--- Log opened Wed Apr 27 00:00:06 2022
03:44 -!- rndhouse [rndhouse@gateway/vpn/protonvpn/rndhouse] has joined #hplusroadmap
04:07 -!- darsie [~darsie@84-113-55-200.cable.dynamic.surfer.at] has joined #hplusroadmap
06:00 -!- yashgaroth [~ffffffff@2601:5c4:c780:6aa0::93] has joined #hplusroadmap
06:04 < kanzure> https://btctranscripts.com/sydney-bitcoin-meetup/2022-03-29-socratic-seminar/
06:04 < kanzure> FROST looks interesting (two round schnorr threshold signature scheme) https://datatracker.ietf.org/doc/draft-irtf-cfrg-frost/
06:09 < kanzure> https://lists.linuxfoundation.org/pipermail/bitcoin-dev/2022-April/020352.html
06:23 < kanzure> .tw https://twitter.com/AcademicsSay/status/1518715880322043907
06:23 < saxo> if you’re worried about billionaires owning publically sourced content on social media wait until you hear about academic publishing (@AcademicsSay)
06:25 < kanzure> where did fenn's robot flowers come from?
07:46 -!- rndhouse [rndhouse@gateway/vpn/protonvpn/rndhouse] has quit [Read error: Connection reset by peer]
07:46 -!- rndhouse [rndhouse@gateway/vpn/protonvpn/rndhouse] has joined #hplusroadmap
07:58 -!- rndhouse_ [rndhouse@gateway/vpn/protonvpn/rndhouse] has joined #hplusroadmap
08:00 -!- spaceangel [~spaceange@ip-78-102-216-202.net.upcbroadband.cz] has joined #hplusroadmap
08:00 -!- rndhouse [rndhouse@gateway/vpn/protonvpn/rndhouse] has quit [Ping timeout: 240 seconds]
08:05 -!- spaceangel [~spaceange@ip-78-102-216-202.net.upcbroadband.cz] has quit [Remote host closed the connection]
08:06 -!- spaceangel [~spaceange@ip-78-102-216-202.net.upcbroadband.cz] has joined #hplusroadmap
08:41 < docl> .tw https://twitter.com/Ben_Reinhardt/status/1519335831542054913
08:41 < saxo> Why aren't there more nanotechnology hackers? (@Ben_Reinhardt)
09:25 < maaku> Who is this Ben Reinhardt fellow?
09:34 < docl> well, he wrote this https://benjaminreinhardt.com/parpa
09:35 < docl> kind of a strategist
09:51 -!- rndhouse_ [rndhouse@gateway/vpn/protonvpn/rndhouse] has quit [Ping timeout: 272 seconds]
09:57 -!- ccdle12 [~ccdle12@243.222.90.149.rev.vodafone.pt] has joined #hplusroadmap
10:11 -!- rndhouse [rndhouse@gateway/vpn/protonvpn/rndhouse] has joined #hplusroadmap
10:26 < muurkha> my thought is that there aren't more nanotechnology hackers because (a) Foresight warned us allowing nanotechnology hackers could destroy the world and (b) hardware is expensive, thus requiring funding that isn't needed for cracking game copy protection and (c) most research funding is in the US, and the national nanotechnology initiative Clinton launched in the US was corrupted from the outset by 
10:26 < muurkha> random scientists rebranding their non-nanotech research areas as "nanotechnology"
10:26 < muurkha> what do you think?
10:33 < docl> Sounds about right... Although I think a few million from the obama era did go to APM projects during the trump era
10:35 < muurkha> yeah, there was definitely real nanotech stuff that happened as part of the national nanotechnology initiative too
10:35 < muurkha> I guess there's also (d) the US mostly no longer knows how to innovate
10:41 -!- Llamamoe [~Llamamoe@178235178127.dynamic-4-waw-k-1-2-0.vectranet.pl] has joined #hplusroadmap
11:04 < maaku> The DOE's advanced manufacturing office is the nanotechnology initiative we were promised but never delivered
11:04 < maaku> But unfortunately they didn't stagger team selection and I think the program has basically wound up?
