2021-06-24 05:12:51 <ttmrichter  > That's the key.  You only pay the cost for the address once: in the token table.
2021-06-24 05:13:16 <ttmrichter  > So if the token is smaller than an address you're saving space (at the expense of speed, typically).
2021-06-24 05:13:33 <ttmrichter  > My STM8 Forth uses token threading because it's just such a cramped space.
2021-06-24 05:15:19 <neuro_sys   > Was it so that token threading limits the number of words you can have?
2021-06-24 05:15:30 <neuro_sys   > To the cell size, that is.
2021-06-24 05:15:55 <neuro_sys   > As in for 1 byte tokens, you could have only 256 words.
2021-06-24 05:17:27 <ttmrichter  > If you're in that cramped a space, that's rarely an issue.  :D
2021-06-24 05:17:36 <ttmrichter  > But there's ways to extend that.
2021-06-24 05:18:11 <ttmrichter  > My tokened Forths are 8-bit tokens, but there are some reserved tokens.
2021-06-24 05:19:31 <ttmrichter  > Essentially F0-FF are hard-coded to words that read the very next token and index into adjunct tables from that.
2021-06-24 05:20:28 <ttmrichter  > Giving me in effect 4096 more words that are slightly slower to access (another layer of indirection).
2021-06-24 05:20:47 <ttmrichter  > I don't think I've ever gone beyond F1, and most of the time I don't even go into F0.
2021-06-24 05:43:28 <neuro_sys   > Makes sense
2021-06-24 05:46:32 <boru        > KipIngram; I have the figForth source (for the 6809) if you want it. It's a scan of a hard listing.
2021-06-24 06:06:27 <cbridge     > <crab> howdy
2021-06-24 06:40:24 <proteusguy  > So in two hours (7:30p GMT+7 BKK Time) we're going to attempt to multi-way live stream a demo of a forth implementation targeting a 6502 running on a javascript emulator. :-)  Read the end of this blog
2021-06-24 09:30:31 <MrMobius    > is there a way to exit a BEGIN/AGAIN loop other than EXIT?
2021-06-24 10:45:36 <crc         > I'm not aware of anything that'd be standard for that
2021-06-24 11:05:59 <KipIngram   > boru: Thanks - that would be cool to have around.  I learned assembly programmingon the 6809; I imagine I remember it fairly well.
2021-06-24 11:06:28 <KipIngram   > proteusguy: How did the stream go?
2021-06-24 11:07:31 <KipIngram   > MrMobius: I'm not sure about standard ways, but it seems like it would be fairly easy to implement a BREAK word.  You're just after a jump to just past the loop?
2021-06-24 11:13:53 <boru        > Heh, I cut my teeth on the 68XX as well.
2021-06-24 11:14:23 <boru        > Let me upload it somewhere.
2021-06-24 11:16:34 <MrMobius    > KipIngram, right. the question though is if that already exists in forth. i dont want to leave it out of the forth im working on if it's something common but seems safe to leave out otherwise
2021-06-24 11:23:11 <cbridge     > -!- ChanServ changed mode/#forth -> +o  proteusguy
2021-06-24 11:27:05 <boru        > KipIngram; https://filebin.net/cmd4jmhyqt2y4225/kip.tar.xz (figForth 6809 source, figForth manual, figForth TinyPascall compiler)
2021-06-24 11:27:55 <boru        > Err, meant to privmsg that. This isn't ##warez...
2021-06-24 13:21:25 <brainfunnel > KipIngram: hmm, in this forth I'm looking at, in the boot.fth it defines CREATE as : CREATE    HEAD 1 1+ , ;  makes me wonder what it'd look like if it weren't token threaded? i suppose simi
2021-06-24 13:21:28 <springb0k   > IBM & CDC were using 'threading' in the 'popular' sense since the '60s.
2021-06-24 13:24:02 <KipIngram   > CREATE itself isn't very hard - it's the "DOES" part of the feature that complicates the implementation and is easier on direct threaded.
2021-06-24 13:24:50 <KipIngram   > springb0k: Well, I take your word for it; I'm not that up on my computing history.  I assume it was Chuck or one of his cronies that coined the Forth usage at some point.
