2026-02-07 02:09:02 is the supervisor something that interrupts to deal with the wireless chip? 2026-02-07 02:09:20 I wonder if KipIngram would like the pi pico 2 better since there's a second core 2026-02-07 02:14:56 MrMobius, depends on his project needs I guess ? 2026-02-07 02:15:32 right. KipIngram mentioned 512K RAM, MB of flash, and wifi so that's why I thought of it 2026-02-07 02:16:55 there are many mcu's with 2 cores and wireless, the stm32wb55 is one for instance 2026-02-07 02:18:05 I know the Pico* is the current focus for a lot of people, but the fact it has external SPI is a show stopper for me 2026-02-07 02:19:11 but a lot of Pico users will only ever have the Pico on a board, premade, and never make their own pcb or buy a standalone RP* mcu chip I suspect 2026-02-07 02:21:06 if one is making a dedicated pcb for a project, the stm32wb55 needs no expernal SPI flash, and has dual cpu, one is a M4 class the other is a M0+ (iirc) dedicated for IO and wireless 2026-02-07 02:22:22 I actually purchased one, with the additional wireless transciever board to evaluate 2026-02-07 02:24:16 ya lots of possibilities if you make your own board 2026-02-07 02:26:36 yeah, and this is the golden age of MCU's, so many low cost ones available 2026-02-07 02:27:17 from Xmos to Parallax with multiple cpu's 2026-02-07 02:28:00 and Xmos has a guaranteed 40ns GPIO max freq output 2026-02-07 02:28:22 so it's a good choice for real time applications 2026-02-07 02:52:56 tpbsd: These ESP32-C6 restrictions you refer to - are these actual restrictions of the hardware, or do you just mean that most vendors have opted to engage secure boot setups in their products as distributed? 2026-02-07 02:53:25 As far as I can tell from an hour or two of perusal, it looks like that's the only way to "lock the thing down." 2026-02-07 02:53:54 If secure boot isn't engaged, you can overwrite the runtime firmware however you like, it seems. 2026-02-07 02:54:48 There's ROM code that runs at boot time, inviolably, but it's just a boot loader as far as I can tell. Sort of a "boot what's there, or download something new" decision maker, and when it's done it retires and takes no further action. 2026-02-07 02:55:24 KipIngram, Ive never used one or looked into it, I'm only parroting what I've read over the years. I think the bootloader is seperate from the hidden supervisor 2026-02-07 02:55:55 The product documentation I've found (for things I can actually buy) is vague - it mentions encryption and secure boot, but it's unclear if they're talking about what they give you or just the fact that those features are there for you to use if you choose. 2026-02-07 02:56:41 and Ive seen at least one DSO pic of the GPIO 'freezing' at a repeatable pace, the poster claimed it was because the supervisor was doing it's thing 2026-02-07 02:57:03 Some of the docs mention two cores - one that can run up to 160 MHz (the "main" core) and then a second that runs at 20 MHz. Maybe this "supervisor" is running on the secondary core on some products? 2026-02-07 02:57:30 And even though they're separate cores, I guess they'd contend for resources. 2026-02-07 02:57:31 yeah, the ESP supervisor is 'hidden' they never talk about it in the litterature 2026-02-07 02:57:54 What is it supervising? 2026-02-07 02:58:22 I dont know, Ive always assumed the subsystems 2026-02-07 02:58:42 The specific one I was looking at is made by an outfit called Seeed Studio. They have a Discord server - I've asked about this there. 2026-02-07 02:58:51 No reply so far, though. 2026-02-07 02:59:53 It's just a little distressing that so few people want to sell you anything in a "Here you go, do what you like" way anymore - everyone wants a hook in you permanently. 2026-02-07 03:00:41 And what's even more distressing is that that fails to produce an absolutely fatal backlash from the market. 2026-02-07 03:01:36 I ordered one. I'll just check and see. It's only $6. 