11:05 < maaku> It does seem to have paid for the aforementioned Zyvex research and some APM tip work that was bought up by CBN Nano
11:08 < docl> yeah, they also funded shafmeister's group
11:10 -!- Molly_Lucy [~Molly_Luc@user/Molly-Lucy/x-8688804] has joined #hplusroadmap
11:11 < maaku> muurkha: regarding lack of a nanohacking community, you give good reasons but I also think a lot of it boils down to not knowing where to start
11:12 < maaku> you can't go build a lab in your garage (too capital intensive), and what else is there to do?
11:12 < maaku> hence my comment about needing a good CAD+simulator
11:12 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has joined #hplusroadmap
11:12 < maaku> if you could at least design something and watch it work, there'd be an onramp to getting people involved
11:20 < muurkha> maybe, yeah
11:21 < muurkha> but the motivation structure would largely be that of a video game
11:25 -!- Molly_Lucy [~Molly_Luc@user/Molly-Lucy/x-8688804] has quit [Read error: Connection reset by peer]
11:27 -!- aedla [~aedla@0f71-0000-0000-0000-5f00-87d8-07d0-2001.dyn.estpak.ee] has joined #hplusroadmap
11:29 < docl> make nanotech hacking adhd-compatible enough and you've mostly solved the problem
11:29 < muurkha> maybe, yeah
11:29 < muurkha> but what I mean is that if what motivates people to participate is video-game motivations
11:30 < docl> yeah that's a good point
11:30 < muurkha> then there's a tendency for participants to drift towards variants with better "gameplay"
11:31 < muurkha> which might be things that are totally non-nanotech-useful like Roblox
11:31 < docl> I've tried setting up schafmeister's cando stack, but never managed to get it to the point where it felt like making a blender model, let alone a video game :/
11:32 < muurkha> or it might be nanocad versions whose physical realism is more questionable, similar to what happened with Bitcoin where things like Ripple and central banks tried to rebrand as "crypto" and "blockchains"
11:33 < muurkha> and now everybody (here in Argentina) is using Binance and Tether
11:35 < muurkha> neither Binance nor Tether is secure against the kinds of probems Bitcoin was designed to withstand, but they're still somewhat constrained by the physical reality of needing to interoperate with actual Bitcoin and the rest of the economy
11:38 < muurkha> so developments that potentially represent genuine advances, like Monero, ZCash, Grin, and even Ethereum, have great difficulty competing for mindshare against outright frauds like Tether, Infura, and OpenSea
11:52 < kanzure> https://freenode.irclog.whitequark.org/bitcoin-wizards/2016-08-14#17248381;
11:52 -!- Molly_Lucy [~Molly_Luc@user/Molly-Lucy/x-8688804] has joined #hplusroadmap
11:53 < kanzure> "Those of you who have not done biology can't realize how primitive things still are. A CS analogy is that we're trying to reverse engineer a non-deterministic alien architecture from the future via a remote debugger over a noisy line that has no error correction and runs at a fraction of a millibaud, where every peek or poke costs hundreds to thousands of dollars."
11:53 < kanzure> muurkha: i think tether demonstrates the insatiable lust for dollars around the world. it's fascinating.
11:55 < maaku> muurkha: I don't see the video-game connection
11:56 < muurkha> maaku: "design something and watch it work" is the motivation structure of Roblox, Minecraft, and the demoscene
11:57 < kanzure> anyone know a good rust library for DHTs
11:58 < maaku> we're talking about a CAD environment, not a game. not so much 'watch it work' as 'validate it works as intended' the simulation need not even be visual except in as much as that improves the UX
11:58 < kanzure> not sure if you remember but fenn had related idea of game for machining and machine construction based on real-world physics.. or maybe that was maaku.
