2026-05-27 01:51:53 <tabemann    > Stalevar, someone's also made a dedicated RPN calculator application for the PicoCalc, if you want something that isn't Forth and is more like a traditional RPN calculator
2026-05-27 01:54:02 <tabemann    > about the PicoCalc, the big thing is that you really should do a lot of research into what you can do with it before you should even consider getting it
2026-05-27 01:55:25 <tabemann    > the software that comes with it is all rather dated and unsupported, I highly recommend replacing the Pico 1H it comes with with a Pimoroni Pico Plus 2W, and you should upgrade the firmware for the STM32 that controls the keyboard, backlight, and power when you assemble yours (if you get one)
2026-05-27 01:56:24 <tabemann    > also, you have to be extremely careful when assembling it to avoid breaking the screen
2026-05-27 01:57:52 <tabemann    > I would say the fragility of the screen is its #1 flaw, but if you assemble it once and keep it assembled from that point on I wouldn't worry about it past the initial assembly process
2026-05-27 02:01:59 <tabemann    > about it being an investment... it of course isn't one in the securities sense (and I'm sure you didn't mean it that way either, despite xentrac's snark), but it's a good choice if you want an open-ended programmable device platform integrating a decent-resolution color screen and a keyboard
2026-05-27 02:02:42 <tabemann    > one important note, though, is that Clockwork is slow about delivering products; I've heard a figure of about three months time recently from order to arrival
2026-05-27 02:02:53 <tabemann    > for the PicoCalc that is
2026-05-27 02:03:03 <tabemann    > the uConsole can take a good portion of a year, OTOH
2026-05-27 02:09:39 <Stalevar    > xentrac, but if I buy this thing I will have it. If not money will be slowly nibbled away by inflation
2026-05-27 02:11:40 <Stalevar    > tabemann, I mean, it kinda seems a bit expensive for the first glance. What is the bill of materials if you order parts yourself rather than buying the kit?
2026-05-27 02:11:59 <Stalevar    > 3D printing the case is relatively cheap, maybe $10
2026-05-27 02:12:39 <Stalevar    > Like, I wanted to estimate if they are ripping people off with their prices or it's quite close to self-cost
2026-05-27 02:14:32 <Stalevar    > tabemann, also I am going to order it from aliexpress or similar, not from CWP, though it'll be more like $130
2026-05-27 02:14:56 <xentrac     > Stalevar: yes, cash in fiat currencies is also a bad investment in many ways, but it is usually not as bad at preserving your capital as spending it on a calculator with low resale value
2026-05-27 02:15:46 <Stalevar    > So it's 1.4 (+40%) of ClockworkPi price
2026-05-27 02:15:55 <Stalevar    > But shipping is going to be about a week
2026-05-27 02:18:03 <Stalevar    > tabemann, I think forth is better than RPN calculator because it has RAM-limited data stack depth while calculators only have four registers
2026-05-27 02:18:52 <Stalevar    > For example you can make sieve of eratosthenes in Forth but can you in RPN calc for PicoCalc?
2026-05-27 02:19:01 <xentrac     > my HP 48GX calculator also had RAM-limited stack depth
2026-05-27 02:19:21 <xentrac     > but you don't want to try to implement the Sieve of Eratosthenes by indexing into the stack
2026-05-27 02:19:24 <Riviera     > Stalevar would probably not call the RPL HP48 "RPN."
2026-05-27 02:19:36 <Riviera     > Rather those like HP42, and similar.
2026-05-27 02:19:40 <xentrac     > RPL does use RPN though
2026-05-27 02:19:54 <Riviera     > xentrac: I am not arguing, just explaining what he likely means.
2026-05-27 02:19:59 <Stalevar    > Hmmm, can you write SoE in RPL?
2026-05-27 02:20:03 <xentrac     > Riviera: understood
2026-05-27 02:20:05 <Riviera     > Of course what he says is wrong, like, obviously, no need to argue about it.
2026-05-27 02:20:12 <Stalevar    > For example I want to know last prime number before 9000
2026-05-27 02:20:17 <xentrac     > yes, you can definitely write SoE in RPL; it has arrays
2026-05-27 02:20:37 <Stalevar    > Of course prime test on odd numbers going down is more efficient
2026-05-27 02:20:47 <xentrac     > maybe
2026-05-27 02:20:52 <Riviera     > There's many people who prefer the 4=level stack of their calculators, and that's what's making the "better" calculator for them, no matter how many prime number sieves they cannot easily program.
2026-05-27 02:20:59 <xentrac     > I'll try it in C
2026-05-27 02:21:01 <Stalevar    > it's probably in range of 8999 to 8899
2026-05-27 02:21:31 <Stalevar    > xentrac, in C it's very easy. But does 48gx run C natively?
2026-05-27 02:21:42 <Stalevar    > or you need to hack it / code on PC and download?
2026-05-27 02:22:23 <Stalevar    > Riviera, because they started with RPN calcs
2026-05-27 02:22:39 <Stalevar    > I have started with GNU dc with however many stack levels you want
2026-05-27 02:22:58 <Riviera     > Stalevar: I do not see how this is a weaker argument than the prime number sieve example you pulled out of your arse.
2026-05-27 02:23:43 <Riviera     > Stalevar: By your logics an ice cream cone is better than your calculators because the ice cream cone can be eaten and enjoyed while the calculators typically can't.
2026-05-27 02:23:50 <Riviera     > Stalevar: anyway, waste of time.
2026-05-27 02:24:33 <Stalevar    > Riviera, what do you mean? We were discussing what firmware might be preferable for PicoCalc. I said forth because it can do arbitrary stuff, just like their default basic, but in RPN
2026-05-27 02:24:39 <Stalevar    > What are you arguing with?
2026-05-27 02:25:27 <Stalevar    > I don't know what kind of RPN calc fw they have for CwPi PC but it's most likely something like Free42, which is cool but probably more limited
2026-05-27 02:27:09 <Stalevar    > https://forum.clockworkpi.com/t/rpn-calculator-for-picocalc-v1/19917/3
2026-05-27 02:27:37 <Stalevar    > Yes, just 100 levels of stack, though I must admit it's more than enough for calculator like applications
2026-05-27 02:30:58 <xentrac     > it's 8999 but on this laptop it takes less than a millisecond of compute
2026-05-27 02:31:09 <xentrac     > I don't have the 48GX anymore
2026-05-27 02:31:47 <xentrac     > given that literally the first number you try is prime I feel like it would be hard for trial division to lose
2026-05-27 02:32:56 <xentrac     > the sieve takes about 62000 CPU instructions more to do 9000 than to do 4
2026-05-27 02:40:47 <xentrac     > trial division only has to do 46 trial divisions for 8999
2026-05-27 02:41:36 <xentrac     > so even for more challenging values of 9000 such as 10000 it beats the sieve handily in finding the last preceding prime
2026-05-27 02:42:12 <tabemann    > back
2026-05-27 02:42:41 <xentrac     > it needs about 2000 amd64 instructions to the sieve's 62000 or so
2026-05-27 02:43:09 <xentrac     > after we try to subtract off the 133000-instruction overhead of just starting a glibc program
2026-05-27 02:43:11 <tabemann    > Stalevar, the screen of the PicoCalc is custom-made -- people have tried to buy it but no one has found a screen quite like it
2026-05-27 02:44:29 <tabemann    > and the keyboard is not a separately-buyable part
2026-05-27 02:45:36 <tabemann    > about the SoE, I agree with xentrac that you don't want to make an SoE by indexing into the stack
2026-05-27 02:46:21 <tabemann    > e.g. the way I've implemented SoE in zeptoforth is by making a bit array in the RAM dictionary where I only store bits for odd numbers
2026-05-27 02:47:24 <xentrac     > compare http://canonical.org/~kragen/sw/dev3/trialdiv.c and http://canonical.org/~kragen/sw/dev3/sieve.c
2026-05-27 02:48:20 <xentrac     > I figured it would be faster to just use bytes rather than try to index a bit array
2026-05-27 02:49:28 <tabemann    > it's faster, but when you're using an RP2040 or RP2350 bit arrays are worth the performance hit (as the CPU is decently fast, but you only get so much SRAM)
2026-05-27 02:50:08 <tabemann    > if I was working in gforth I'd use bytes, but that's because I'd have oodles of RAM at my disposal
2026-05-27 02:50:57 <tabemann    > also, optimizing to only store odd numbers is worth it, as storing even numbers is essentially wasted memory
2026-05-27 02:51:42 <xentrac     > it didn't occur to me how to do the math to do that
2026-05-27 02:51:57 <xentrac     > in some sense storing multiples of 3 or multiples of 5 is also wasted memory ;-)
2026-05-27 02:52:50 <tabemann    > except that testing for evenness is as simple as 1 AND 0= and does not require a modulus operation
2026-05-27 02:54:19 <tabemann    > whereas what you're suggesting adds a significantly greater level of complexity and a definite performance hit (as moduli and divisions are typically much slower than AND's and bitshifts)
2026-05-27 02:54:49 <Stalevar    > xentrac, how do you put stuff onto canonical org?
