just piss me off - like the ANS forther-volken. 07:55:57 not really.. THey inherited that from the "for(;;)" of FIG, I am sure 07:56:14 PoppaVic: huh? 07:56:21 PoppaVic: what's "for(;;)" gotta do with it? 07:56:23 forever-loop 07:56:32 PoppaVic: oh yeah 07:56:55 "when all else fails" scream and recycle 07:56:59 PoppaVic: is : an immediate word? can you nest definitions? can you have a word that compiles a word that compiles a word that compiles a word? 07:57:14 I dunno why not 07:57:19 PoppaVic: hehe, i can add vibration and sound effects to my error messages? :-) 07:57:36 PoppaVic: can you nest : defs? 07:57:39 Unlike some folks, I always embraced "scream and die" 07:57:45 no. 07:57:50 PoppaVic: I could *literally* make it scream! 07:58:16 PoppaVic: Robots (it's a game that comes with GNOME) has a really annoying bloodcurling-scream sound effect! 07:58:18 folks with gui's will argue, but even in a gui - doze has lead the world in "Scream and die" 07:58:35 PoppaVic: i'll have a native SCREAM bytecode! 07:58:49 ..in a gui, you'd get an alert-dialog, and on closure you'd STILL need to recycle 07:59:23 ashorternick: write an extensible variant on errno/strerror() 07:59:23 --- quit: saon ("brb") 07:59:44 PoppaVic: it's done with @#$% exceptions, they can hold a message 07:59:58 I don't like exceptions and avoid them 08:00:16 PoppaVic: true, but they're much faster and easier for me ;-) 08:00:20 ..I know it's wrong, but I never liked the mechanisms I've been offered 08:00:25 PoppaVic: great, now I have to bleep out @#$% and !@#$%^&*() 08:00:54 PoppaVic: forth rocks 08:00:59 PoppaVic: it's such an elegant language 08:01:08 --- join: saon (i=1000@unaffiliated/saon) joined #forth 08:01:21 saon: welcome back 08:02:58 forth has a lot of interesting features, I agree 08:03:25 PoppaVic: it seemed like the only language simple and small enough to implement a full dev environment in @#$% on a !#@$ %^&*( 08:03:36 otoh, the developers have held it back for decades. 08:03:50 sure - see #openbios 08:04:26 I also believe that FIG was prolly the most portable damned forth for at least a generation 08:04:47 PoppaVic: biosen are insane, there was an article in linux journal about linuxbios and porting it to new chipsets... having to initialize a memory controller using only rom and registers is nuts 08:04:56 PoppaVic: is there a *cowers* spec? 08:05:02 for? 08:05:11 PoppaVic: FIG forth? 08:05:18 the #openbios folks are not only firendly, they are intelligent 08:05:49 FIG never had a "spec" - they wrote a master file and all that needed changes were some very low primitives. 08:06:12 crap, I dunno even how to read or write a spec anymore. 08:06:32 PoppaVic: just describe what all the words do ;-) 08:06:57 ashorternick: that was integral to FIG, and over 90% of the words were written in forth. 08:07:01 PoppaVic: and how the parser works... just go through the program and answer every questoin someone might have 08:07:09 PoppaVic: that's awesome ;-) is it available in source form online? 08:07:25 sure - they still offer FIG elsewhere - it just isn't ANS 08:08:28 THe prob with the Forth Interest Group became the too-pricey membership and the never-ending arguments in the (too infrequent) "newsletter" 08:09:40 PoppaVic: lol 08:10:24 I've still got all my "FORTH Dimensions" from the late 70's - I got sooo goddamned tired of the arguments and "clever people" 08:11:19 otoh, I plan to exhme them to add actual "macros" when I get back to my imp 08:11:30 PoppaVic: what did they argue about? everything? 08:11:30 "exhume" 08:12:11 christ yes, binary - formats, stacks, how controls worked - what stacks... forth-78, 79, fig, f83 - it never ended 08:13:04 about every OTHER issue (and we only got 6/year) had a neat idea. 08:14:50 I can recall first reading about "locals" and "macros". 08:15:01 anyway.. brb - lemme recycle the isp 08:15:05 --- quit: PoppaVic ("Pulls the pin...") 08:18:25 --- join: PoppaVic (n=pete@0-2pool198-32.nas30.chicago4.il.us.da.qwest.net) joined #forth 08:18:49 PoppaVic: welcome back 08:18:50 SO much for THAT sob 08:18:53 thanks 08:19:33 hmm.. It looks likely that my config format needs an if/else 08:20:19 PoppaVic: what config format? 08:20:28 oh.. 08:20:55 http://rafb.net/paste/results/bnCsvE67.html 08:21:16 this format would mean a total code rewrite, but it's looking more useful 08:22:04 PoppaVic: what's it for? 08:22:28 I'm working on a total-system that lets me never again EVER suffer autoshit 08:22:52 the make/shell would be in a forthish 08:23:13 ..combitorial: there is no reason for all the whacked variants 08:23:19 PoppaVic: cool 08:23:54 advisory: the "shell" would not be designed for the same interactivity of bash and sh and whatever 08:24:16 ..I'm targeting the make/sh/m4 stupidities 08:24:38 PoppaVic: a colon does what? parses a word, creates a header for it, sets the xt to the code-space here pointer, and sets state to 1? 