11:59 < muurkha> maaku: right, validate that it will probably work as intended if the simplifying assumptions underlying the simulation turn out to be correct
12:00 < kanzure> muurkha: have you pondered the problem of (sadly permanent) local minimas in "open-ended" artificial life evolution simulators?
12:01 < kanzure> there seems to be some undocumented assumption that the world has that allows for large degrees of novel forms and complexity to emerge that is hard to capture or not yet captured for a-life evolution simulators
12:05 < muurkha> kanzure: not boring global minima but permanent local minima?  presumably you can eliminate permanent local minima with a fractal archipelago architecture
12:06 < maaku> muurkha: we're at the point now where ab-initio quantum simulation of parts consisting of a few thousand atoms is doable, even for hobbyists
12:06 < muurkha> nice
12:07 < muurkha> I thought we couldn't yet do ab-initio quantum simulation of helium's emission spectrum, just hydrogen
12:07 < kanzure> by permanent local minima i mean these simulations reach homeostasis at some point and stop doing interesting things
12:07 < muurkha> kanzure: how do you know that isn't the global minimum?
12:08 < kanzure> i think the real world is evidence that the substrate of reality seems to support substantially more expressive forms than whatever my dumb simulators stop at
12:09 < maaku> muurkha: you're conflating exact solutions with numerical methods
12:09 < maaku> there's only exact solutions for hydrogen, but we can use numerical codes to simulate larger quantum systems of course
12:10 < kanzure> what sort of errors happen in nonexact ab initio simulators?
12:10 < muurkha> ah, I see
12:10 < maaku> https://en.wikipedia.org/wiki/Ab_initio_quantum_chemistry_methods
12:10 < maaku> https://en.wikipedia.org/wiki/List_of_quantum_chemistry_and_solid-state_physics_software
12:11 < lsneff> Lots of interesting ML stuff happening in the qm simulation field
12:12 < kanzure> maaku: can i trouble you to talk about secure elements for a bit?
12:12 < maaku> kanzure: sure but probably not today. you'd probably want to talk to christopher allen too
12:13 < kanzure> already am
12:24 < docl> see, this is why spiroligomers are cool. more rigid, less messy physics
13:01  * L29Ah fills docl's mouth with quantum foam
13:03 < muurkha> heh
13:09 < docl> .t https://www.youtube.com/watch?v=8X69_42Mj-g
13:09 < saxo> Clasp: Common Lisp using LLVM and C++ for Molecular Metaprogramming - YouTube
13:11 < fenn> the machine construction game was me, and i'm still working on it
13:11 < muurkha> cool!
13:11 < fenn> i ran into the problem of "i can't compile hello world on windows no matter what i try" and gave up after a week of that
13:13 < muurkha> fenn: you may be interested in http://canonical.org/~kragen/sw/w32dry/
13:13 < fenn> so then i've been trying to use panda (shitty graphics library) instead of OGRE (less shitty graphics library) because panda is in python and doesn't need anything to be compiled
13:13 < fenn> maybe eventually i'll use armory3d when i get a good computer, but i still want a version of the game client to be able to run on low end hardware so i will probably have two versions in the end
13:14 < muurkha> I think things built with w32dry still run on current Windows
13:14 < muurkha> however it doesn't quite reach the level of "shitty graphics library" I think
13:15 < fenn> o cam
13:15 < fenn> i can't be doing this right now, sorry
13:15 < muurkha> a nearly non-shitty alternative might be Godot
13:15 < fenn> i'd love to sit down and try to compile your code etc, but it would result in the day being used up
13:15 < fenn> yeah godot has some things going for it, but it's really trying hard to be something i don't actually want
13:16 < muurkha> yeah
13:16 < muurkha> but it solves the hello-world-on-windows problem.  also the screen-tearing-on-Linux problem which is actually important for 3-D animations
13:17 < fenn> no i mean literally i can't compile void main {printf("hello world!")}
13:17 < muurkha> I know
13:17 < fenn> gcc pukes, visual studio needs me to sign up with MS, etc
13:17 < muurkha> with Godot you can, as long as you don't insist on writing it in C
13:17 < fenn> oh, right
13:17 < muurkha> or outputting to the console
13:17 < fenn> but then i'm perpetually stuck in godot-land
13:17 -!- ccdle12 [~ccdle12@243.222.90.149.rev.vodafone.pt] has quit [Quit: Client closed]
13:17 < muurkha> which is a bad idea on Windows anyway
13:17 < muurkha> yes
13:18 < muurkha> Godot also solves the how-the-fuck-do-I-load-a-shader problem
13:18 < muurkha> which is relevant even on low-end hardware these days. maybe more relevant
13:18 < fenn> i want it to be cross platform, and i figure starting early is better than trying to backport code later, but this means i have reached a point of zero code progress
13:18  * fenn mopes
13:19 < muurkha> yes, plausibly
13:19 < fenn> godot's examples don't work on ym low end hardware
13:19 < muurkha> can you just use HTML5?