2026-05-27 02:55:16 <tabemann    > by having bowed down to the greatness that is Mark Shuttleworth?
2026-05-27 02:56:27 <Stalevar    > Also I see there is tcl calc
2026-05-27 02:56:41 <xentrac     > no, by having refused to sell him the domain when he started up his company
2026-05-27 02:56:46 <xentrac     > he's a good guy though
2026-05-27 02:57:06 <Stalevar    > The first program I wrote was calculator, but I put buttons onto word document and typed stuff in VBA editor
2026-05-27 02:57:14 <xentrac     > that's awesome
2026-05-27 02:58:05 <Stalevar    > That's because there was a weird book about computers which my parents bought had VB as only thing which programming was discussed for
2026-05-27 02:58:36 <Stalevar    > But I didn't release you need to install VB separately and thought that VBA IDE was it and was suprised it didn't have compile button described
2026-05-27 03:01:12 <Stalevar    > xentrac, what else would you program for if you have computer just bought from company and it only has windows xp, drivers and MS office??
2026-05-27 03:05:46 <tabemann    > that's kind of like how back in the 1980's everyone had BASIC because that's what came with their machine
2026-05-27 03:06:56 <tabemann    > I have mixed feelings about BASIC because I hated it as a kid (I knew from using LOGO at school and from reading books from the local library that much better things were possible), but now the vast majority of people don't get a prompt at which they can simply start coding anymore
2026-05-27 03:08:07 <xentrac     > Stalevar: hmm, does Windows XP come with DEBUG.COM?
2026-05-27 03:08:45 <highchurch  > tabemann were you a LOGO hacker?
2026-05-27 03:08:53 <Stalevar    > Dunno, I don't have it anymore. When I switched to Linux I did the ritual deletion of Windows XP system files
2026-05-27 03:09:01 <xentrac     > yes: https://en.wikipedia.org/wiki/Debug_(command)#Overview
2026-05-27 03:09:06 <Stalevar    > Like rm -rf /mnt/c:/Windows
2026-05-27 03:09:30 <xentrac     > VBA is more rewarding though
2026-05-27 03:09:46 <xentrac     > Excel is also pretty decent as a programming environment even without VBA
2026-05-27 03:09:53 <Stalevar    > Actually you have VBscript and JScript if you insist programming on bare Windows
2026-05-27 03:10:08 <tabemann    > highchurch: I played around with LOGO and was aware that you could do stuff like it such as define your own procedures/functions, but I never got the fullest out of it because even though I cajoled a summer school teacher out of a LOGO boot disk I didn't really have any documentation
2026-05-27 03:10:50 <xentrac     > I think we could do a lot more to improve programming environments that have a higher ceiling with facilities from Excel.  including, actually, improving VBA itself
2026-05-27 03:11:14 <xentrac     > Windows XP comes with a web browser, so JS is probably the best answer
2026-05-27 03:11:20 <highchurch  > I know very little LOGO, but it's one of those things that seem like there is a branch to an alternate/better present where LOGO was the dominant beginner language
2026-05-27 03:11:23 <tabemann    > my big problem with Applesoft BASIC (my family and my local elementary school had Apples) was that it didn't have structured programming or local variables
2026-05-27 03:11:45 <tabemann    > highchurch: I wish I had Forth as a little kid
2026-05-27 03:11:48 <highchurch  > me too
2026-05-27 03:12:17 <tabemann    > as a kid I basically went BASIC (and a bit of LOGO and assembly) > Hypercard > C++ > C > Python > OCaml
2026-05-27 03:12:19 <xentrac     > highchurch: agreed
2026-05-27 03:12:31 <highchurch  > i dabbled in QuickBASIC as a kid, then only hopped back into programming as a late teen with Common Lisp (which I've had to relearn since then)
2026-05-27 03:12:32 <xentrac     > tabemann: you had better taste than I did!
2026-05-27 03:13:12 <tabemann    > well I lucked out in that the previous owner of the used Mac my family got as our second computer was a programmer, and they gave us *all* their programming stuff
2026-05-27 03:13:26 <highchurch  > been looking into game development, specifically the love2d library for Lua.  But every time i see Lua code (or python) my brain flatlines out of boredom.  Am I the only one who experiences this?
2026-05-27 03:13:35 <tabemann    > so I got things like books on C++ and a C++ IDE and compiler
2026-05-27 03:13:52 <tabemann    > but it took me a while to get into that, so in the meantime I'd muck about with Hypercard on the Mac
2026-05-27 03:13:54 <xentrac     > highchurch: I like Lua
2026-05-27 03:13:56 <highchurch  > that's awesome tabemann
2026-05-27 03:14:10 <highchurch  > xentrac: I *want* to like Lua.  It seems clean and tidy.
2026-05-27 03:14:14 <xentrac     > I liked Python better when it didn't intentionally break backward compatibility every release
2026-05-27 03:14:19 <xentrac     > Lua is a lot messier than Python
2026-05-27 03:14:23 <highchurch  > is it?
2026-05-27 03:14:43 <highchurch  > because you have to shoehorn tables into being all data structures?
2026-05-27 03:15:03 <xentrac     > also you have implicit nil, 1-based addressing, and a couple of other things
2026-05-27 03:15:21 <tabemann    > I like the idea of Lua, but I'm kind of turned off by implicit nil, 1-based indexing, and like
2026-05-27 03:15:23 <highchurch  > haha.  I hack around with APL, so I'm comfy with 1-based
2026-05-27 03:15:27 <xentrac     > shallow equality
2026-05-27 03:15:44 <tabemann    > my favorite lightweight high-level language is Scheme
2026-05-27 03:15:52 <highchurch  > ⎕IO←1
2026-05-27 03:16:08 <xentrac     > e.g. {3}=={3} is false
2026-05-27 03:16:42 <highchurch  > I like how CL has 17 different equality checks.  Wasn't it Sussman who said "equality is a complicated thing"
2026-05-27 03:16:52 <xentrac     > I like Scheme enough to write a self-compiling compiler in it: http://canonical.org/~kragen/sw/urscheme/
2026-05-27 03:16:54 <tabemann    > ideally you support both shallow and deep equality
2026-05-27 03:17:44 <tabemann    > I tried writing a Scheme interpreter in Haskell, but gave up due to the complexities imposed by, well, trying to do all of R7RS-small 'right' in Haskell
2026-05-27 03:17:59 <xentrac     > but I feel like it's a bit clumsier than Lua or Python
2026-05-27 03:18:03 <tabemann    > (I had a big Haskell phase in the mid 2010's)
2026-05-27 03:18:11 <xentrac     > oh, I didn't do all of R5RS, much less do it right
2026-05-27 03:18:28 <xentrac     > I only did the parts the Ur-Scheme compiler needed to compile itself
2026-05-27 03:18:52 <tabemann    > the thing is if you don't try to do Scheme 'right' Scheme is easy-peasy to implement
2026-05-27 03:19:18 <xentrac     > I really like how I can compile standalone executables from Scheme on my phone with Chicken CSC
2026-05-27 03:19:35 <tabemann    > it gets more complicated when you try to implement things like equality of recursive data structures (which R7RS-small implementations are supposed to properly implement!)
2026-05-27 03:19:46 <xentrac     > and efficient continuations
2026-05-27 03:20:19 <tabemann    > I even wrote an algorithm in Haskell for solving that specific problem
2026-05-27 03:20:50 <highchurch  > xentrac is Chicken your preferred Scheme
2026-05-27 03:21:05 <tabemann    > every so often I play around with Racket
2026-05-27 03:21:32 <MrMobius    > Stalevar: TI Nspire and HP Prime are locked down but Casio fc-CG50 is not
2026-05-27 03:21:46 <tabemann    > whereas if I ever write a program in C or C++ and need a high-level language to embed in it I'd probably use Lua
2026-05-27 03:22:21 <tabemann    > just for the record -- Lua's C API is very, very nice, whereas I've dealt with Python's C API and it's one fucking PITA, at least it was in Python 2.x
2026-05-27 03:22:42 <tabemann    > (I've dealt with both in my day jobs)
2026-05-27 03:22:57 <xentrac     > highchurch: it's just the only one that I've figured out how to build executables with on my phone
2026-05-27 03:23:21 <Stalevar    > MrMobius, you cannot buy casio fc-CG50 on aliexpress
2026-05-27 03:23:44 <Stalevar    > but Prime and Nspire are there (though expensive, don't worth it, since PicoCalc cost the same)
2026-05-27 03:24:42 <tabemann    > Stalevar: one advantage of buying your PicoCalc from Clockwork is if you do break your screen assembling it they will send you a new screen for very little
2026-05-27 03:24:49 <Stalevar    > If you chose between PicoCalc, Prime and Nspire and also 84 plus CE, all have color screen and keyboard, what you pick, price difference is negligible, everything about 120-130?