2026-02-07 03:08:20 According to what I've found so far, you can also re-program that second cpu. It's typically used to implement ultra-low power "logic while sleeping." A big part of its role would be to decide if and when to wake up the main cpu. 2026-02-07 03:14:22 KipIngram, my online AI (Kimi-K2) disagrees with me! see https://bpa.st/YS7Q 2026-02-07 03:15:12 KipIngram, I like SeedStudio a lot, Ive purchased some excellent biards from them in the past 2026-02-07 03:17:53 KipIngram, I think marketing has kept pace with technology and is a very polished and well tested art nowdays! 2026-02-07 04:19:40 I have a guess about this. When this product comes out of the box, it's got firmware on it. Includes WiFi and Bluetooth stacks and assorted other "arrangements." I bet that second processor is set up to do something, but the most readily exposed programming interface doesn't deal with it - it's just "out there." Supervising. I think (I hope) in my case, though, I'll be able to take it over 2026-02-07 04:19:42 and decide what to do with it too. Best I can tell it is more like a peripheral that you configure and launch than a fully independent processor - if you don't start it, it won't do anything. So if you don't have an application that needs it for anything, it's easiest just to leave it off. 2026-02-07 04:23:53 KipIngram, well, you can explore that situation now youve ordered one, and be the definitive opinion! 2026-02-07 04:26:54 KipIngram, I do recall the DSO screenpic now, it showed a ESP* GPIO outputting a squarewave, but at regular intervals the waveform was stretched asyncronously and regulary. The poster said it was just simple code toggling the gpio. Ive seen lots of such waveforms when the cpu is 'busy' doing something else on a regular basis 2026-02-07 04:27:44 and it was that screenpic that stuck in my mind at the time, and was supported by claims of this 'hidden supervisor' 2026-02-07 04:28:16 The above would explain that, though - if he wrote the code for that in Python or C running on the stock setup. That second processor could be doing something periodically that stole a little time. 2026-02-07 04:28:23 I'm still not fully convinced that there isnt one, even tho AI's are becer wrong ;-) 2026-02-07 04:28:44 lol... No, of course not. :-) 2026-02-07 04:29:49 thats what I was thinking at the time, it looked like it had a garbage collector running 2026-02-07 04:30:42 I cant recall what type of OS he was running, tho I was only interested in Forth at the time, and so it may have been 'punyforth' 2026-02-07 04:33:13 even tho I have 2x ESP8266 units a friend gave me here, I just cant get enthuiastic enough to test them as Mecrisp-Stellaris on STM32 has everything I need including the stm32wb55 which I also have and havent tested yet 2026-02-07 04:34:03 the stm32wb5has a m4 and M0+ cpu, plus low power wireless 2026-02-07 04:36:32 ... and 'mecrisp-cube' also runs on it https://spyr.ch/twiki/bin/view/MecrispCube by Peter Schmid 2026-02-07 04:37:35 this allows C code and that GUI I/O config thing by for STM to run on it 2026-02-07 06:15:31 veltas: I've been using NetBSD, on an on and off basis, about since 9.0 came out, though my primary system has been Debian GNU/Linux for about two decades now. Being increasingly disappointed with Debian, I'm thinking of adopting pkgsrc as the source of applications on top of either a "base" install of Debian, or perhaps http://tinycorelinux.net/ . 2026-02-07 06:19:00 iv4nshm4k0v, pkgsrc is one of my favorite package managers, I once bootstrapped it to Slackware 2026-02-07 06:19:48 iv4nshm4k0v, I think that pkgsrc is vastly underappreciated as it's amazing watching it at work 2026-02-07 06:25:13 The way I see it, pkgsrc is more of a build system than a package manager. But yes, definitely a software worth looking at. Going to bug wiz for http://pkgsrc.org/wip/ commit access in a couple of weeks. 