2022-07-09 11:22:40 <KipIngram   > Hey, it just occurred to me that it would be possible to set things up so that you could assign a custom "colon handler" to each vocabularly.  The vocabulary the word went into would determine it's runtime action.  Then you could use : and ; for different purposes.  For example, in vocabulary PRIMITIVES you'd get primitive behavior - you wouldn't need somethiing like :code and ;code anymore.
2022-07-09 11:23:04 <KipIngram   > Anyway, it adds a touch of complexity and doesn't let you do anything you can't otherwise do, but... just an idea that I hadn't thought of before.
2022-07-09 11:23:49 <KipIngram   > It would let you use the same : ... ; format for both host and target in a tethered system, too.
2022-07-09 12:03:50 <KipIngram   > MrMobius: I think I'll amend what I said the other day.  I would consider using mecrisp-across if my goal was to deploy a tiny little micro that was too small to host my own tools.  It's a well-thought out idea.  Of course, I might try to integrate some of those ideas into my own tools, too.  But if I needed to get a tiny micro going *right now*, mecrisp-across looks like a damn good way to do it.
2022-07-09 12:07:31 <KipIngram   > And it's right there, ready to go.
2022-07-09 12:09:26 <KipIngram   > dave0: Hey, how is the next == ret Forth coming along?
2022-07-09 12:23:06 <KipIngram   > I toyed a little just now with what the code just before your definition lists would need to look like; the best I can come up with is something like 'lea <reg1>, [rip+3]; jump <reg2>'.  That would deliver you to your docol code (addressed by reg2), with the address of the list in reg1.
2022-07-09 12:23:38 <KipIngram   > That's 10 bytes; you'd probably want to pad it to 12, and use offset of 5 instead of 3.
2022-07-09 12:24:40 <KipIngram   > This is still the best direct threading idea I've ever seen, I t hink; I've been intrigued by it ever since you first told me about it.
2022-07-09 12:25:35 <dave0       > hey KipIngram
2022-07-09 12:27:19 <dave0       > KipIngram: i got distracted by sectorFORTH   https://github.com/cesarblum/sectorforth/
2022-07-09 12:27:40 <dave0       > KipIngram: i managed to make a 417 byte forth
2022-07-09 12:29:05 <dave0       > jeez it's already 2:30am
2022-07-09 12:32:55 <KipIngram   > Well, sectorforth is also a pretty cool thing.
2022-07-09 12:33:06 <KipIngram   > Where are you?
2022-07-09 12:33:20 <KipIngram   > Down under?
2022-07-09 12:33:46 <dave0       > yup
2022-07-09 12:33:56 <dave0       > 1 hour south of sydney
2022-07-09 12:33:57 <KipIngram   > Given the way computers boot themselves these days, sectorforth is DAMN cool.
2022-07-09 12:34:14 <KipIngram   > I'm not too far from Houston.  11:34am here.
2022-07-09 12:34:26 <dave0       > ah cool, having a good saturday?
2022-07-09 12:34:44 <KipIngram   > So far, yeah.  Still have Monday off, so still kind of basking in vacation-dom.
2022-07-09 12:35:00 <dave0       > ah cool... long weekend!
2022-07-09 12:35:13 <KipIngram   > I was off all last week, and the Friday before, so a big one.
2022-07-09 12:35:36 <KipIngram   > I usually take the week of July 4th off, but this year we got given two extra days as a sort of "bonus" for a project well-executed.
2022-07-09 12:35:44 <KipIngram   > I "tacked them on."
2022-07-09 12:36:12 <dave0       > nice :-)
2022-07-09 12:36:12 <KipIngram   > I also hit 10 years service earlier this year, so I have a whole additional week of vacation that I haven't had in previous years.
2022-07-09 12:36:34 <dave0       > you gonna hang out on your vacation?
2022-07-09 12:36:54 <KipIngram   > Usually after July 4th I'd hunker down until Thanksgiving - that was my longest stretch without any time off.  So I'll probably drop the extra week down in the middle of that somewhere, sometime in September.
2022-07-09 12:37:07 <KipIngram   > Yeah, I like staycations.
2022-07-09 12:37:13 <dave0       > :-)
2022-07-09 12:37:18 <KipIngram   > I'm kind of a home body these days.