2026-05-27 03:24:53 <tabemann    > and they ship it quickly, much quicker than the PicoCalc itself
2026-05-27 03:26:58 <tabemann    > xentrac, I don't recommend embedding Python in anything
2026-05-27 03:27:15 <tabemann    > embedded Python really sucks to deal with
2026-05-27 03:27:35 <Stalevar    > tabemann, if I break it I can take a picture and send it to seller claiming it was broken during shipping. It's cheating, but it can get you new screen for free
2026-05-27 03:27:40 <MrMobius    > Stalevar: you can't really program an HP-48G in C. Someone tried to make a compiler but it never caught on. You can use UserRPL which is like forth but more sophisticated with types or SysRPL which is much faster but doesn't contain any type checking and may crash the calculator. Also assembly on the calculator or on PC
2026-05-27 03:27:45 <tabemann    > whereas Lua is specifically designed for ease of embedding
2026-05-27 03:27:55 <Stalevar    > If they rip off extra 40% compared to ClockWorkPi, I guess it's OK
2026-05-27 03:28:04 <tabemann    > Stalevar, they'll ask for your order number
2026-05-27 03:28:25 <Stalevar    > tabemann, you send it via platform chat, it has order number
2026-05-27 03:28:49 <Stalevar    > They cannot verify if it's broken during assembly or shipping. But really, I'd rather just not break the screen
2026-05-27 03:29:13 <Stalevar    > Because even free replacement = waiting extra two weeks
2026-05-27 03:30:03 <tabemann    > oh, they don't care if you broke the screen yourself -- just provide the order number and they'll send you a new one for something on the order of $15 USD (with the price including the shipping) right away
2026-05-27 03:30:28 <MrMobius    > Stalevar: you may want to do some research on Pico calc keyboard quality. I don't know anything to say it's bad but one main selling point of some calculators is they are nice to type on
2026-05-27 03:30:29 <tabemann    > however, these days they include precut tape in the package for taping down the screen during assembly to minimize the risk of breakage
2026-05-27 03:30:53 <MrMobius    > whereas many devices like that are more like typing on a remote control
2026-05-27 03:31:34 <Stalevar    > Anyway, I got an RPN calculator already, I'm not certain I will get another one at all or not
2026-05-27 03:31:44 <Stalevar    > MK-52 but dead
2026-05-27 03:31:55 <Stalevar    > I am going to replace capacitors and see if it works
2026-05-27 03:32:31 <tabemann    > MrMobius, I personally don't mind the PicoCalc's keyboard myself, and while some people have had bad stories about the keyboard, oftentimes that's because of software issues with the keyboard interface (and I've personally found that the keyboard works fine *provided* you don't try pressing the keys at weird angles)
2026-05-27 03:32:31 <Stalevar    > I got it in antique store for $6
2026-05-27 03:33:07 <MrMobius    > tabemann: are the keys rubber or hard plastic?
2026-05-27 03:33:28 <tabemann    > whereas my keyboard driver for zeptoforth works quite well (aside from a minor issue that others have reported that once in a very blue moon the keyboard will simply stop responding, requiring a power cycle)
2026-05-27 03:33:31 <Stalevar    > MrMobius, given they are on injection mold, it must be same ABS as the case
2026-05-27 03:33:34 <xentrac     > tabemann: yeah, I've written Python extensions in C, and I understand the embedding API is even worse
2026-05-27 03:34:36 <MrMobius    > Stalevar: hopefully. HP always used plastic then switched to rubber for HP49G which everyone hated so went back to plastic for HP50G
2026-05-27 03:34:42 <tabemann    > MrMobius, the keys are rubber and have a vertical part in the middle that presses on an SMT switch component
2026-05-27 03:34:52 <tabemann    > xentrac, the Lua C API is a breeze in comparison
2026-05-27 03:35:03 <Stalevar    > Hm... Rubber?
2026-05-27 03:35:09 <xentrac     > so I hear but I haven't actually used it
2026-05-27 03:35:23 <MrMobius    > ick
2026-05-27 03:35:24 <Stalevar    > To be honest I don't know if plastic or rubber key are better for pocket device
2026-05-27 03:35:25 <xentrac     > LuaJIT's FFI is out of this world thoug
2026-05-27 03:35:26 <xentrac     > h
2026-05-27 03:35:36 <xentrac     > Stalevar: STAINLESS STEEL
2026-05-27 03:35:45 <Stalevar    > What's the point of making rubber keys at all?
2026-05-27 03:35:54 <Stalevar    > Plastic is usually cheaper?
2026-05-27 03:36:05 <tabemann    > Stalevar, they're one piece
2026-05-27 03:36:20 <tabemann    > if they were plastic they'd have to fabricate lots of little moving parts
2026-05-27 03:36:21 <MrMobius    > Stalevar: if you plan on typing things into it for an hour or two, you want something where you can feel the key register and only have to press once. you don't usually get that with rubber
2026-05-27 03:36:45 <tabemann    > MrMobius, these are better than that
2026-05-27 03:37:08 <xentrac     > how about brass?
2026-05-27 03:37:12 <tabemann    > these have a definite "click" to them, and I've found that they only miss keystrokes when I'm pressing on the keys at a weird angle
2026-05-27 03:37:28 <xentrac     > if you're charging US$90 for the calculator you might as well include some brass in the UI
2026-05-27 03:37:53 <tabemann    > xentrac, what, have a brass key close a circuit electrically?
2026-05-27 03:39:15 <xentrac     > no, just use brass instead of plastic over the rubber domes
2026-05-27 03:39:15 <tabemann    > as I've mentioned before, some people have reported missed keystrokes, but part of that is the software rather than the hardware (both on the Pico side and on the STM32 side -- there are better, newer versions of the STM32 firmware available than that it comes with out of the box)
2026-05-27 03:39:41 <xentrac     > so the keys are shiny and bacteriostatic
2026-05-27 03:39:46 <tabemann    > xentrac, what, for extra bling factor?
2026-05-27 03:39:55 <xentrac     > and durability
2026-05-27 03:40:24 <tabemann    > this doesn't use rubber dones but rather little metal domes that close a connection when pressure is applied
2026-05-27 03:41:02 <xentrac     > that can work, though often those require too much actuation force
2026-05-27 03:41:33 <tabemann    > 
ACTION has never, ever heard of a keyboard with shiny metallic keys, even those über-expensive high-end keyboards that use little magnets that go through coils

2026-05-27 03:43:50 <Stalevar    > tabemann, SwissMicros DM42n
2026-05-27 03:44:00 <Stalevar    > Though it costs like a medium range laptop
2026-05-27 03:44:25 <Stalevar    > But they have stainless steel
2026-05-27 03:47:00 <tabemann    > Stalevar, stainless steel is pretty, but doesn't have the bacteriostatic aspect of xentrac's favored bling
2026-05-27 03:47:52 <MrMobius    > xentrac: the HP48 and similar have metal domes and are about the best calculator you can type on
2026-05-27 03:48:05 <Stalevar    > tabemann, for bacteriostatic, sterling silver is better
2026-05-27 03:48:37 <Stalevar    > And it's not all that expensive, maybe $1 for gram
2026-05-27 03:48:51 <tabemann    > Stalevar, the problem with silver, though, is that it reacts with sulfur compounds in the air
2026-05-27 03:49:01 <Stalevar    > copper alloys darken too
2026-05-27 03:49:23 <tabemann    > e.g. I've seen old silver objects that are simply black
2026-05-27 03:49:24 <Stalevar    > tabemann, silver could be anodized with oxyde layer or something
2026-05-27 03:49:51 <tabemann    > I didn't know you could anodize silver
2026-05-27 03:50:08 <Stalevar    > tabemann, or just heated in oxygen atmosphere?
2026-05-27 03:50:09 <tabemann    > I was familiar with anodizing being primarily an aluminum thing
2026-05-27 03:50:19 <Stalevar    > Whatever works, anyway
2026-05-27 03:51:06 <Stalevar    > tabemann, steel is typically treated with controlled oxidation for example, that's why weapons are black
2026-05-27 03:51:16 <tabemann    > one way or another, I think the $75 USD (that's what it was when I got mine) I spent on my PicoCalc was worth it
2026-05-27 03:51:54 <tabemann    > Stalevar, yeah, blueing and such
2026-05-27 03:52:04 <tabemann    > or phosphating
2026-05-27 03:52:10 <Stalevar    > tabemann, what do you use it for mostly?