13:19 < fenn> no, of course not, don't be silly
13:19 < muurkha> why not?
13:19 < fenn> because webgl doesn't work on my low end hardware
13:20 < fenn> also, jesus fucking christ are you stupid
13:20 < fenn> do i really have to explain what shitshow HTML5 is
13:20 < fenn> is "HTML5" even "a thing" and not just a passing transient state of confluence among vying corporate interests
13:21 < fenn> no, i don't want to tether my life purpose to that
13:22 < fenn> in case it wasn't made exceedingly clear, "web apps" make me angry
13:22 < fenn> please don't suggest that i build a web app
13:22 < fenn> also, language sucks
13:23 < fenn> i have no problems with using dynamic HTML to present database queries or whatever
13:24 -!- Llamamoe [~Llamamoe@178235178127.dynamic-4-waw-k-1-2-0.vectranet.pl] has quit [Quit: Leaving.]
13:25 < muurkha> obviously I'm stupid; my nickname is literally "muurkha"
13:26 < fenn> consider yourself to have succeeded in poking the bear
13:26 < muurkha> haha
13:26 < muurkha> don't consider yourself a bear, you might meet a real one
13:26 < muurkha> I'm surprised there is still hardware without webgl support
13:27 < fenn> as far as i can tell, my hardware will never be supported
13:27 < muurkha> I mean, that can run a browser
13:27 < muurkha> I agree that HTML5's stability leaves a lot to be desired
13:28 < fenn> my point was that it's not a standard made by a standards body, and it's not attempting to be
13:28 < muurkha> yeah, that's reasonable
13:28 < muurkha> and I think it's a good idea to not tether your life purpose to a passing transient state of confluence among vying corporate interests
13:29 < docl> .wik List_of_game_engines
13:29 < saxo> " /  / Game engines are tools available for game designers to code and plan out a video game quickly and easily without building one from the ground up." - https://en.wikipedia.org/wiki/List_of_game_engines
13:29 < muurkha> but if you get your UI running in webassembly with webgl you can probably run it on anything else into the future
13:29 < fenn> by the time i get to v1.0 "HTML5" will no longer exist
13:29 < muurkha> unless your desired low-end hardware can't run Mesa?
13:29 < fenn> everything can run Mesa, that's the point
13:29 < fenn> but, well, it sucks to do that
13:30 < fenn> okay everybody you can stop trying to help now
13:30 < muurkha> because if you write your own 3-D renderer it will perform better than Mesa on your low-end hardware?