08:25:18 creates a TENTATIVE header, and then starts running the equiv of tick and literal 08:25:30 PoppaVic: literal? 08:25:42 : flub 123 .s foo ; 08:25:52 123 is a literal destined for the dstack 08:26:00 OrngeTide: oh ok 08:26:03 PoppaVic: whoops 08:26:05 Hey ashorternick. Sounds good. 08:26:05 OrngeTide: sorry 08:26:08 Quartus: thanks. 08:26:22 Quartus: btw ashorternick = thinkinginbinary but it's shorter for PoppaVic's xchat window :-\ 08:26:27 Quartus: my forth can interpret stuff 08:26:31 This is good. 08:26:33 * PoppaVic appreciates it 08:26:54 PoppaVic: besides, then i can have two of myself if i need to drum up consensus for an idea >:-) 08:26:54 ashorternick, grab the coretest from somewhere; that's your target. 08:27:00 ashorternick: ok 08:27:04 Quartus: ok 08:27:09 Quartus: (talking to myself lol) 08:27:12 heh 08:27:31 ashorternick: yer getting overwhelmed and confuzzled ;-) 08:27:34 Quartus: does qf compile part of its code at startup as forth, or are all core words implemented as native code? 08:27:48 PoppaVic: yeah, giddy with delight. 08:27:56 yer skipping a possibility 08:28:11 PoppaVic: what? 08:28:18 PoppaVic: implementing them as forth but including compiled copies? 08:28:32 PoppaVic: like a binary image of a fully-initialized VM? 08:28:43 you can compile native; you can generate/layout 'forth', and you can feed a stream to the system. 08:28:53 ashorternick, it's all implemented as native code, but much of the native code source is written in a metacompiled Forth. So yes, and no. 08:29:05 PoppaVic: oh, so instead of using a buffer, pretend it's real input? 08:29:09 PoppaVic: that's a good idea 08:29:13 yeah, metacompiling is the "fourth" option 08:29:16 PoppaVic: oh 08:29:34 PoppaVic: well I "metacompile" by calling the native definitions of my forth functions in native code, occasionally taking shortcuts 08:29:41 PoppaVic: so it's basically forth expressed in @#$% 08:29:41 nonono 08:29:49 ashorternick, I do something similar to that. 08:29:51 PoppaVic: it acts just like the real forth, but it's faster 08:30:00 PoppaVic: i don't have to run through the bytecode switch constantly 08:30:02 Q can explain metacompiling better than I, I suspect 08:30:12 Quartus: so i'm on the right track? 08:30:20 ashorternick, if it's working, you're on the right track. :) 08:30:20 --- quit: skylan_ (Connection reset by peer) 08:30:58 THe thing with "metacompiling" is that it is sooo close to "cross compiling" 08:31:10 Well, my build environment is also a cross-compiler. 08:31:20 sure 08:31:30 I'm trying to explain to him, Q 08:31:33 Quartus: how do you hide the new definition of a word while compiling it so you can use the old definition? my dict entries are stored in a @#$% hashtable, so i can't have multiple entries. is that a bad idea? 08:31:55 Quartus Forth itself runs on native 68K Palm devices, and generates native 68K code; but when running under Palm OS 5, which are ARM chips, it still generates 68K code that runs under the PACE emulator on those devices, so it's a native compiler and a cross compiler at the same time. :) 08:31:57 Think in terms of a master HOST and TARGET voc/dict 08:32:18 Quartus: spiffy! 08:32:27 Quartus: no Universal Binaries for Palms yet? ;-) 08:32:28 ashorternick, while defining I'd set the xt to some invalid value in the dictionary header, making that xt result in unfindable entries. Set it right at ; time./ 08:32:43 oddly, the idea of host/target can also be the same machine: just not the same dictionaries. 08:32:50 ashorternick, the 68K code is the closest thing to universal binaries on the Palm. 68K is the VM. 08:33:03 Quartus: I guess this is a forbidden @$%$ question, but how do I keep track of the two different entries in a Hashtable? 08:33:11 Two? Forbidden? 08:33:16 Quartus: or should I *not* add it to the Hashtable until it's done? 08:33:21 Quartus: the old and the new 08:33:34 You can certainly add it to the hashtable at ; time. 08:33:35 Quartus: let's say I do : FOO 2 ; : FOO 1 FOO + ; 08:34:11 Quartus: Right now I keep the dictionary in two parallel ways... a Hashtable for fast word lookup and a Vector for fast "id" (what _header returns) lookup. 08:34:21 ugh 08:34:33 Quartus: so perhaps _header will add it to the vector only, and something like _make 08:34:42 Quartus: ..._make_visible will put it in the hashtable. 