2026-05-27 03:52:37 <tabemann    > Stalevar, play around in zeptoforth, write little programs to do random things like that clock example
2026-05-27 03:53:21 <Stalevar    > tabemann, is there any reason to prefer RPN calculator over pocket forth?
2026-05-27 03:53:28 <tabemann    > unlike my other boards I don't need to carry around my laptop to use zeptoforth when I've got it, and it also provides a nice screen and keyboard too
2026-05-27 03:53:39 <Stalevar    > I think forth is already an RPN calculator but with less alien syntax
2026-05-27 03:53:54 <Stalevar    > like space instead of enter
2026-05-27 03:54:19 <tabemann    > RPN calculators are specifically designed for simple calculations, whereas Forth is more suited for writing arbitrary, complex programs
2026-05-27 03:55:02 <Stalevar    > tabemann, but in forth you see entire expression and can edit it until you press enter. And if you don't use if/while/loop you don't need to put it into word either
2026-05-27 03:55:30 <Stalevar    > What's the point of RPN calc if forth does the same thing but better? Besides the fact you can't put forth in calculator in 80s
2026-05-27 03:56:19 <tabemann    > Stalevar, that is true (and in zeptoforth, you can, with some limitations, use conditionals and loops without starting a word)
2026-05-27 03:56:35 <Stalevar    > tabemann, does zeptoforth have readline-like editing or at least cursor keys and insert mode? Or it's as broken as tclsh without rlwrap?
2026-05-27 03:57:20 <tabemann    > Stalevar, not quite out of the box (because readline-like editing breaks mass code uploading, so it's not on by default), but it's trivial to enable -- all you need to do is execute ENABLE-LINE
2026-05-27 03:57:47 <Stalevar    > Then is there a single minus of using forth instead of free42 or whatever?
2026-05-27 03:57:49 <tabemann    > and after you do that you can turn it off again by pressing F1, and re-enable it by, at the start of a line, pressing F2
2026-05-27 03:58:14 <tabemann    > Stalevar: you don't get decimal arithmetic or absolutely huge floats
2026-05-27 03:58:37 <Stalevar    > Hm... but you can implement them, can't you?
2026-05-27 03:58:53 <tabemann    > like that SwissMicros you mentioned comes with decimal and with *quadruple precision floats*
2026-05-27 03:59:06 <tabemann    > quad-fucking-ruple
2026-05-27 03:59:33 <Stalevar    > Yeah, but dc is arbitrary precision. Want 1000 digits of Pi? Easy
2026-05-27 03:59:42 <tabemann    > Stalevar, of course you can implement them if you really feel like it, but Forths normally don't come with it out of the box
2026-05-27 04:00:13 <Stalevar    > tabemann, imagine forth but with dc-like arithmetic
2026-05-27 04:00:22 <Stalevar    > And you can add sin / cos / tan as words
2026-05-27 04:01:08 <tabemann    > Stalevar, if you simply want a really powerful RPN calculator for cheaper than one of those swissmicros, what I'd almost suggest is buy a PicoCalc, but instead of installing a Pico 2 or like, instead follow the directions some people have on the forum and use it with a 3d-printed back and an RPi Zero
2026-05-27 04:01:19 <Stalevar    > So would there be a single downside compared to dm42 if you get that working (math functions and arbitrary precision similar to bc/dc)?
2026-05-27 04:01:20 <tabemann    > and with linux on that RPi Zero use dc
2026-05-27 04:02:27 <tabemann    > Stalevar: there wouldn't be a single downside, except that you'd have to implement all that stuff yourself
2026-05-27 04:02:54 <Stalevar    > tabemann, but can't you extend forth to dc level? I mean does zeptoforth let you change the way numbers are interpreted so if you type 1451..1514 (100 digits) it will put them in stack as is?
2026-05-27 04:04:40 <Stalevar    > I don't know much about forth yet. I know you can redefine + - × ÷ but I don't know if you can change the way literals are interpreted
2026-05-27 04:04:54 <Stalevar    > or you'd need to enter numbers as strings with S" "S ?
2026-05-27 04:05:00 <tabemann    > zeptoforth's basic data type is the cell, i.e. a 32-bit signed or unsigned integer; it also supports 64-bit double cells (signed or unsigned integers), 64-bit S31.32 fixed-point numbers as double cells, optionally S15.16 fixed-point numbers as single cells, and on the RP2350 it also supports single precision floating-point numbers
2026-05-27 04:05:34 <tabemann    > each 'type' has its own set of words, except that addition and subtraction for fixed-point numbers use the same words as addition and subtraction for integers of the same size
2026-05-27 04:06:42 <tabemann    > for instance, if I want to get the square root of 1,000,000 I'd to 1E6 VSQRT V.
2026-05-27 04:07:28 <tabemann    > the V prefix is because F is already taken by S31.32 fixed-point numbers, which were implemented before I got around to supporting floating point, and because the underlying ARM instructions are prefixed with V's
2026-05-27 04:08:35 <tabemann    > however, if I wanted to get the square root of 1,000,000 as an S31.32 fixed-point number, I'd instead to 1000000,0 SQRT F.
2026-05-27 04:08:51 <tabemann    > yes, that's right, the decimal place is marked with a comma
2026-05-27 04:11:09 <Stalevar    > tabemann, how do you change the way plain numbers are interpreted in forth in forth itself?
2026-05-27 04:11:17 <Stalevar    > or you need to edit the kernel?
2026-05-27 04:11:39 <tabemann    > Stalevar, you can register a routine that parses numbers
2026-05-27 04:11:51 <tabemann    > and in practice there is a chain of such routines
2026-05-27 04:11:56 <tabemann    > one calling the next
2026-05-27 04:12:06 <tabemann    > or more like one calling the previous
2026-05-27 04:12:25 <tabemann    > the kernel itself only knows about single-cell integers
2026-05-27 04:12:50 <tabemann    > it does know about prefixes for hex, decimal, octal, and binary numbers though
2026-05-27 04:13:01 <Stalevar    > 89324759020498248095 . -2908961348049509985  ok
2026-05-27 04:13:03 <tabemann    > ($, #, /, and % respectively)
2026-05-27 04:13:07 <Stalevar    > in gforth I get this
2026-05-27 04:13:23 <Stalevar    > Can I write something in gforth to make it remember entire number?
2026-05-27 04:13:36 <tabemann    > that's because you stuffed in a huge number
2026-05-27 04:14:14 <Stalevar    > Yes, I know, it overflows 2^63
2026-05-27 04:14:32 <Stalevar    > My question is what routine or whatever is called when I do that
2026-05-27 04:14:52 <tabemann    > usually Forths don't have overflow detection
2026-05-27 04:15:48 <Stalevar    > what does gforth do when I type a plain integer with just decimal digits and nothing else? What word is executed or whatever? I imagine it's probably in kernel?
2026-05-27 04:15:58 <tabemann    > and this kind of thing is very Forth-specific -- I personally don't know how gforth does it
2026-05-27 04:16:11 <Stalevar    > ok, what does zeptofoth do?
2026-05-27 04:16:34 <Stalevar    > You type digits, press enter. What code is called?
2026-05-27 04:17:06 <tabemann    > it calls the xt in HANDLE-NUMBER-HOOK
2026-05-27 04:17:20 <tabemann    > first it attempts to parse it as a word
2026-05-27 04:17:40 <tabemann    > if it cannot find any word by that name, then it calls that xt to try to parse it as a number
2026-05-27 04:18:34 <Stalevar    > So the repl parser is usually implemented in forth itself?
2026-05-27 04:19:16 <tabemann    > in practice each HANDLE-NUMBER-HOOK routine has a VARIABLE where the previous HANDLE-NUMBER-HOOK routine is saved, which it calls, and then only attempts to parse the number itself if that routine returns FALSE
2026-05-27 04:19:51 <tabemann    > in zeptoforth the core of the REPL is implemented in assembly, but there are a number of hooks that allow portions to be implemented in Forth
2026-05-27 04:19:58 <Stalevar    > OK, so you can redefine xt to parse 1000-digit number like it nothing
2026-05-27 04:20:28 <Stalevar    > I mean, to parse it without overflow
2026-05-27 04:20:28 <tabemann    > Stalevar, the problem, though, is in Forth there is no such thing as a 'type' in the sense of C or Python
2026-05-27 04:20:58 <Stalevar    > tabemann, similar to Tcl I imagine?