2026-02-07 06:28:56 iv4nshm4k0v, of course youre right, pkgsrc is far more than a package manager 2026-02-07 06:30:22 I love how it will jump between compiling src and installing binaries as well as trying many urls for a hard to find a specific version of an app 2026-02-07 06:31:06 iv4nshm4k0v, in fact youve inspired me to try a non systemd Linux distro and pkgsrc next 2026-02-07 06:32:56 pkgsrc-wip (work in progress) is a project to get more people actively involved with creating packages for pkgsrc, a portable packaging system coming from NetBSD. It is the default packaging system for NetBSD, SmartOS, and MINIX 3 and has also been ported to many additional operating systems, including Linux, Darwin, FreeBSD, OpenBSD, Illumos, and others. 2026-02-07 06:33:25 wow! it's grown a lot in the 20 years since I last tried it 2026-02-07 06:34:08 Illumos with pkgsrc is enticing 2026-02-07 06:34:13 as is SmartOS 2026-02-07 07:17:25 I didn't get that far with tiny core Linux. it's more of a demo than a real OS in my opinion. You couldn't download gedit for example because of a broken dependency that the package manager can't fix 2026-02-07 07:17:36 The package list is also really short 2026-02-07 07:19:07 I was interested in creating VMs for various architectures with an assembler and gdb pre configured to help newbies and briefly used tinycore 2026-02-07 07:27:03 MrMobius, what package manager does tiny core have ? 2026-02-07 07:28:02 MrMobius, in my opinion pkgsrc would be the ultimatebuild system for Linux 2026-02-07 07:28:55 I think I'll try and use it with a decent non systemd Linux soon on my spare pc 2026-02-07 07:30:45 https://www.pkgsrc.org/#quickstart 2026-02-07 07:31:33 Code exists in pkgsrc to support 20+ Unix-like operating systems and 15+ CPU architectures. 2026-02-07 09:33:46 iv4nshm4k0v: What sort of specs does your NetBSD system have? 2026-02-07 09:47:53 veltas, as it uses pkgsrc, it would have more packages available than any other system 2026-02-07 09:49:04 https://pkgsrc.se/ categories are on the left 2026-02-07 09:50:28 205 editors for instance 2026-02-07 09:50:54 413 programming languages 2026-02-07 09:54:53 Sorry I mean like the hardware like RAM, CPU etc 2026-02-07 09:55:50 I reimplemented a package manager in Lua that was maintained by the computing society at my university 2026-02-07 09:56:20 I thought you might, but used your question to show how many apps pkgsrc has :) 2026-02-07 09:56:25 Was very nice and simple and I added dependency cycle detection 2026-02-07 09:56:38 cool, I like Lua 2026-02-07 09:58:13 Ive just finished a pop up CMSIS-SVD search window for neovim in Lua, uses a Sqlite db for the data. It's so easy to use 2026-02-07 09:58:37 https://github.com/veltas/dcs-lua 2026-02-07 09:59:02 Wow 10 years ago, I definitely approach stuff differently now 2026-02-07 09:59:09 I probably would have written in C today 2026-02-07 10:01:23 neovim is written in Lua so my pop up in Lua is a good fit 2026-02-07 10:02:14 I'm thinking about efficient lookup of a table of key-value pairs, where I have to search the key-value pairs many more times than I insert them. So I think I'll keep the keys in a sorted array and do binary search, seems like the simplest overall method 2026-02-07 10:02:40 I wrote a general purpose memory burner in C for the PC back around 1995, along with the hardware design and released it all under the GPL 2026-02-07 10:03:13 I was considering a hash table but I am writing this to run on an 8-bit platform so the binary search may be faster and it's definitely less code 2026-02-07 10:04:15 veltas, I used Sqlite db for speed as the CMSIS-SVD file can easily be 1MB of XML 2026-02-07 10:04:46 8-bit platform though 2026-02-07 10:04:51 and a DB means less programming which suits me as Im not a real programmer 2026-02-07 10:04:56 ahh 2026-02-07 10:05:03 I missed that 2026-02-07 10:05:27 All my FURS apps run on a fast PC as well 2026-02-07 10:05:44 On a fast PC I'd use a linked list 2026-02-07 10:06:26 Less effort than linking and initialising / talking to sqlite 2026-02-07 10:07:22 The last time I touched XML at work I wrote the parser manually with scanf, it turned out to be less work than using libexpat or libxml 2026-02-07 10:08:15 I have become conversant with XSLT 2026-02-07 10:08:20 It's amazing how often it's easier to write some hacky code to do something than it is to integrate a general-purpose 'industry standard' library 