2026-04-05 13:03:21 <Guest9579   > hey, I didn't know `#forth' was an exclusive place
2026-04-05 15:20:44 <KipIngram   > My understanding of the # vs. ## thing is that #<something> is reserved for "official projects."  I wasn't aware that there even was a #forth.
2026-04-05 15:39:34 <tpbsd       > I just finished my third test of using AI for my embedded Forth projects
2026-04-05 15:40:31 <tpbsd       > test 1: describe the purpose of some unnamed Forth source: this went well, Id rate it at 90%
2026-04-05 15:42:41 <tpbsd       > test 2: describe the function of a electronics circuit by analysing the allegro netlist from a lepton schematic, I'd rate it at 90% correctless also
2026-04-05 15:44:47 <tpbsd       > test 3: generate a Lepton .sch file of a simple LED light circuit. Big FAIL! the parts, while correct are not connected, their is no meaningful schematic
2026-04-05 15:59:48 <KipIngram   > Based on what I've encountered, you likely can eventually get it to where you want it, but probably after a "struggle."
2026-04-05 16:01:32 <KipIngram   > I've had exactly one stron success with any sort of "vibe development."  I wanted my little ESP32 monitor to be able to dump RAM content in a standard "hex + ascii" pretty printed way, and I asked it to give me a C function for that.  I was able to just drop it in after replacing the printf() statements with my existing "put_line()" function, and it just worked - perfectly.
2026-04-05 16:01:52 <KipIngram   > Other than that one thing it's never failed that I've had to haggle with the thing for hours to get anything close to what I want.
2026-04-05 16:05:52 <tpbsd       > agreed, tests 1 and 2 results were impressive but test 3 was a total fail that demonstrated the AI cant as yet create schematics
2026-04-05 16:06:26 <tpbsd       > it's not like schematics are easy anyway, billions of humans cant do them either
2026-04-05 16:08:17 <tpbsd       > KipIngram, I see people are now trying to get AI to autoroute PCB layouts with zero success so far, but they dont seem to know that pcb autorouter apps have been around for years and are really good, so AI isnt needed there
2026-04-05 16:09:47 <tpbsd       > KipIngram, maybe AI would be better used to improve the pcb autorouter programs themselves ?
2026-04-05 17:08:03 <KipIngram   > Yeah, there's a difference between solving a problem with an actual ALGORITHM vs. just throwing a brute force probabilistic process at it.  Anytime an algorithm is applicable, it's the way to go.  Methods akin to LLMs etc. should always be a "last resort."
2026-04-05 17:08:33 <KipIngram   > But - to create an algorithm a person has to actually know what they're doing.  The dream everyone is chasing is for people to no longer have to know anything themselves.
2026-04-05 17:09:20 <KipIngram   > I don't believe it will ever succeed, but even if it did it would be catastrophic.
2026-04-05 17:09:51 <KipIngram   > Why do we have such a desire to no longer use the one real gift our species was given?
2026-04-05 17:10:15 <iv4nshm4k0v > The dream every rich person is chasing is having profits without any employees.
2026-04-05 17:10:32 <KipIngram   > Yes, that too.
2026-04-05 17:11:03 <KipIngram   > Take that to it's logical extreme.  A handful of people controlling all of the methods of production with no employees.  Who's going to buy the stuff?
2026-04-05 17:11:58 <KipIngram   > Or, rather, how will the people that have to buy the stuff get the money to do so?  Everyone overlooks the "second" function of any economy.  It's purpose is not only to bring goods and services into existence, but also to arrive at some sort of at least reasonable DISTRIBUTION of those goods and services.
2026-04-05 17:12:05 <KipIngram   > It has to do both, or it's a fail.
2026-04-05 17:12:47 <KipIngram   > Note I didn't say "optimum."  We couldn't all agree on what an optimum distribution was anyway.   But it has to do that job at least well enough for the system to slog forward.
2026-04-05 17:14:11 <KipIngram   > I think those businessmen are short-sighted - actually creating a "zero emnployee" economy is basically a perfect formula for unavoidable socialism.
2026-04-05 17:14:58 <KipIngram   > And I doubt that's what those businessmen actually want, so what they're really doing is painting themselves into a corner.
2026-04-05 17:15:22 <KipIngram   > It's a game theory thing - each one of them wants to have no employees HIMSELF while everyone else DOES have emnployees.
2026-04-05 17:16:47 <KipIngram   > It's the same mistake poor people make when they think the government could just give them money.  It works great if the government gives ONE PERSON money - at least great for that person.  But if you give EVERYONE money all you get is higher prices.
2026-04-05 17:16:58 <KipIngram   > Because you didn't magically create more "stuff."
2026-04-05 17:17:36 <KipIngram   > Though you do inject some chaos into the system while the prices adjust.
2026-04-05 17:50:34 <pgimeno     > "<MrMobius> you could use an 8051 if you want lots of RAM"    <- all 128 bytes of it?! please tell me that was sarcasm
2026-04-05 17:50:47 <pgimeno     > and hello everyone again
2026-04-05 18:24:24 <nmz         > So I'm reading this old book by wirth Algorithms+data structures=programs and it calls a queue a (stack)
2026-04-05 18:24:31 <nmz         > but is a queue a stack?
2026-04-05 18:42:10 <iv4nshm4k0v > Queues can be first-in-first-out (FIFO) and last-in-first-out (LIFO.)  Stack (at least as understood by Forth programmers) is the latter one.
2026-04-05 18:42:41 <nmz         > can be or is?
2026-04-05 18:44:22 <nmz         > wikipedia defines a stack as strictly LIFO and queues as FIFO
2026-04-05 18:45:54 <nmz         > this is what I get for reading old books
2026-04-05 19:46:44 <Guest4010   > hey!, I've been looking back in time, and it seems Chuck already used constructions in forth that are not postfix, any idea of why?
2026-04-05 19:47:41 <Guest4010   > I've identified docolon, variable, constant, tick, create, etc.
2026-04-05 21:07:24 <Guest4010   > when a definition is compiled via docol, it's run-time behaviour is stored in the code field, but that same definition would use `does>' where does that code gets stored?
2026-04-05 21:41:16 <pgimeno     > I can talk about the Jupiter. Both the pre-does> and the post-does> code are stored in the parameter field of the word itself. The post-does> code has its own mini-header.
2026-04-05 21:43:11 <Guest4010   > "the jupiter" is some forth implementation?
2026-04-05 21:44:03 <Guest4010   > I'm reading now "create does>" is something introduced by the standard, and that chuck moore doesn't use, so I'm now curious about how he implemented it, where can I find this info?
2026-04-05 21:45:35 <Guest4010   > I actually wonder if is there any paper, talk or whatever where chuck explains it enough so as to implement at least a minimal (and original) forth, even before fig-forth
2026-04-05 23:09:52 <MrMobius    > pgimeno: 8051 is one of the few MCUs with an external bus so you can connect a 512K SRAM or EEPROM