2026-05-27 04:21:02 <tabemann    > the closest thing to a 'type' is that many Forths (but not zeptoforth) have special stacks for floating point numbers
2026-05-27 04:21:24 <tabemann    > Tcl operates in terms of strings, Forth operates in terms of cells
2026-05-27 04:21:42 <Stalevar    > in Tcl everything is a string (semantically, internally integers are stored as integers usually, but it's speed hack)
2026-05-27 04:22:15 <Stalevar    > The treatment depends on command word you apply
2026-05-27 04:22:28 <tabemann    > yes
2026-05-27 04:22:29 <Stalevar    > for example word string treats argument as string, word expr as number
2026-05-27 04:22:52 <Stalevar    > So Forth is same, values are untyped, instead words are typed
2026-05-27 04:23:00 <tabemann    > for instance if I want to add two cells I use + whereas if I want to add two double-cells or S31.32 fixed-point numbers I use D+
2026-05-27 04:23:20 <Stalevar    > tabemann, similar to OCaml
2026-05-27 04:23:25 <Stalevar    > you use .+ to add floats
2026-05-27 04:23:36 <tabemann    > yeah, I remember that about OCaml
2026-05-27 04:24:17 <Stalevar    > Also similar to C, char could be number from -128 to 128 or actual character
2026-05-27 04:24:50 <MrMobius    > Stalevar: to offer a different viewpoint on calculators vs forth, an HP48G or similar does allow you to write very long and complex programs and has a programming language like forth but much more sophisticated. I agree that the DM42 programming may not be as good
2026-05-27 04:25:35 <tabemann    > MrMobius, what's your opinion of HP's RPN calculator language whose name I can't remember at this moment?
2026-05-27 04:25:36 <Stalevar    > I have HP 39gs but I haven't tried to program
2026-05-27 04:25:50 <MrMobius    > tabemann: RPL? that's what im refering to
2026-05-27 04:25:55 <tabemann    > MrMobius, yeah that
2026-05-27 04:27:01 <MrMobius    > Stalevar: the advantages of the calculator are portability (which doesnt much matter these days now that there are competing devices), better battery life, better to type on and wont crash under normal operation. you probably dont want to use a regular forth on a pico calc if it crashing is unacceptable
2026-05-27 04:27:24 <MrMobius    > so of course most people dont want or need a calculator these days. it's just a fun thing for 99% of the cases now
2026-05-27 04:28:49 <Stalevar    > MrMobius, and if they do, they usually get simple calculator for $3
2026-05-27 04:29:00 <Stalevar    > I have a couple of those laying around
2026-05-27 04:29:05 <tabemann    > for the record, if one stays away from @ and ! and >R and R> and like and one makes sure to not explode or implode either stack zeptoforth is quite stable
2026-05-27 04:29:49 <Stalevar    > In calculator stack is fixed set of variables so you can't get stack underflow even if you tried
2026-05-27 04:30:08 <Stalevar    > + will pop the numbers and you'll just get zeros at the end
2026-05-27 04:30:23 <tabemann    > of course, I've found the easiest way to require my PicoCalc to be power cycled is by overflowing the stack and having it overwrite the task control block
2026-05-27 04:31:26 <tabemann    > if I'm lucky I can attach a USB connection to my PicoCalc, pull up picocom, and access my emergency-shit-has-gone-down console to postmortem what happened and reboot without a powerr cycle
2026-05-27 04:32:12 <tabemann    > what happens is that I put a guard value at the end of each TCB
2026-05-27 04:32:27 <tabemann    > and when the multitasker is invoked it checks that value to make sure it hasn't changed
2026-05-27 04:33:35 <tabemann    > if it *has* changed it then determines that the system is in an unrecoverable state, and tries to drop down to a limited console accessible over UART0 (which is redirected to the USB-C on the PicoCalc) to give the user a chance to diagnose what happened and reboot
2026-05-27 04:36:45 <tabemann    > MrMobius, in RPL, is the stack limited only by memory, or is it the classic 4-depth stack?
2026-05-27 04:39:14 <Stalevar    > tabemann, I have just checked + behavior on MK-61 emulator. It seems if you do 1 ↑ 2 ↑ 3 ↑ 4 + + + you get 10 as expected, but if you keep using + you get 11, 12, 13 and so on
2026-05-27 04:39:25 <Stalevar    > So final 1 isn't popped for some reason
2026-05-27 04:39:50 <Stalevar    > It stays in the Y (stack variable) forever
2026-05-27 04:40:33 <tabemann    > if you do 1 2 3 4 + + + + in zeptoforth it will yell at you about stack underflow (which is far less fatal in zeptoforth than stack overflow)
2026-05-27 04:41:17 <tabemann    > note that this is only if you do it from the interpreter
2026-05-27 04:41:58 <Stalevar    > $ dc
2026-05-27 04:41:58 <Stalevar    > 1 2 3 4 + + + + p
2026-05-27 04:42:03 <tabemann    > if you do : bad-example ( -- ) 1 2 3 4 begin + again ; in zeptoforth you'll surely crash your system
2026-05-27 04:42:41 <tabemann    > well, that and you then call BAD-EXAMPLE of course
2026-05-27 04:43:57 <Stalevar    > in gforth it doesn't crash even when you underflow though word but it doesn't seem to have again?
2026-05-27 04:44:26 <Stalevar    > : AGAIN
2026-05-27 04:44:27 <Stalevar    >   dest? again-like 4306032 compile, <resolve ;
2026-05-27 04:44:46 <Stalevar    > It shows something for see again, but I don't know what it is
2026-05-27 04:45:17 <Stalevar    > : underflow 1 2 3 4 + + + + + ;  ok
2026-05-27 04:45:17 <Stalevar    > underflow
2026-05-27 04:46:15 <Stalevar    > tabemann, depending on how data stack is implemented, via CPU push/pop or manually via linked list one can probably add code to check undeflow
2026-05-27 04:46:52 <Stalevar    > And probably if you put data stack at the bottom of memory (or top) it will try to write to non-existant memory and won't break things?
2026-05-27 04:47:13 <Stalevar    > How does data stack grow from top to bottom or bottom to top?
2026-05-27 04:48:31 <Stalevar    > What happens on rp2040 when you try to access addresses above physical ram?
2026-05-27 04:48:49 <Stalevar    > Does it have MMU?
2026-05-27 04:50:29 <tabemann    > back
2026-05-27 04:50:31 <Stalevar    > With MMU you can probably put data stack in separate segment so it will be a hardware exception on under/overflow
2026-05-27 04:50:56 <tabemann    > the problem is that on the RP2040 the MPU is very restricted in what it can do
2026-05-27 04:51:17 <tabemann    > e.g. you can only protect units of memory powers of two 256 or larger in size
2026-05-27 04:52:05 <tabemann    > so you'd have to put a 256 byte block of memory between the data and return stacks that would likely be wasted without some rather creative programming
2026-05-27 04:52:25 <tabemann    > and the TCB'd have to take up a power of two 256 bytes or larger in size
2026-05-27 04:52:41 <Stalevar    > 256 bytes isn't that much to care about?
2026-05-27 04:54:13 <tabemann    > it is when you've got 16 tasks and only 264 KiB of SRAM
2026-05-27 04:54:56 <tabemann    > in that case you've wasted 4 KiB just to keep the return stack of each task from overrunning the data stack
2026-05-27 04:55:10 <tabemann    > and another 4 KiB on the extra space to protect all the TCB's
2026-05-27 04:55:43 <Stalevar    > Hm, 16 tasks? I thought you turn it on, get one REPL and that's it?
2026-05-27 04:56:09 <Stalevar    > Or basic IDE if you have default fw
2026-05-27 04:56:10 <tabemann    > the other problem is that these have to be aligned to 256-byte power of two boundaries, which restricts the sizes of the stacks and other things
2026-05-27 04:56:28 <tabemann    > nah, zeptoforth is multitasking and multicore
2026-05-27 04:56:42 <Stalevar    > Do you need that?
2026-05-27 04:56:56 <tabemann    > there's only one *REPL*, but zeptoforth can support many tasks
2026-05-27 04:57:38 <Stalevar    > Maybe there are other forth`s for rp2??0 ?
2026-05-27 04:57:58 <tabemann    > for instance, on the PicoCalc there's one task for the "BIOS" (read: STM32) driver, one for the sound driver, a bunch for the terminal emulator...