2021-06-24 13:26:04 <KipIngram   > brainfunnel: I once spent a couple of days trying to "generalize" create/does into an arbitrary number of levels, but I didn't really get it to go anywhere very interesting.
2021-06-24 13:26:49 <springb0k   > I'm sure they did.  Just pointing out that the rest of the world didn't take a word from the Forth world and redefine it.  Vice versa.
2021-06-24 13:28:45 <brainfunnel > KipIngram: haha, I suppose thats why "DOES" isn't implemented in this particular boot.
2021-06-24 13:31:05 <KipIngram   > :-)
2021-06-24 13:35:54 <neuro_sys   > From the Wiki article, the PDF is OCR'ed, you can search for "thread". http://web.mit.edu/Saltzer/www/publications/MIT-MAC-TR-030.ocr.pdf
2021-06-24 13:38:16 <neuro_sys   > Although it's also referring to the meaning of the thread in Forth sense I think.
2021-06-24 13:38:31 <neuro_sys   > A metaphor for a sequence of instructions.
2021-06-24 14:12:29 <MrMobius    > KipIngram, interesting. does me do a jump or jsr?
2021-06-24 14:13:13 <MrMobius    > ya Fortran started out in the mid 50s and became threaded at some point
2021-06-24 15:42:18 <KipIngram   > ME does a jump, MrMobius.
2021-06-24 15:42:59 <KipIngram   > Always to the beginning of the most recently compiled word - it's not an "adjustable" jump.
2021-06-24 15:43:22 <KipIngram   > I have it in conditional forms too, like 0=me and <=me and so on.
2021-06-24 15:43:44 <KipIngram   > The same "complete" set of conditionals that I have for conditional single and double returns.
2021-06-24 15:44:06 <KipIngram   > Those items - conditional recursion, conditional return, are really the full extent of my conditional control flow words.
2021-06-24 15:44:13 <KipIngram   > I don't have the classic structures.
2021-06-24 15:44:31 <neuro_sys   > The standards don't make a distinction between colon words and primitives right?
2021-06-24 15:44:46 <KipIngram   > The compiler is smart enough to use a flag in the header to know whether to put an offset to start of word after the word it just compiled, so I don't need immediate words for any of those.
2021-06-24 15:45:00 <KipIngram   > There's not supposed to be a distinction - correct.
2021-06-24 15:45:23 <KipIngram   > That's one of Forth's "graceful features" - that words you add behave in every way like the "built-in" words.
2021-06-24 15:46:45 <neuro_sys   > KipIngram: Is your Forth available online?
2021-06-24 15:58:54 <MrMobius    > KipIngram, so you basically have to make a whole new word for your BEGIN loops?
2021-06-24 16:13:53 <KipIngram   > neuro_sys: No, not yet.  I want to get it to a more complete state first.
2021-06-24 16:14:17 <KipIngram   > MrMobius: Well, I don't HAVE to.  I usually do.  But I can do this if I want to:
2021-06-24 16:14:33 <KipIngram   > : foo ... code ... .: bar ... code me ;
2021-06-24 16:14:42 <KipIngram   > and the loop will go back to bar instead of foo.
2021-06-24 16:14:57 <KipIngram   > By using .: instead of : I can later delete the name bar from my dictionary.
2021-06-24 16:15:07 <KipIngram   > I would usually write that like this, though:
2021-06-24 16:15:12 <KipIngram   > : foo ...code...
2021-06-24 16:15:19 <KipIngram   > .: bar ... code ... me ;
2021-06-24 16:15:57 <KipIngram   > But yes, all of my jumps go to the beginning of the code associated with the most recent name in the dictionary.
2021-06-24 16:16:29 <KipIngram   > I've been using that model for a while  now, and don't find it to be inconvenient in any way.  I'm generally looking for opportunities to factor, and that helps me do that.
2021-06-24 16:22:20 <dave0       > maw
2021-06-24 16:26:17 <crc         > maw
2021-06-24 16:26:23 <dave0       > maw crc
2021-06-24 16:57:06 <KipIngram   > maw dave0
2021-06-24 17:02:09 <dave0       > maw KipIngram
2021-06-24 21:15:05 <cbridge     > <Jo_Anna> hi proteusguy