2026-02-07 10:08:33 I do all my XML transforms in it 2026-02-07 10:08:41 Yeah I know you are an XSLT person, maybe I should have used that 2026-02-07 10:09:02 But then we're a C shop so I'd be expecting whoever maintains it to learn XSLT, so the short C program is probably a better fit for us 2026-02-07 10:09:13 I always use the FLOSS app designed to do each job, if it exists 2026-02-07 10:09:34 I've tried that and I just find quite often that results in more work, and a larger solution 2026-02-07 10:10:08 I certainly don't recommend reinventing the wheel though, if you don't like expat and write your own general-purpose XML parser then you're a fool 2026-02-07 10:10:11 being a electronics guy I'm ambivalent about that, I just spend the time needed 2026-02-07 10:11:11 Im not qualified to go off designing my own apps or id end up knee deep in slop 2026-02-07 10:11:33 I designed the pop-up because there isnt one already 2026-02-07 10:12:04 That's the difference though, I didn't write a general XML parser, I wrote a parser for the XML file I was expecting. I added a tiny bit of flexibility in e.g. whitespace but would reject any XML macro stuff I didn't recognise 2026-02-07 10:12:14 And it was just a lot less code and a lot more readable than using expat 2026-02-07 10:12:22 and I designed a Forth LSP and it works well, but it used pygls a general purpose LSP, ot only needs config 2026-02-07 10:12:47 This is part of the Forth philosophy of solving the problem you have, not some imagined problem 2026-02-07 10:13:17 I'd never attempt to create a package manager myself, theyre way beyond my skill level 2026-02-07 10:13:33 And beyond your interest I'm assuming 2026-02-07 10:14:06 Ive been doing Forth since 2014 and Ive definitely developed a 'Thinking Forth' mindset 2026-02-07 10:14:16 yes you're correct 2026-02-07 10:14:42 Package managers can be extremely simple, I mean the slackware package manager is essentially `tar xf package.tar.gz`, Arch Linux's package manager is almost that simple but will check for overwritten files and signatures 2026-02-07 10:15:00 But if you're not interested then it doesn't really matter how simple it is, it's not the problem at hand 2026-02-07 10:15:00 plus I first tried pkgsrc around 1999 and it's simply awesome 2026-02-07 10:15:40 yeah, the Slackware package manager really isnt one (I used Slackware for about 5 years) 2026-02-07 10:16:33 I heard recently that Germany is paying for Arch Linux's package manager to be rewritten in Rust for like half a million dollars, and I chuckled, that's *way* too much money for that job 2026-02-07 10:16:46 pkgsrc will auto build from src or binary, even switching between them as required depending whats agailable 2026-02-07 10:17:15 But good luck to whoever gets that contract lol, easy money 2026-02-07 10:17:29 and pkg src will go thru long lists of urls until it locates the right version/checksum etc 2026-02-07 10:17:46 I get the feeling that Rust is here to stay 2026-02-07 10:18:25 Yeah but "here to stay" amongst C, C++, Java, node.js, Python, Go, et al 2026-02-07 10:19:02 It's simply not general-purpose enough to 'replace' any of these, and if C++ couldn't replace C then nothing can, that was the best shot 2026-02-07 10:19:17 I have a son who programs as a hobby and he's very negative about Rust, but before it became popular I met Rust embedded guys online and I was impressed at how friendly and capable they were. They never once hassled me about Forth either 2026-02-07 10:19:48 A language with higher-level features than C, that's mostly backwards compatible with C so you can just convert your codebase to C++ by swapping gcc for g++ and fixing the manageable list of errors 2026-02-07 10:19:58 so I have a posiive view of the Rust prople 2026-02-07 10:20:08 It's not about the people, it's about the language 2026-02-07 10:20:26 only to a programmer ;-) 2026-02-07 10:20:46 A programmer being anyone who writes in a language 2026-02-07 10:21:02 You would hit these problems if you tried using Rust, anyone would 2026-02-07 10:21:07 The borrow checker is fundamentally