2026-05-27 04:58:40 <Stalevar    > By the way I got rp2350 thing on a flash drive-like
2026-05-27 04:58:48 <tabemann    > there are, but zeptoforth is the only one I am aware of that supports multicore or that runs on the PicoCalc (even though last time I checked single-core support for the PicoCalc was in the works for Mecrisp-Stellaris)
2026-05-27 04:59:27 <tabemann    > and IMO running on the RP2040 or RP2350 without multicore is a waste of perfectly good hardware
2026-05-27 04:59:28 <Stalevar    > I have tried to plug it and connect to micropython with picocom but didn't find what to do because it simply gives you same python repl as you get by typing python3 so I found no point and shelved it
2026-05-27 05:00:41 <Stalevar    > I guess I can try to load some forth on it or whatever. I don't know why I bought it at all, but it was relatively cheap ($5)
2026-05-27 05:01:43 <tabemann    > the point of MicroPython is so you can control embedded systems with Python, not to replace Python on your PC
2026-05-27 05:02:47 <tabemann    > oh, BTW, one reason to get a PicoCalc and install a Pimoroni Pico Plus 2W (or failing that, a Pico 2W) is then you get to use WiFi with it
2026-05-27 05:03:03 <tabemann    > as zeptoforth has optional support for either IPv4 or IPv6
2026-05-27 05:05:54 <Stalevar    > tabemann, how do you do that tree thing on PicoCalc which is drawn on the screen in shop title picture?
2026-05-27 05:06:08 <Stalevar    > I guess it's chaotic bifurcation thing but I forgot the formula
2026-05-27 05:06:47 <tabemann    > it's done with a PRNG
2026-05-27 05:07:33 <tabemann    > where I adjust the length of a line and the angles of each branch, which inherits a certain proportion of the length of the parent branch, according to the PRNG
2026-05-27 05:14:13 <Stalevar    > https://github.com/clockworkpi/PicoCalc/blob/master/Bin/PicoCalc%20SD/bifdiag.bas
2026-05-27 05:14:19 <Stalevar    > i think it was this one?
2026-05-27 05:15:24 <tabemann    > oh, that I'm not familiar with
2026-05-27 05:15:43 <tabemann    > I don't think I've even looked at the programs that come with the Pico 1H that comes with the PicoCalc
2026-05-27 05:16:04 <tabemann    > I immediately put a Pico Plus 2W in mine and installed zeptoforth on it
2026-05-27 05:16:58 <tabemann    > (in the few months before my PicoCalc arrived I was busy preparing zeptoforth for it using an emulated setup involving a standalone screen, serial to emulate the keyboard, and an SD card reader)
2026-05-27 05:17:31 <Stalevar    > Wait, it was you who ported zeptoforth?
2026-05-27 05:18:39 <tabemann    > yes
2026-05-27 05:18:53 <tabemann    > I'm the author of zeptoforth
2026-05-27 05:18:58 <tabemann    > the primary author at least
2026-05-27 05:21:01 <Stalevar    > Hm, I guess I can put zeptoforth-1.16.2/bin/1.16.2/rp2350_1core/zeptoforth_kernel-1.16.2.uf2 into my USB thing and see if it's any better than gforth
2026-05-27 05:21:02 <MrMobius    > tabemann: limited only by memory
2026-05-27 05:21:22 <tabemann    > Stalevar, no, don't do that
2026-05-27 05:21:27 <Stalevar    > Why not?
2026-05-27 05:21:50 <tabemann    > the rp2350_1core versions are only special-purpose ones for if you want to run a different application on the second core
2026-05-27 05:21:53 <Stalevar    > It won't overwrite the bootloader
2026-05-27 05:22:15 <Stalevar    > RPI_PICO2-20250415-v1.25.0.uf2 is probably what was on default
2026-05-27 05:22:27 <Stalevar    > So I can restore it any moment
2026-05-27 05:22:32 <MrMobius    > Stalevar: not sure on MK-61 but on old HPs, the last register was "sticky" so it copied the last value it held to the next slot when stack items were dropped
2026-05-27 05:23:13 <Stalevar    > MrMobius, anyway I should try to fix my MK-52 and see in hardware
2026-05-27 05:23:17 <tabemann    > Stalevar, don't worry about overwriting anything as long as you have the original UF2 on hand (and a flash nuke to erase everything before you reinstall the previous UF2 if you so desire)
2026-05-27 05:23:21 <Stalevar    > Its same as MK-61 but with ROM chip
2026-05-27 05:24:13 <tabemann    > Stalevar, what you want is zeptoforth-1.16.2/bin/1.16.2/rp2350/zeptoforth_full_usb-1.16.2.uf2
2026-05-27 05:25:12 <tabemann    > this will give you a complete setup aside from loadable extras such as zeptoed and will give you a USB console
2026-05-27 05:26:47 <tabemann    > unless
2026-05-27 05:26:53 <tabemann    > if you have 16 MiB of flash
2026-05-27 05:26:56 <tabemann    > you want
2026-05-27 05:27:09 <tabemann    > zeptoforth-1.16.2/bin/1.16.2/rp2350_16mib/zeptoforth_full_usb-1.16.2.uf2
2026-05-27 05:27:21 <tabemann    > note that if your device has only 2 MiB of flash it will not be supported
2026-05-27 05:28:17 <Stalevar    > https://ae01.alicdn.com/kf/S93e48c6223284ff29d0e725ed39348b78.jpg
2026-05-27 05:28:25 <Stalevar    > It has 16M flash chip
2026-05-27 05:28:54 <Stalevar    > rp2350 has its own flash with uf2 bootloader afair
2026-05-27 05:30:35 <tabemann    > it has a 32 Mbit flash chip
2026-05-27 05:30:44 <tabemann    > which means 4 MiB
2026-05-27 05:30:57 <tabemann    > the bootloader is in mask rom
2026-05-27 05:31:04 <tabemann    > it cannot be overwritten
2026-05-27 05:32:02 <MrMobius    > Stalevar: I remember MK-52 but never liked the landscape mode
2026-05-27 05:32:51 <Stalevar    > MrMobius, yet if you don't want to lose your program by power cycle it's only option out of soviet stuff, except MK-85, 90, 98, but those are rarity
2026-05-27 05:34:54 <Stalevar    > MicroPython v1.24.0-preview.201.g269a0e0e1 on 2024-08-09; Raspberry Pi Pico2 with RP2350
2026-05-27 05:34:57 <Stalevar    > OK, I got this
2026-05-27 05:35:31 <MrMobius    > I remember the MK-90 too and used to look on ebay but one never came up
2026-05-27 05:35:51 <MrMobius    > I didnt know about MK-98
2026-05-27 05:36:44 <tabemann    > are these old soviet calculators you're speaking of?
2026-05-27 05:37:21 <Stalevar    > Yes
2026-05-27 05:37:32 <Stalevar    > Though MK-90 and 98 more like Russian
2026-05-27 05:37:44 <Stalevar    > because the company existed some time after USSR dissolved
2026-05-27 05:39:37 <tabemann    > lemme guess... did some of these calculators print "yeggog" on an error?
2026-05-27 05:42:26 <MrMobius    > eggog, yes
2026-05-27 05:46:06 <tabemann    > except that to a russian, initial "e" is pronounced "ye"
2026-05-27 05:46:21 <MrMobius    > that's right
2026-05-27 05:47:37 <tabemann    > apparently it was easier for the makers of russian calculators to make their calculators print english "error" than the equivalent in russian
2026-05-27 05:47:50 <tabemann    > but of course, english "error" is meaningless to the average russian
2026-05-27 05:55:08 <Stalevar    > ошибка
2026-05-27 05:55:24 <Stalevar    > I don't think you can do this convincingly on 6 segment
2026-05-27 05:55:48 <tabemann    > yeah
2026-05-27 05:55:56 <Stalevar    > ЕГГ0Г fits perfectly
2026-05-27 05:56:09 <Stalevar    > ОШИБКА still nope
2026-05-27 06:00:15 <tabemann    > I've read that many a russian was bemused by "ЕГГ0Г"
2026-05-27 06:04:41 <MrMobius    > how about "бЛИН?
2026-05-27 06:04:42 <MrMobius    > ?
2026-05-27 06:06:01 <tabemann    > the problem is displaying "l" and making it not look like "p" on an n-segment display
2026-05-27 06:06:28 <tabemann    > and I don't see how you'd print "И" on a 7-segment display
2026-05-27 06:09:17 <MrMobius    > tabemann: it looks like an English U when you write it out
2026-05-27 06:09:43 <tabemann    > MrMobius, I thought that was cursive/italic
2026-05-27 06:10:05 <MrMobius    > you mean cursive that you don't write out?
2026-05-27 06:10:56 <tabemann    > what I mean is that it looks like Latin "U" when it isn't printed
2026-05-27 06:11:43 <tabemann    > and that the form that looks like backwards Latin "N" is the printed non-italic form
2026-05-27 06:12:42 <tabemann    > it feels really... odd to me that you'd use a form that's normally italic or handwritten on the screen of a calculator
2026-05-27 06:13:05 <tabemann    > better to stick with "ЕГГ0Г"
2026-05-27 06:16:57 <tabemann    > okay, I should hit the sack soon
2026-05-27 06:17:54 <MrMobius    > italic?
