2024-06-02 04:04:11 <zelgomer    > zelgomer> i think i don't like forth's memory model
2024-06-02 04:04:59 <zelgomer    > nevermind, it's fine. my problem is just that i need multiple dictionaries, and i have an idea for that
2024-06-02 04:05:56 <zelgomer    > i think i'm going to use the top four bits of an address as an index into an array of dictionaries that i grow with realloc as you write to them
2024-06-02 04:08:38 <KipIngram   > https://www.youtube.com/watch?v=6OuqnaHHUG8
2024-06-02 04:10:22 <zelgomer    > about to head off but i'll just leave this here in case anyone has any input: the problem is that i'm cross compiling, so i need to be able to stage sections of a target image before i write them to a file. so far i've been getting by with simply "create buf 1000 allot" and then hope my compiled program doesn't exceed 4k, but i'm not very happy with that
2024-06-02 04:11:29 <zelgomer    > so just looking for another way to implement target allot and , and friends without tripping all over the host's dictionary, too. i think carving up the address space is probably a good way to do it.
2024-06-02 04:23:00 <zelgomer    > actually i'm also going to leave with one more thought currently simmering: sort of in the spirit of veltas's "weak" qualifier, i'm thinking of passing options to module loads. maybe something like    " trig vec2d" options " math" load  ( then math.f has some way of testing options and compiles sin, cos, and 2-d vector arithematic words but not 3-d vector or matrix words)
2024-06-02 04:24:26 <zelgomer    > this way i can write a common module for a few apps, but pieces of them don't need to be included in binaries that don't use them
2024-06-02 04:24:46 <zelgomer    > maybe, i don't know. good night.
2024-06-02 14:46:24 <KipIngram   > I just saw a video about "How transistors remember information."  That was slightly annoying somehow - a *transistor* doesn't remember anything.  The memory emerges from the behavior of collections of transistors.  A single transistor just has a state that's a function of its input voltages.
2024-06-02 14:47:11 <KipIngram   > I didn't watch the video - they probably got it right and it was likely just a naming quirk.
2024-06-02 15:18:38 <GeDaMo      > https://www.tomshardware.com/tech-industry/semiconductors/robert-dennard-father-of-dram-is-deceased-also-known-for-his-foundational-dennard-scaling-theory
2024-06-02 15:18:38 <GeDaMo      > "He began researching how he might store a bit of information in one  transistor instead of six. That insight, investigation, and  experimentation would eventually lead to the invention of dynamic  random-access memory (DRAM.)"
2024-06-02 16:14:58 <KipIngram   > Well, that's a fair point, but I went and looked at the video and that's not what they talked about.
2024-06-02 16:15:06 <KipIngram   > They talked about latches and flip flops.
2024-06-02 16:17:09 <KipIngram   > They mention dynamic RAM in the last ten seconds or so, and cite it as using one transistor and a capacitor.
2024-06-02 16:17:31 <KipIngram   > The bulk of the video is a build-up through static RAM.
2024-06-02 16:17:58 <KipIngram   > I would still say that in dynamic RAM it's not the transistor that's retaining the information - it's the capacitor.
2024-06-02 16:19:51 <KipIngram   > Though I imagine there are some clever designs that get them veyr closely related to one another.
2024-06-02 16:53:47 <veltas      > It's a bit like saying 'how people build spaceships', nobody builds a spaceship on their own, they do it together
2024-06-02 16:54:37 <veltas      > And as you said, most of the video is about sram
2024-06-02 17:18:28 <KipIngram   > It was a minor quibble anyway - I just sometimes worry that people wholly unfamiliar with the tech will see things like that title and get the wrong idea.
2024-06-02 17:20:30 <zelgomer    > if we're going to get pedantic, everything has inertia and thus some memory
2024-06-02 17:21:05 <zelgomer    > transistors don't switch instantaneously, they're storing memory!
2024-06-02 18:12:32 <veltas      > KipIngram: Fair
2024-06-02 19:33:37 <KipIngram   > It's like when pop physics people talk about an electron being in "several places at once."  It's not - not really.  They know what they mean and I've figured out what they mean, but I worry that a lot of people will take it too literally.  It feeds right into the whole "quantum woo" thing.
2024-06-02 19:33:54 <KipIngram   > What it REALLY means is that there are several places where a future measurement MIGHT FIND the electron.
2024-06-02 19:34:07 <KipIngram   > But it's wrong to say that it's already in any of those places.
2024-06-02 19:34:44 <KipIngram   > It's not really anywhere until you measure it and force it into a state that actually has a position.  And then it's in that one place.
2024-06-02 19:36:53 <KipIngram   > Feynman used to stress that talking about values you hadn't measured yet was a way to get in the weeds.
2024-06-02 20:35:18 <KipIngram   > veltas: The thumbnail on this one is even worse:
2024-06-02 20:35:20 <KipIngram   > https://www.youtube.com/watch?v=NKfW8ijmRQ4
2024-06-02 20:35:26 <KipIngram   > "How do smartphones think?"
2024-06-02 20:35:31 <KipIngram   > THINK!?
2024-06-02 20:35:38 <KipIngram   > 
ACTION facepalms

2024-06-02 20:40:45 <lf94        > Yo guys :)
2024-06-02 20:41:10 <lf94        > https://gist.github.com/lf94/2700524fa42b76e86b047f35febd6370 I'd love to see everyone try implementing a traffic light state machine
2024-06-02 20:41:34 <thrig       > easy. red all the ways! screw the cars!
2024-06-02 20:43:21 <lf94        > lmao
2024-06-02 20:44:37 <lf94        > for real though :)
2024-06-02 23:36:42 <lf94        > playing with outer interpreter is tough
2024-06-02 23:50:30 <KipIngram   > I find this to be an interesting looking gadget:
2024-06-02 23:51:10 <KipIngram   > Oh, sorry - never mind.  That's a file on my computer.  I don't have the online link anymore.  Anyway, it's a Lattice datasheet for the MachX02 family CPLDs.
2024-06-02 23:52:29 <KipIngram   > lf94: I've never found state machines particularly "graceful" in Forth, but I may just have not thought hard enough about it.
2024-06-02 23:59:39 <lf94        > It seems there are a few nice solutions