too harsh and too basic 2026-02-07 10:22:03 to me it's about the people, for instance, Ive been building CMSIS-SVD tooling since 2016 and the only people who started making CMSIS-SVD tooling for non ARM embedded archs were the early Rust people 2026-02-07 10:22:50 People matter but I'm saying in terms of the big picture, I think the gloss is going to wear very quickly on Rust 2026-02-07 10:22:58 I barely hang on my the skin of my nails programming, remember Im not a real programmer, Im real electronics tech tho 2026-02-07 10:23:05 Because people will find out the language is deeply flawed 2026-02-07 10:23:19 ahh 2026-02-07 10:23:22 You'll see a lot of stuff being written in Rust right now, but it won't be the hyped language forever 2026-02-07 10:23:37 I guess nothing ever is ? 2026-02-07 10:23:48 But that Rust stuff will probably stick around, a lot of it will die though because it's a moving target and not everyone can maintain things like a full time job 2026-02-07 10:23:59 One of my favorite editors is Helix, and thats written in Rust 2026-02-07 10:24:15 Helix hopefully is active enough to continue being maintained as Rust changes 2026-02-07 10:24:48 and it was so very easy to configure Helix for my Forth LSP, as easy as eating icecream and apple pie 2026-02-07 10:25:13 Well that's not a feature of Rust, that's a feature of the developers who wrote Helix 2026-02-07 10:25:45 Helix is on another level I think. While neovim works and is easy to drive, it's a bit of a patchwork quilt, disorganised, hard to debug 2026-02-07 10:25:56 neovim isn't a rewrite of vim though 2026-02-07 10:26:07 no, nor is helix 2026-02-07 10:26:21 Helis has a lot of differenced to Vim 2026-02-07 10:26:33 I mean a 'reinterpretation' then? 2026-02-07 10:26:53 neovim is literally a fork of vim with Lua scripting and nicer defaults, if I remember right 2026-02-07 10:27:03 So they're a bit hamstrung by upstream 2026-02-07 10:27:09 but it's super fast. Neovim can choke on huge files (being Lua) but helix being Rust is like lightening 2026-02-07 10:27:31 Does neovim use luajit? 2026-02-07 10:27:37 yeah the keys for Neovim are much the same as Vim 2026-02-07 10:27:50 I'm not sure 2026-02-07 10:28:10 Might be coded inefficiently then, or something luajit can't optimise well for 2026-02-07 10:28:23 I wouldn't assume 'Rust' is the reason for performance there, that's making a lot of assumptions 2026-02-07 10:28:41 It's quite possible the Helix team created a much more efficient way to do this, credit to them 2026-02-07 10:28:45 tp@fbsd15:~% fabric "Does neovim use luajit?" 2026-02-07 10:28:45 Short answer 2026-02-07 10:28:46 - **On most platforms Neovim is built against LuaJIT 2.1** (the OpenResty fork). 2026-02-07 10:29:38 1MB is about the right size to expose inefficient algorithms on a modern PC 2026-02-07 10:29:41 my ignorance level is stellar, so assumptions-r-us 2026-02-07 10:30:06 When you get up to 'millions' of things you tend to get hit hard by O(n^2) stuff etc 2026-02-07 10:30:40 But 1MB is 'tiny' in a sense to your computer, if the algorithms and data structures are chosen correctly 2026-02-07 10:30:45 my choice od sqlite and Lua seems ok for Neovim 2026-02-07 10:31:08 Im not aware of waiting for search hits 2026-02-07 10:32:10 when called the popup opens listing all the tables, one for each peripheral in the STM32XX 2026-02-07 10:32:43 It's sqlite's job to choose decent algorithms and data structures for your queries, it's a good tool for something like that certainly 2026-02-07 10:32:51 one selects a table and hits enter, then all the registers_fields for that table are diplayed over the tables 2026-02-07 10:33:55 search or arrow key to what you want, hit enter and the peripheral_register_field is pasted into the embedded source 2026-02-07 10:34:50 it doesnt get any easier than that in my experience, and the list of tables (peripherals) allows finding what you want quickly 2026-02-07 10:35:22 so Its a bit like a man pags for the STM32 Technical Manual 2026-02-07 11:27:58 FWIW, my preferred vi-like editor is the Debian http://packages.debian.org/trixie/vim-tiny build. I can use NetBSD vi as well, though my Vim habits get in the way here. NetBSD vi also seems a tad buggy - e. g., I've had some glitches with multi-level undo in a buffer containing UTF-8. 