08:34:56 hide/show 08:34:58 PoppaVic: hey, remember my architecture before you gag... (never mind, it will make you gag itself) 08:35:01 PoppaVic: ok 08:35:06 ashorternick, it sounds like you're overcompilicating your dictionary, but ok. 08:35:09 PoppaVic: hey, this makes :NONAME words easy! 08:35:29 yeah. You can go "over the top" on the dictionaries 08:35:36 Quartus: nah, it seems to work pretty well... just think of the Hashtable as an optimization. i could iterate backwards through the Vector 08:35:49 ashorternick, while you're building your complicated dictionary, consider wordlists; you may want to build them in now, they're handy. 08:36:02 I would recommend them 100% 08:36:05 Quartus: wordlists = separate, selectable dictionaries? 08:36:15 see ANS and gforth 08:36:18 Conceptually, yes. But they can all be built into one dictionary. 08:36:32 a dictionary would be built upon a wordlist 08:36:52 diction/vocabul 08:37:17 Whatever is simplest, implementation-wise. A wordlist is a single-cell numeric identifier associated with a group of words. An overriding search-order, supporting at least 8 wordlists, controls which wordlists are searched and in what order. 08:37:48 ashorternick: I did a test a few years ago... Using wordlists, I could define words that parsed and looked-up words w/o vocs - interesiting 08:37:56 Quartus: whatever is simplest among what choices? 08:38:21 ashorternick, I'd do it with one dictionary, perhaps prepending the wordlist to the name. 08:38:35 Quartus: wordlists are in DPANS section 16? 08:38:41 I believe that's the section. 08:38:43 Quartus: ok 08:38:48 Quartus: so it's not CORE? 08:38:52 Right. 08:38:54 Quartus: i'll finish CORE first. 08:39:05 But since you're putting so much overhead into the dictionary now, I thought it might be prudent. 08:39:15 Quartus: it can be easily grafted on later ;-) 08:39:34 In Quartus Forth the wordlist number is XOR'd against the hash value, which is very slick; takes no space in the dictionary, and one hashtable serves all. 08:39:34 Quartus: ooh, i can use a space as the separator between wordlist and word, since it's illegal in word names 08:39:46 Quartus: ooh, that's AWESOME! 08:40:03 Quartus: i assume you found a relatively fast hash function? 08:40:12 I did. Quite a simple one. 08:40:52 hmm 08:41:09 It limits the max number of wordlists to the size of the hashtable, but that's not an issue. 08:41:15 Quartus: not a big deal 08:41:40 Think of a voc(wordlist) like this: struct context and members, or class context and members or memb-func. 08:41:51 I find that a confusing way to think of it. 08:41:58 They're just separate namespaces. 08:42:03 right 08:42:12 namespaces are a good way to think of it, i think 08:42:17 they arenot "vocs", but they relate to the contexts 08:42:24 I reserve a few in Quartus Forth; one of them is used to track INCLUDEd filenames, so I can implement NEEDS -- NEEDS won't include a file that's already been included, which is very handy. 08:42:29 PoppaVic: what's a vocabulary that a wordlist isn't? 08:42:58 any voc would hold a wordlist; only special words would have namespaces 08:43:04 Another reserved wordlist tracks module names, for module/private:/public:/end-module. 08:43:12 right, not bad ideas 08:43:49 ashorternick: it would have to do with the "context" - it always does anyway 08:44:07 I fear PoppaVic's explanations may be confusing you, ashorternick -- they confuse me! 08:44:17 Quartus: lol 08:44:25 ..the diff would (generally) be: you can't "extend" those 'namespaces' - if you could, they really should be vocs 08:44:40 sorry, I run in circles and thru the years 08:44:48 Quartus: okay, i have _SHOW and _HIDE now 08:44:57 Quartus: or should they not be "secret"? 08:45:07 ashorternick, I thought you were going to add the header at ; time? 08:45:32 Quartus: yeah, _SHOW will take the "dictionary index" and add its entry to the hashtable. 08:45:34 the ; should be the point where you expose the new stuff 08:45:39 and _HIDE? 08:45:43 Quartus: until _SHOW, the header exists, but is not lookup-able 08:45:48 Quartus: it removes an entry from the hashtable 08:45:52 no 08:45:53 Quartus: just to be symmetric 08:45:55 why do you need that? 08:45:57 Quartus: i dunno 08:45:59 nono 08:46:13 ashorternick, don't add things you don't need :) 08:46:13 PoppaVic: why not? 08:46:15 Quartus: ok 08:46:17 Quartus: gone 08:46:20 THe idea of hide/show is "building - can fail" and "passed - expose" 08:46:39 PoppaVic: but you never explicitly hide things? they're hidden until shown forever? 08:46:51 I am sure some forths use 'hide' for other things, but that is the gist 08:46:53 ashorternick, that's generally the nature of it, yes. 08:46:57 right 08:47:07 Java should have a "secret" access modifier that lets nothing access a method/variable, not even the containing class ;-) 08:47:13 For hiding methods you might want later. 08:47:45 ashorternick, that's Java thinking, hopefully you can rid yourself of it at some juncture :) 08:47:50