13:30 < fenn> i'm not working on this at the moment and should be doing other things
13:30 < muurkha> okay.  sorry to have touched a sore point
13:31 < muurkha> platform choice is a difficult problem and I agree with you that there isn't an answer to it right now that's even reasonably acceptable
13:31 < fenn> in order to be popular enough to attract a critical mass of players, it NEEDS to run on windows. i WANT it to run on linux, because i hate windows
13:32 < fenn> having two computers running all the time is a pain
13:32 < muurkha> and you want it to run on hardware that doesn't even support WebGL and never will
13:32 < muurkha> and it needs to do real-time 3-D rendering because you're designing machinery
13:32 < fenn> right
13:32 < muurkha> brb writing a 3-D renderer for this ATMega168
13:33 < muurkha> uh ATMega328
13:33 < fenn> ten years ago this would have been a perfectly sane and normal set of requirements
13:33 < fenn> i'm sorry your standards are screwed up
13:33 < fenn> also btw i have been working on this in various incarnations for over ten years, i just haven't made a lot of progress
13:34 < muurkha> no, no, I agree with your set of requirements, and I agree that it's screwed up that it's hard to meet them
13:34 < fenn> then please don't make sarcastic comments about ATMega328s
13:35 < muurkha> it was hard ten years ago too though, that's why I wrote w32dry
13:35 < muurkha> that wasn't sarcastic, I really am going to do that, I just haven't finished yet
13:35 < fenn> i'm pretty sure i can just compile some C that calls OGRE and it will work on every platform OGRE runs on
13:36 < fenn> but, stupid compiler problems, things that ought to just work, don't
13:36 < fenn> burnout
13:37 < fenn> it's hard to keep momentum working on your own, but if you try to rope other people in, they have a different idea of what it is that should be done, and it's not what you wanted to do in the first place, so then you spend more time fixing contributions than writing new code
13:38 < fenn> i guess i need to do more drugs, but i don't know how to get them
13:38 < muurkha> yeah, I dont know how to work with other people
13:38 < muurkha> a 3-D renderer for an ATMega328 I can do though
13:39 < fenn> the point of running on low end hardware isn't as some kind of demo-scene maximum efficiency computing stunt, it's to enable people who can't afford high end hardware to participate and contribute data from their economic perspective
13:39 < fenn> this game is about building machines from scratch
13:39 < fenn> if you could afford a $50k CNC mill, it's not for you
13:41 < fenn> often i'm exasperated by the fact that it's ME that has to do some data contribution to the knowledge commons, out of the 8 billion humans on planet earth. it seems so unreasonable that somebody else hasn't already done this
13:41 < fenn> what are all those people doing?
13:41 < fenn> do they even exist?
13:42 < fenn> but maybe they're stuck behind artificial barriers created by the disparity between san francisco macbook culture, and 20 year old PS2 keyboard plugged into a USB adapter running on a discarded walmart linux machine
13:42 < fenn> just, stuff that nobody ever considered because they were chasing the bleeding edge
13:43 < fenn> i run into these issues all the time
13:43 < fenn> cloudflare hates me, 403s on 50% of websites now because i dare to use an older browser
13:43 < fenn> it's absurd. and nobody even knows about it
13:44 < muurkha> when I was volunteering a couple of years ago in the slums near here the kids there had better cellphones than I do
13:44 < fenn> yes, i should just give up and get a newer computer and install a new OS with systemd and god knows what else shitware has infested open sores in the last decade
13:44 < lsneff> frankly, trying to build something that runs well on both 20 year old hardware and modern hardware sounds like a waste of time
13:44 < muurkha> they definitely support WebGL and they cost US$30
13:44 < lsneff> Computers are cheap these days
13:44 < fenn> general purpose computers are still expensive
13:45 < muurkha> a little, but a lot of androids are rootable
13:45 < muurkha> one of those kids published let's plays on youtube
13:45 < fenn> even an old desktop machine with a GPU card is in the hundreds of dollars range
13:45 < fenn> android is a waste of time
13:45 < muurkha> he used speech synthesis so his neighbors couldn't find out it was him
13:45 < jrayhawk> if it is not self-hosting, it is not a general-purpose computer
13:46 < muurkha> agreed, jrayhawk 
13:46 < muurkha> not sure if android is a waste of time but it's certainly hard to self-host
13:46 < fenn> in principle, you could hack the z80 in your microwave to run GCC, but in practice you won't
13:46 < lsneff> A laptop with an integrated gpu that can run vulkan shaders is a few hundred bucks
13:46 < muurkha> I have the same 403 problem.  archive.today/archive.is/archive.fo helps sometimes
13:46 < muurkha> GCC needs more than 64K of RAM to run, fenn
13:47 < fenn> well, is that more or less difficult than slashing through the bloatware jungle that is android
13:47 < muurkha> android hardware is certainly powerful enough to do self-hosted development
13:47 < fenn> i can figure out how to attach a RAM bank to a z80
13:47 < muurkha> I got Jupyter running on my cellphone under Termux
13:48 < muurkha> I was surprised to see a Fortran compiler running on my cellphone
13:48 < muurkha> but then the Termux people broke Termux on my phone, and I think Termux is being broken on new Android?