2026-02-07 11:27:58 veltas: I have several (currently largely unused) i586 boxes (Intel Pentium MMX, AMD K6 - that kind of thing) and I've became interested in NetBSD specifically because it was said to run on that kind of hardware. Hadn't found the time to actually try it myself. 2026-02-07 11:33:24 (I also run NetBSD/amd64 on AMD A8, but I've set up pkgsrc for building packages with -m32 -mcpu=i486 as that's what I've been interested in in the first place.) 2026-02-07 11:41:53 iv4nshm4k0v, perhaps lua has replaced Perl because the current netbsd devs dont know Perl well ? 2026-02-07 11:42:31 iv4nshm4k0v, FreeBSD say thay have no one who knows Forth, so theyre changing the /boot stuff to Lua 2026-02-07 11:43:04 iv4nshm4k0v, and the Forth /boot files are pretty old, like 2012 - 2015 2026-02-07 11:43:26 the Lua /boot stuff is around 2018 2026-02-07 11:44:17 I once uses Perl for everything on PC, a lot of automation software and it has everything 2026-02-07 12:40:34 Lua is first mentioned in NetBSD 6.0 changelog record. Perl isn't mentioned at all, so I suppose it was never part of NetBSD. Lua /might/ be lighter-weight and (or) more portable than Perl, I think. 2026-02-07 13:05:25 I think modern forth are severely underrated for embedded use, size, simplicity and speed!  But those things do not care any more... 2026-02-07 13:30:30 A classic (threaded) Forth could certainly be simple, but it comes with a performance penalty due to the regular calls to the 'next' routine. Conversely, there /are/ performant (proprietary, I think?) Forths that avoid that, but they are no longer 'simple.' 2026-02-07 13:49:05 The fact that Lua is a first-class language on NetBSD is interesting and probably fun 2026-02-07 13:50:33 My experience is NetBSD doesn't work well on old computers because the drivers are a bit stale and get updates that aren't tested too well 2026-02-07 13:50:56 And the support doesn't exist, you need to self-support or fix the drivers yourself 2026-02-07 13:51:46 Might be different now but I didn't get a ton of help, I had mentioned I was a firmware engineer by trade and was willing to try and fix my drivers but didn't get far and nobody was really willing to help 2026-02-07 13:52:00 If I had student-tier time I'd probably figure it out 2026-02-07 13:52:26 But as a busy father of young children with a full time job I don't 2026-02-07 13:53:12 Maybe LLMs will help get further, don't know how you feel about them 2026-02-07 13:55:48 I'm not sure I understand how drivers being "a bit stale" could be a problem for /old/ computers. Do you suggest that, say, Linux has newer drivers for SoundBlaster 16 ISA cards? Other than that, it's a volunteer effort. It's possible that the person who did the code you've had issues with simply wasn't there at the time. NetBSD developers tend to have full-time jobs, too. 2026-02-07 13:56:19 I suspect the issue is they change the kernel and the driver gets updated to fit the new API and isn't tested 2026-02-07 13:56:41 If I already knew NetBSD kernel dev then it wouldn't be such an issue, but I don't 2026-02-07 13:57:22 Testing drivers for all your devices you support is notoriously difficult even for big projects, NetBSD is one of the smallest OS projects 2026-02-07 13:57:58 I'll probably get BSD fever and try NetBSD again some time, hopefully this time I get further 2026-02-07 13:59:48 Linux has the same issue, except their policy tends towards removing interfaces needed for performance on older hardware, and again introducing bugs that never get tested 2026-02-07 14:00:53 I can't remember what it's called but I've got a laptop with a Rage II GPU and the driver is extremely glitchy now, and I've been told it's something to do with the kernel removing support for the better tested interface that helped accelerate graphics 2026-02-07 14:02:55 "user-space mode setting" maybe? 