2026-05-27 06:18:26 <tabemann    > but yeah, Stalevar, flash your board with a "universal flash nuke" (e.g. https://github.com/Gadgetoid/pico-universal-flash-nuke), then install zeptoforth-1.16.2/bin/1.16.2/rp2350/zeptoforth_full_usb-1.16.2.uf2 on it
2026-05-27 06:20:17 <tabemann    > MrMobius, I know it's not called "italic" in Russian; IIRC it's called something like "kursiv" borrowed from German or like
2026-05-27 06:21:35 <tabemann    > Stalevar, once you have done that you will get a /dev/ttyACMx (if you're using Linux) or COMx (if you're using Windows) device that you can connect to with your favorite terminal emulator
2026-05-27 06:22:06 <tabemann    > you won't need to mess with USB-serial dongles or anything of that sort
2026-05-27 06:25:10 <Stalevar    > MrMobius, pancake
2026-05-27 06:25:28 <tabemann    > pancake?
2026-05-27 06:25:54 <MrMobius    > Stalevar: I know but it could also be what you put instead of "ЕГГ0Г" :P
2026-05-27 06:26:48 <Stalevar    > I don't think they would put it on a serious scientific device
2026-05-27 06:27:10 <Stalevar    > Remember MK-52 was professional device, given to ship navigators, cosmonauts and such
2026-05-27 06:27:47 <Stalevar    > For example they carried some MK-52 with space navigation programs on Soyuz spacecraft in case of main computer failure
2026-05-27 06:28:07 <Stalevar    > Apollo astronauts carried some HP calcs afair?
2026-05-27 06:28:51 <Stalevar    > tabemann, I have started with dumping flash in case I ever need to restore it
2026-05-27 07:32:16 <Stalevar    > You won't believe what I found on flash chip from https://ae01.alicdn.com/kf/S93e48c6223284ff29d0e725ed39348b78.jpg this rp2350 flash
2026-05-27 07:33:17 <Stalevar    > If you need assistance with your therapy device contact your healthcare provider or Philips.
2026-05-27 07:33:34 <Stalevar    > Lines like this, from some therapy mask for PAP therapy
2026-05-27 07:33:54 <Stalevar    > Which is supposed to have Wi-Fi and SD card
2026-05-27 09:18:47 <MrMobius    > 0_0
2026-05-27 09:22:03 <veltas      > tabemann: I think americans say "cursive" to mean hand-written words where letters are joined together.  I think in UK it's just called "joined-up writing"
2026-05-27 09:22:20 <veltas      > If it's German then that's not surprising, americans seem to borrow a lot more German words than Brits
2026-05-27 09:24:02 <veltas      > One of the more dominant styles of handwriting was "Italian" or "Italic" which among other things was more slanted (although it was all slanted, because that works better with a quill)
2026-05-27 09:24:41 <veltas      > And I guess the convention to call the slanted font face "Italic" may have not been followed everywhere
2026-05-27 09:24:59 <veltas      > Especially in countries without the same alphabet
2026-05-27 09:26:32 <veltas      > I would guess "Italic" makes sense in the Catholic/Protestant part of the world rather than Orthodox
2026-05-27 09:29:37 <Stalevar    > Also I was pondering if I should reflash the device or preserve for history. I have reflashed and wiped with nuke uf2
2026-05-27 09:29:57 <Stalevar    > Forth works, though I don't know what to do with it yet. I didn't read the docs
2026-05-27 09:30:42 <Stalevar    > 2 3 + . 5  ok
2026-05-27 09:43:14 <veltas      > What more could you want
2026-05-27 09:43:28 <veltas      > You know what Stalevar, that's always my problem with embedded hobby stuff
2026-05-27 09:43:47 <veltas      > I don't have any ideas for projects at all, I always get the kit out and can't think of anything useful
2026-05-27 09:44:09 <Stalevar    > veltas, I have ideas but they are too hard for me to implement
2026-05-27 09:44:17 <veltas      > I've had one good idea but I don't have time for it, which is to write something to translate between the two communication standards used by my boiler and thermostat
2026-05-27 09:44:33 <veltas      > I've moved house since so that's no longer necessary
2026-05-27 09:44:55 <veltas      > And the simple on/off relay worked well enough anyway
2026-05-27 09:45:44 <Stalevar    > I would like to have something like Altair 8800 or IMSAI 8080 but much simpler. Their clones are sold but overly expensive. So my idea was to maybe get Z80, static ram and some chips for front panel
2026-05-27 09:45:57 <veltas      > Awesome
2026-05-27 09:46:15 <veltas      > That sounds cool but would you do more than just a front panel?
2026-05-27 09:46:18 <Stalevar    > Yes, but I bought most of component and never went to actual building
2026-05-27 09:47:12 <Stalevar    > It seems that I'd need some chips as buffers for front panel, but I don't know what exact chips are the best
2026-05-27 09:47:34 <Stalevar    > I have 32k SRAM and Z80 and lots of switches and led
2026-05-27 09:47:54 <veltas      > Why SRAM?
2026-05-27 09:47:58 <Stalevar    > Also Arduino Mega which I planned to use as front panel replacement
2026-05-27 09:48:15 <Stalevar    > temporarily
2026-05-27 09:48:38 <veltas      > And why do you need a buffer for the front panel?
2026-05-27 09:49:08 <Stalevar    > veltas, because you can get 32K in one chip and it's more than enough, the rest 32k could be used for another ram chip or ROM with library of routines
2026-05-27 09:49:37 <Stalevar    > veltas, as far as I get it it's impossible to implement front panel by wiring leds and switches directly to CPU legs
2026-05-27 09:50:30 <Stalevar    > Z80 because it's cheaper and easier to wire than original Intel i8080
2026-05-27 09:50:41 <Stalevar    > And it can run 8080 code
2026-05-27 10:21:39 <veltas      > Okay that makes sense
2026-05-27 10:28:34 <Stalevar    > .s  [ ] ok
2026-05-27 10:28:34 <Stalevar    > : f ( n -- n! ) dup if dup 1 - .s f * else 1 then ;  ok
2026-05-27 10:28:52 <Stalevar    > I can't just get it to calculate simple factorial
2026-05-27 10:29:15 <Stalevar    > I try to print stack before every recursive call and get nonsense
2026-05-27 10:33:34 <veltas      > Does it allow a recursive word that way?
2026-05-27 10:33:42 <veltas      > Not using 'recurse'?
2026-05-27 10:33:50 <Stalevar    > how do I use recurse?
2026-05-27 10:33:57 <Stalevar    > just put it instead of f?
2026-05-27 10:34:04 <veltas      > Yes
2026-05-27 10:34:21 <veltas      > Putting f usually means "call the old def of f" because f isn't visible until ;
2026-05-27 10:37:28 <Stalevar    > The problem is that case 0 breaks the stack : f ( n -- n! ) dup if dup 1 - recurse * else drop 1 then ;  ok
2026-05-27 10:37:45 <Stalevar    > I needed to drop 0 because I don't use it in else branch
2026-05-27 10:41:06 <veltas      > The loop's easier to write
2026-05-27 10:41:13 <veltas      > But I guess that's not the point
2026-05-27 10:43:32 <Stalevar    > Loop is harder to write, but it's not eating return stack
2026-05-27 10:43:49 <Stalevar    > And harder because I don't remember how to do loops in forth
2026-05-27 10:46:29 <veltas      > limit-plus-1 initial ?do i my-step loop
2026-05-27 10:47:24 <veltas      > Something like  : f 1 swap 1 max 1 ?do i * loop ;
2026-05-27 10:47:49 <veltas      > I guess 1+ the limit because it will not multiply by n otherwise
2026-05-27 10:48:13 <veltas      > So   : f 1 swap 0 max 1+ 1 ?do i * loop ;
2026-05-27 10:49:07 <veltas      > Factorial at least you're not worried about eating all return stack because you can't really calculate larger numbers anyway
2026-05-27 10:49:14 <Stalevar    > How do I fetch index variable in ?do loop?