13:48 < fenn> i have debian running on my nook
13:48 < fenn> it's totally possible to make ONE instance of android work for you, with a lot of work
13:49 < fenn> but trying to replicate that automatically across many different kinds of android hardware is a fool's errand
13:49 < muurkha> yeah
13:49 < fenn> also, every time i touch android i end up wasting hours somehow
13:49 < muurkha> maybe we could do installfests and FLISOL kinds of things for that
13:49 < fenn> it's supremely unsatisfying
13:49 < muurkha> android devices have a couple of big self-hosting problems not related to the software
13:50 < docl> you can apparently get an 8gb laptop with a 512GB NVMe on newegg for $400
13:50 < lsneff> ^
13:50 < muurkha> docl: that's still out of the budget of the slum kids
13:50 < docl> yeah
13:50 < muurkha> but there's plenty of hardware that supports WebGL out of the box in the US$30 price range
13:50 < lsneff> There
13:50 < muurkha> one problem, though, is that they're all battery-powered, and the batteries all break in a year or two
13:51 < lsneff> There's a middle ground between appealing to every possible user and starting the prroject
13:51 < fenn> yeah it's real dumb that you can't tell the microcontroller to stop charging at 4.0V/cell
13:51 < muurkha> and then they're hard to replace, either because they're weird prismatic shapes designed for that specific device or because they're literally glued to the case
13:51 < muurkha> the other problem is that they're physically fragile
13:52 < fenn> lsneff your "middle ground" is way off in webassembly land, over the web app mountain range
13:52 < muurkha> yeah, webassembly land is a good place to be!
13:52 < lsneff> Sure, what's wrong with that
13:52 < muurkha> or so I hear.  I haven't gone there yet myself
13:52 < fenn> i'm trying to stay solidly in the shire of x86 and openGL
13:52 < lsneff> Why?
13:52 < muurkha> why?  x86 hardware is all expensive and power-hungry
13:53 < lsneff> And opengl is no longer supported by a lot of hardware
13:53 < fenn> they're things i think i understand, and there's lots of documentation available, and stable software that "ought" to "just work"
13:53 < muurkha> webassembly doesnt require webapps
13:53 < fenn> are you serious?
13:53 < lsneff> OpenGL was outdated a decade ago.
13:53 < lsneff> Macs don't support it anymore, I believe
13:54 < fenn> but.. why
13:54 < fenn> it's like "ascii text is no longer supported. you have to use unicode"
13:54 < lsneff> 'cause it's bad compared to modern gpu apis
13:54 < fenn> so what
13:54 < lsneff> ascii text is unicode
13:54 < muurkha> only Macs don't support it, and you can still run OpenGL on them on Vulkan
13:54 < lsneff> opengl is a much worse match for modern graphics hardware than modern apis like vulkan
13:54 < muurkha> also I think there's a shim that lets you run OpenGL on top of Metal
13:55 < muurkha> my plan with the Zorzpad is to support a good UI in submilliwatt territory so I don't need to worry about batteries or charger ports
13:55 < fenn> arrrrgh
13:55 < muurkha> (Metal is the modern MacOS replacement for OpenGL)
13:55 < muurkha> Android 3-D is still OpenGL ES IIRC?