2026-02-07 14:07:41 https://en.wikipedia.org/wiki/Direct_Rendering_Manager#History 2026-02-07 14:27:20 With the majority of GPUs requiring proprietary microcode software to function, I've decided I'd try to make sure I do not use accelerated graphics at all. So, yes, I've heard of issues - not just with kernels, but with X.org changing driver APIs seemingly every few months - but so far as I can tell, it's been very rarely that I've been directly affected. 2026-02-07 14:27:20 For years, I've been using a Socket AM2 mainboard that had an in-chipset GPU without such a flaw, but it sadly died last year. I think that my DDR2-based Intel Atom boards (one of them Intel D510MO) are also "free software-friendly" (i. e., would support accelerated graphics with an FSF-endorsed distribution), but all else I have requires proprietary OS parts for accelerated graphics. 2026-02-07 14:34:28 I reallly don't like hardware vendors that won't tell us how their hardware works at an INTERFACE level. We don't need all the implementation secrets, but it would really be right and proper for us to know how to INTERACT with the hardware. 2026-02-07 14:35:54 The case that offends me the most is FPGA vendors refusing to disclose the details of their config bit streams. I mean, just by looking at what our possible options are gives us a pretty good idea of what information must be present there, so it's not like some holy secret is being hidden. 2026-02-07 14:36:13 But not knowing HOW the information is laid out makes it impossible to innovate tool chains. 2026-02-07 14:37:26 And I get it they want to sell their software, but using secrecy of the bitstream to prop that up is very close to what Microsoft already got in trouble for legally - they lost a case that involved them having undocumented Windows features that gave MS Office a leg up over competitors. 2026-02-07 14:37:50 It's considered illegal competition tactics. 2026-02-07 14:42:46 I think the minute they chose to try to monetize that software, there should have been a requirement that they publish everything needed for compeitition. 2026-02-07 14:42:56 Real, robust competition. 2026-02-07 14:43:27 And I do NOT believe that it would be impossible to do that without "giving away hardware design secrets." Like I said, the nature of the required information is already pretty clear. 2026-02-07 15:11:24 IMO that's over-regulating 2026-02-07 16:33:33 I have no familiarity with FPGAs, but they're certainly something I might be interested in. Yet those FPGAs that /require/ proprietary development tools are a no-go for me, so I certainly hope at least /some/ vendors are sensible. If not, well, I guess it won't hurt me to put my signature under a petition for /some/ regulation in the field. 2026-02-07 16:33:33 As to GPUs, my approach has been about as follows: a. use fbdev X video driver; b. make sure the X server has no access to /dev/dri if it exists; c. make sure no proprietary microcode is installed in the system; d. use vga= / vesa= boot(loader) option if the in-kernel driver doesn't support the video adapter without said microcode. 2026-02-07 17:20:51 I don't know that any of the vendors are sensible - I think to the extent third party tools exist they're reverse engineering efforts. 2026-02-07 17:21:19 The really major chip makers keep their bit stream format pretty secret, or at least did up through the time I was paying real attention. 2026-02-07 17:30:20 Maybe that's a business niche, an open FPGA 2026-02-07 17:34:37 iv4nshm4k0v are you trying to use only free software, are you part of the free software movement? 