2026-05-27 10:49:32 <veltas      > I
2026-05-27 10:59:45 <Stalevar    > 34 f . 0  ok
2026-05-27 10:59:45 <Stalevar    > 33 f . -2147483648  ok
2026-05-27 10:59:58 <Stalevar    > : f ( n -- n! ) dup 1 ?do i * loop ;
2026-05-27 11:50:17 <veltas      > You need to increment the limit because it stops when I equals the limit
2026-05-27 11:50:39 <veltas      > I guess 33! is too big
2026-05-27 11:50:49 <veltas      > Or 32! even
2026-05-27 12:54:24 <MrMobius    > veltas: I think the point is tabemann doesnt know Russian and briefly googled it without knowing much
2026-05-27 12:56:07 <MrMobius    > Stalevar: 74HC373 or 74HC374 would work well if you want lights on your front panel
2026-05-27 12:56:32 <MrMobius    > or 74HC244 might work if you just want to show address or data bus
2026-05-27 15:04:29 <veltas      > KipIngram: We're talking about OS's like they are this abstraction over the hardware providing portability etc and then we want to do something as simple as getting the CPU temperature and we realise the OS provides almost no value
2026-05-27 15:04:46 <veltas      > It feels like the main business cases for OS's are most optimised and then almost everything else is just a hack
2026-05-27 15:06:01 <veltas      > Entertaining threads in here https://stackoverflow.com/questions/23314886/get-cpu-temperature
2026-05-27 15:22:46 <highchurch  > mornin'
2026-05-27 15:31:10 <veltas      > Morning
2026-05-27 15:31:20 <veltas      > Afternoon here
2026-05-27 15:36:09 <xentrac     > dawn here
2026-05-27 15:37:16 <xentrac     > you wouldn't really want to expose the CPU temperature to unprivileged processes usually
2026-05-27 15:38:31 <xentrac     > (1) it's a sidechannel for wallbanging. (2) it changes slowly enough that the system call overhead is insignificant. (3) it's a source of nondeterminism, and sometimes you want to guarantee reproducible execution
2026-05-27 15:38:45 <xentrac     > EGGOG
2026-05-27 15:40:24 <veltas      > What's wallbanging?
2026-05-27 15:43:18 <veltas      > I'm not penalising Linux here for reporting in sysfs, I'm just thinking it's not well documented how you're meant to get the temp
2026-05-27 15:43:45 <veltas      > The whole sysfs thing seems well designed and just lacks proper documentation, it seems to have just enough docs to be maintained and not really enough to be applied
2026-05-27 15:44:32 <veltas      > The situation on Windows is even worse
2026-05-27 15:45:03 <veltas      > It's all a lot more work than I would expect from what seems like an innocent enough question
2026-05-27 15:51:04 <xentrac     > wallbanging is exfiltrating information through a sidechannel by using lots of CPU
2026-05-27 15:52:37 <xentrac     > the context is that you have some piece of code that is able to observe some piece of confidential information, such as your password, your TLS private key, or your disk encryption key, and you are trying to use OS facilities to prevent it from communicating that information to other processes, so that you don't have to audit its code for backdoors
2026-05-27 15:53:00 <xentrac     > so-called "mandatory access control"
2026-05-27 15:53:04 <xentrac     > or "confinement"
2026-05-27 15:58:58 <xentrac     > hmm, I think https://docs.kernel.org/driver-api/thermal/index.html looks reasonable, and in particular https://docs.kernel.org/driver-api/thermal/sysfs-api.html documents not just how to write the driver but also what it should look like from userspace?
2026-05-27 16:07:37 <veltas      > Then what /sys/class/hwmon/hwmon*/name are you looking for?
2026-05-27 16:09:28 <veltas      > Also that sysfs stuff is buried down the page, it starts presenting itself as how to write drivers, this isn't where people who want to write userspace apps would think to look, but yes it's in there
2026-05-27 16:09:45 <xentrac     > yeah, I found it by grepping
2026-05-27 16:10:18 <veltas      > I did find all this myself and have what works well enough for my case, but it wasn't pretty and what I've got wouldn't work e.g. on AMD
2026-05-27 16:11:42 <veltas      > A more direct way to access for CPU temp on intel is via the coretemp entry in /sys/devices, which is only for Intel-brand CPUs apparently
2026-05-27 16:12:54 <veltas      > If this is documented I don't know where it is, it's easy to find on a live system but what if I want to write it portably for any (or 90% of) x86 CPUs?
2026-05-27 16:13:37 <veltas      > And then it may dawn on baremetal devs it's probably easier to look up where it's kept on most Intel and AMD CPU's, all this abstraction has just obfuscated something that wasn't too hard
2026-05-27 16:20:15 <Stalevar    > veltas, actual limit is much lover, 13! or 14!, I just found it interesting that 34! ≡ 0 (mod 2^32), though it makes sense. 17 even numbers including 8 multiples of 4 including 4 multiples of 8 including 2 multiples of 16 and 32:  17 8 4 2 1 + + + + . 32  ok
2026-05-27 16:22:19 <Stalevar    > veltas, uhm, CPU temperature can be taken from /proc or /sys I think with simple data read?
2026-05-27 16:25:03 <xentrac     > veltas: good point yeah
2026-05-27 16:25:12 <xentrac     > I usually use the `sensors` command
2026-05-27 16:26:13 <xentrac     > strace says this is reading stuff from /sys
2026-05-27 16:28:08 <veltas      > Stalevar: The issue is there's little documentation on how to specifically get CPU temp on a range of machines/arches
2026-05-27 16:29:00 <veltas      > Answer I guess is to read lm-sensors source
2026-05-27 16:29:12 <Stalevar    > 65 f . -9223372036854775808  ok
2026-05-27 16:29:23 <Stalevar    > On big brother I can go up to 65
2026-05-27 16:29:49 <Stalevar    > And then 66! ≡ 0 (mod 2⁶⁴)
2026-05-27 16:36:34 <Stalevar    > : 33f2p 33 16 8 4 2 1 begin + depth 1 = until ; 33f2p . redefined 33f2p  64  ok
2026-05-27 16:39:04 <Stalevar    > I find dc easier, no explicit loops constructs at all, you simply define string with `[ ]` then save in register with `d sr x` and execute with conditionals like >r or quit with conditional and execute with lrx
2026-05-27 16:39:23 <xentrac     > watching the strace may be adequate without reading the source
2026-05-27 16:43:34 <veltas      > It's adequate for getting my own CPU temp, which is just as good as spelunking in my own /sys, and is not helpful to achieve what I've said which is to get the value portably
2026-05-27 16:45:56 <veltas      > But I'm assuming the source is capable of reading it for most x86 CPUs
2026-05-27 16:51:22 <Stalevar    > https://learnxinyminutes.com/files/learnforth.fs this one doesn't have while loop
2026-05-27 16:52:20 <Stalevar    > begin ... while ... repeat
2026-05-27 16:54:35 <xentrac     > Stalevar: maybe read Starting Forth
2026-05-27 16:55:31 <Stalevar    > : sum ( a ... -- a+... ) begin depth 1 > while + repeat ; 33 16 8 4 2 1 sum . redefined sum  64  ok
2026-05-27 16:56:40 <Stalevar    > xentrac, it has lots of fluff which I already know, such as what is reverse polish notation. Such books are for absolute beginners. I already know stack languages just want reminder on syntax
2026-05-27 16:58:50 <xentrac     > maybe Thinking Forth then
2026-05-27 17:05:40 <Stalevar    > xentrac, maybe forth reference manual?
2026-05-27 17:09:33 <Stalevar    > gforth info actually works, seem
2026-05-27 17:10:01 <cleobuline  > https://www.google.com/search?q=mottronix
2026-05-27 17:10:18 <Stalevar    > You can use `THEN' instead of `ENDIF'. Indeed, `THEN' is standard, and `ENDIF' is not, although it is quite popular. We recommend using `ENDIF', because it is less confusing for people who also know other languages (and is not prone to reinforcing negative prejudices against Forth in these people). Adding `ENDIF' to a system that only supplies `THEN' is simple:
2026-05-27 17:10:36 <Stalevar    > That's excerpt from /opt/gforth/share/info
2026-05-27 17:11:08 <Stalevar    > Hm... do you think endif is better than than?
2026-05-27 17:12:50 <Stalevar    > 1 if 2 else 3 endif . unable to parse: endif
2026-05-27 17:12:50 <Stalevar    > 1 if 2 else 3 then . 2  ok
2026-05-27 17:12:58 <Stalevar    > zeptoforth doesn't know endif
2026-05-27 17:13:16 <xentrac     > no
2026-05-27 17:15:06 <Stalevar    > Adding `ENDIF' to a system that only supplies `THEN' is simple:   : ENDIF   POSTPONE then ; immediate
2026-05-27 17:15:57 <xentrac     > yes
2026-05-27 19:48:58 <vms14       > veltas: i made a how-to guide, although is kind of meh
2026-05-27 19:49:04 <vms14       > and there is also a repl :0
2026-05-27 19:49:13 <vms14       > https://gitlab.com/ohmycat/oh/-/blob/main/how-to.md
2026-05-27 19:49:27 <vms14       > it's incomplete
2026-05-27 19:49:39 <vms14       > and i should remove all and try to make a proper tutorial instead
2026-05-27 19:49:59 <vms14       > you can evalute "words" to push a list of word names on the stack, and "r" resets the stack
2026-05-27 19:55:25 <vms14       > the readme explains internals a bit btw