2022-07-09 12:38:23 <dave0       > i still walk 6 days a week, but it's nice to have fridays off :-)
2022-07-09 12:38:58 <dave0       > i saw my friends today, they are going good
2022-07-09 12:40:07 <dave0       > my friend jane works a lot she's a nurce, and one of her employees went overseas for 3 months, so friend jane took over doing her job ... so jane is up at 4am every day and in bed by 9pm at the latest
2022-07-09 12:40:11 <dave0       > nurse*
2022-07-09 12:40:38 <dave0       > i forget the name of the employee that went overseas, but she's coming back on the 17th
2022-07-09 12:40:54 <dave0       > which i think is a friday
2022-07-09 12:41:04 <dave0       > might have been the 15th
2022-07-09 12:41:43 <dave0       > but employee probably wont be back to work until monday, so friend jane still has 4am starts for at least another week
2022-07-09 12:58:44 <KipIngram   > Our new kitten is hid]i\
2022-07-09 12:58:50 <KipIngram   > Oops.
2022-07-09 12:58:54 <KipIngram   > Walking on my keyboard, that is.
2022-07-09 12:59:07 <KipIngram   > He was hiding behind the screen assaulting the fingers I stuck around the edges.
2022-07-09 12:59:20 <KipIngram   > Then he hiked around and walked over the keyboard.
2022-07-09 13:03:24 <KipIngram   > After all, it is HIS keyboard of course.
2022-07-09 13:04:23 <KipIngram   > My wife ran across a great dog/cat tidbit the other day:
2022-07-09 13:04:39 <KipIngram   > Dog: They play with me.  They feed me.  They take care of me.  They must be gods.
2022-07-09 13:04:53 <KipIngram   > Cat: They play with me.  They feed me.  They take care of me.  I must be a god.
2022-07-09 13:12:55 <KipIngram   > dave0: Actually regarding that bit of code to nest to new words in the next==ret system, you might be able to look at the code that it jumped to and move some of it to before the jump, instead of padding with unused bytes.  Doesn't save much - it really just saves one copy of that code.
2022-07-09 13:13:07 <KipIngram   > And wouldn't change the timing at all.
2022-07-09 13:13:18 <KipIngram   > But if you were "offended" by padding...
2022-07-09 14:42:09 <imode       > it's probably not available, but do we have schematic diagrams for the layout of the GA144 chip? I'm wondering how this thing does local timing between nodes.
2022-07-09 14:42:41 <imode       > i.e how it's actually asynchronous. what kind of handshaking does it use.
2022-07-09 14:42:44 <imode       > etc.
2022-07-09 14:54:39 <imode       > I guess this is moreso between the F18 and the rest of the chip..
2022-07-09 15:22:03 <KipIngram   > There's a good bit of information about that guy here:
2022-07-09 15:22:05 <KipIngram   > https://etherforth.org/
2022-07-09 15:22:26 <KipIngram   > That's the best docs on it I've seen; it at least shows how that system is laid out across the chip, and talks some about the "ether."
2022-07-09 15:22:34 <KipIngram   > May not be as detailed as you're looking for.
2022-07-09 15:22:43 <KipIngram   > I know the thing is entirely acynchronous.
2022-07-09 15:25:18 <imode       > the F18 core's docs are basically just outlining the block diagram of the chip.
2022-07-09 15:25:23 <imode       > which is pretty bog-standard.
2022-07-09 15:30:28 <KipIngram   > I decided that the best way to learn to program that thing would be by studying actual significant examples; that etherforth system is a good such example.
2022-07-09 15:30:42 <KipIngram   > It strikes me as an entirely different mindset from "ordinary" embedded programming.
2022-07-09 15:33:25 <KipIngram   > In some ways it seems more similar to designing a circuit than to conventional programming.
2022-07-09 15:35:28 <imode       > because of the addition of locality and a limited amount of local resources.
2022-07-09 15:35:40 <KipIngram   > Yeah.
2022-07-09 15:37:45 <imode       > it doesn't feel like programming because we don't program with tons of tiny processors.
2022-07-09 15:37:54 <imode       > or tons of tiny processES.
2022-07-09 15:38:25 <KipIngram   > Yes, exactly right.  But putting tiny bits of logic together into networks is exactly what designing a digital circuit is.
2022-07-09 15:39:11 <imode       > yep. it's more like biology.
2022-07-09 15:40:34 <KipIngram   > It's the kind of way you'd need to think for BitGrid as well.
2022-07-09 15:40:41 <KipIngram   > Except with BitGrid even more, I expect.
2022-07-09 18:39:54 <imode       > I don't think we understand how to program fifty million processors, but that doesn't mean we shouldn't try. scrap whole-system autonomy for local autonomy. agent models.
2022-07-09 18:53:26 <KipIngram   > Yeah - I agree.  It's always good to try.