2026-02-07 17:43:00 From where I stand, anyone who uses free software - and that certainly includes myself - is "part of the movement." Other than that, I try to only /rely/ on software I can edit - which is a distinct category, but in practice, it mostly contains free software for me. 2026-02-07 17:43:01 I see no problem with using proprietary software on occasion, per se, but do it often enough, and I'd say it becomes a case of software using you. 2026-02-07 17:44:04 Sorry I should say I'm not about to character assassinate you or anything I'm just interested 2026-02-07 17:44:39 I'm not really interested in the politics/drama, just I do like GNU software and I just have a natural affinity with open source and free software 2026-02-07 17:44:52 I wouldn't pick GPL for most code but not really for a strong reason 2026-02-07 17:45:32 I don't consider it a 'freedom' per-se but certainly it's a 'freedom' that *could* be granted by a state, I don't really see it happening soon but maybe one day 2026-02-07 17:46:14 I've just noticed you seem to be taking a lot of concessions to avoid proprietary software so was wondering why, but I do understand the inclination 2026-02-07 17:46:32 I feel the same way I just have mostly given up on that goal, I don't really see a practical way to do so 2026-02-07 17:46:58 I think it's quite typical of a Forther though to prefer setups you can hack, and understand, if you're forced to. We all get the advantages of that 2026-02-07 17:51:23 And it seems also like you prefer simpler systems rather than over-complicated systems 2026-02-07 17:51:58 Well, to paraphrase, "every system must be as simple as possible, but not simpler." 2026-02-07 17:53:55 The problem with complicated systems (over- or not) is that the more complex they are, the more are the chances I won't be able to edit them - which is something I do care about. If Bash stops working the way I like it, I can patch it, possibly start fork. Can I do the same with LibreOffice? I'm frankly not as confident in my skills there. 2026-02-07 17:55:57 And like I've said, I don't /really/ mind proprietary software; what I /do/ mind is when stopping to use a particular tool ruins your life. I've been playing xbattle - one from http://ftp.gwdg.de/pub/x11/x.org/R5contrib/ - recently for a bit (mainly just seeing that I can make it run correctly on my NetBSD setup.) I can stop doing it at any moment with no trouble. 2026-02-07 17:55:58 Now consider that I would've been using Wi-Fi - with an adapter that uses proprietary microcode. I one day decide to stop using that microcode - and I've lost my access to LAN. In my opinion, it's better to just never start doing that. Same with DRM: you stop using that one DRM implementation, and you've just lost access to dozens of your books. 2026-02-07 18:00:24 Of course, if I'm using a non-free tool on behalf of my employer - it'd be their problem, not mine, if anything stops working. So I'll offer my share of constructive criticism, but if they're setting themselves for a fall, I'm not going to deprive them of the freedom to do that. 2026-02-07 18:00:25 (As an aside, the way I see it, 'freedoms' cannot be given. The only thing state can do is to recognize and /protect/ them.) 2026-02-07 19:45:06 Looks like NetBSD 5.0 in 2009 removed support for 80386-tier processors 2026-02-07 21:54:58 LOL iv4nshm4k0v NetBSD 3rd go may come sooner than expected, tried using my desktop today and systemd is totally messed up and probably needs recovery but it's annoyed me enough I want to replace Arch Linux 2026-02-07 22:02:46 veltas, there are also other possibilities, for instance PKGSRC with a non Systemd Linux distro ? 2026-02-07 22:04:22 veltas, sadly Linux From Scratch has now gone Systemd only due to Gnome and Kde Systemd dependencies 2026-02-07 23:50:10 Good news on the ESP32-C6: https://pastebin.com/MMhqJWJM 2026-02-07 23:50:23 Encryption off, secure boot off. I can take total control. 2026-02-07 23:52:13 As far as I can tell it comes with NOTHING on it other than its built-in bootloader. Normally it will boot from flash, but there's a button on it (tiny) you can push, or a signal you can hold down, that will cause it to slurp an image over the serial line and flash it. I downloaded a MicrorPython I can put on it - I'll probably do that and tinker with it a little before starting to look into my 2026-02-07 23:52:16 own things. 2026-02-07 23:52:49 I think I can actually develop without even flashing it - the software has a "load_ram" command that will download a binary image into ram and run it. 2026-02-07 23:56:59 Nice