2022-07-09 18:53:35 <KipIngram   > Sooner or later somebody will figure something out.
2022-07-09 18:54:57 <KipIngram   > At some level something like chemistry will come in - chemists / chemical engineers don't plan reactions at the level of individual molecules.  Well, I mean in a certain way they do - but the large processes are designed around macroscopic variables; temperature, pressure, etc.
2022-07-09 18:57:32 <thrig       > TK421: Why Aren't You at Your Post?
2022-07-09 18:57:33 <KipIngram   > The processes don't "track" individual molecules - that's what I'm trying to get at.
2022-07-09 18:58:10 <KipIngram   > Those macro quantities are controlled, and the details far below left to tend to themselves.
2022-07-09 19:23:47 <imode       > biology has the answers, or at least some parts of them.
2022-07-09 19:24:25 <KipIngram   > Yes.
2022-07-09 19:24:51 <KipIngram   > Once the numbers are big enough.
2022-07-09 19:24:54 <imode       > take a look at this, for example. https://twitter.com/slava__bobrov/status/1545395485053165568?t=zOvYmJjonn-i_8xAG0NyzA
2022-07-09 19:25:28 <imode       > the structural components of a cell.
2022-07-09 19:25:38 <imode       > look at the membrane boundaries.
2022-07-09 19:25:59 <imode       > all of those signalling molecules, drifting along in the cellular membrane.
2022-07-09 19:26:39 <imode       > kay was right in the context of message passing, but that's a small part in the larger assemblage of real-world concurrent systems, i.e biological ones.
2022-07-09 19:27:34 <imode       > membranes as boundaries for computation and resources, bond sites ("ports") that adhere to eachother.
2022-07-09 19:28:06 <imode       > everything is there. communication, isolation and locality.
2022-07-09 19:29:17 <imode       > have you ever taken a look at wang tiling, KipIngram?
2022-07-09 19:33:47 <KipIngram   > Only at a fairly high level.  Scientific American articles and that kind of thing.
2022-07-09 19:34:13 <KipIngram   > Definitely not enough to get at "computational connections."
2022-07-09 19:37:19 <imode       > the basic concept is that you have a set of tiles, and each tiles has a set of edges, and each of those edges have a color/label.
2022-07-09 19:37:41 <imode       > you also have a set of rules that define what edges match to what edges.
2022-07-09 19:37:53 <imode       > red to blue, green to black, orange to orange, etc.
2022-07-09 19:38:59 <imode       > by virtue of those rules, you can assemble a grid of tiles by chucking them all together and watching them bond according to the rules.
2022-07-09 19:39:22 <imode       > turns out that's all you need to be turing complete. you can encode a turing machine in a set of wang tiles.
2022-07-09 19:40:00 <imode       > the state of your turing machine computation is "self-assembling" from the old state.
2022-07-09 19:40:37 <imode       > this is the basis of DNA-based methods for computing.
2022-07-09 19:41:07 <imode       > cells follow this rule. if you examine this: https://www.cellsignal.com/pathways/cellular-landscapes/cellular-landscape-adhesion
2022-07-09 19:41:42 <imode       > two membranes are anchored together, bound by protein structures that are threaded through the membranes.
2022-07-09 19:45:51 <imode       > that link can also facilitate signalling on both sides of the link.
2022-07-09 19:47:56 <KipIngram   > Oh, I didn't know about the Turing complete part - that's fascinating.
2022-07-09 19:48:39 <KipIngram   > So, yeah - that's exactly the sort of thing I think would help us along with astronomically high "core counts."
2022-07-09 19:48:55 <imode       > I've got a model.
2022-07-09 19:49:27 <imode       > software and hardware.
2022-07-09 19:49:31 <imode       > that looks _like that_.
2022-07-09 19:50:47 <imode       > it behaves like that. all of these tiny units of computation, acting locally... your programs need to learn how to walk, like kinase.
2022-07-09 19:51:20 <imode       > sorry, not kinase, kinesin.
2022-07-09 19:51:51 <imode       > https://youtu.be/y-uuk4Pr2i8 it _walks_.
2022-07-09 23:45:45 <KipIngram   > Proteins are *amazing*.  They really are like little machines - I didn't realize how literally for a long time, then I started seeing videos like that.  Little pumps,  motors, etc.  Fascinating.
2022-07-09 23:46:43 <KipIngram   > I read somewhere that if you have a protein that's 150 amino acids long (and that's small for a protein), then only 1 in about 10^77 possible chains actually represents a functional protein.