13:55 < fenn> lsneff: you ever have this problem that you find a good pair of shoes, and they fit well, but then the next time you go to buy a new pair because yours are worn out, you find out they stopped making those shoes, and now there are these almost identical ones that you have to run the gamble of them not fitting the same
13:55 < muurkha> and that's the vast majority of PCs
13:56 < fenn> they could have just kept making the old shoes, but they had to fuck with it
13:56 < fenn> but EVERYTHING is like this
13:56 < fenn> it's a never ending assault of stupid compatibility problems
13:56 < muurkha> agreed
13:57 < fenn> you can't even get concrete and steel that are the same as they used to be
13:57 < muurkha> although if you dislike compatibility problems you'll HATE the Zorzpad
13:57 < muurkha> because even running C on it will be a struggle
13:57 < muurkha> with major performance compromises
13:57 < fenn> opengl ES is something different
13:58 < muurkha> a little yeah
14:00 < fenn> almost every old laptop uses the same 18650 li-ion cells, so the batteries are repairable at least
14:00 < fenn> not sure what $30 hardware you're talking about though. probably some ARM tablet
14:01 < muurkha> yes, ARM tablets and cellphones that ship with Android and Chrome
14:01 < fenn> in "rainbows end" there's this guy still using an old thinkpad in 2040, because of the war on general purpose computing
14:01 < muurkha> yeah, old laptops have much more repairable batteries
14:01 < muurkha> yeah
14:02 < muurkha> hopefully instead I'll be using a Zorzpad
14:02 < fenn> oh it's actually set in 2025, but it feels like at least a decade in the future still
14:02 < muurkha> I don't think most old Thinkpads will last that long
14:02 < muurkha> oh, they'll last that long
14:02 < fenn> please explain what "Zorzpad" is
14:04 < lsneff> fenn: yeah, I really hate when that happens
14:04 < lsneff> In this case, I’m actually glad that opengl is on its way out, but I understand your frustration
14:11 < muurkha> fenn: the Zorzpad is the autonomous personal computer I'm building for myself.  SLC Flash, an Ambiq Apollo3 CPU, integrated keyboard, and memory-in-pixel LCD displays
14:12 -!- spaceangel [~spaceange@ip-78-102-216-202.net.upcbroadband.cz] has quit [Remote host closed the connection]
14:12 < fenn> have you looked at e-ink displays
14:12 < fenn> transflective LCD can be low power but they're usually low resolution
14:13 < muurkha> yes.  the memory LCDs use 1000 times less energy and have a 100 times higher frame rate than e-ink
14:13 < fenn> including backlight?
14:13 < muurkha> there is no backlight.  these are reflective because in the applications for which memory LCDs are most appealing a backlight is way outside the power budget
14:14 < fenn> ok i have never heard of this technology before
14:14 < muurkha> they have amazing contrast though.  the black is like looking into a hole
14:14 < muurkha> though, a hole with glass over it
14:14 < muurkha> the "white" is of course gray, about 50% albedo
14:15 < muurkha> the particular display I'm using is the LS027B7DH01, which is 400×240 at 175 dpi
14:16 < fenn> looks like the same product niches as e-ink
14:16 < muurkha> yeah
14:16 < muurkha> it's also used in the Playdate; other memory LCDs have been used in the Pebble smartwatch and some Mifree smartwatches
14:16 < muurkha> maintaining the static display is not free like with e-ink; it fades in about 30 seconds when I unplug it
14:17 < muurkha> the datasheet says it's 50 μW but I haven't masured that yet
14:17 < muurkha> *measured
14:18 < maaku> fenn: congrats on getting old
14:18 < muurkha> so my plan is to pot the displays with some water-clear neutral-cure silicone a few millimeters under a thick layer of polycarbonate, with the CPU potted underneath, inside a hard case
14:18 < docl> I'm trying to get perry metzger to join :)
14:18 < maaku> docl: any relation to phil metzger (the planetary scientist)?
14:18 < fenn> TMI i guess
14:19 < maaku> in all seriousness it is true that OpenGL is deprecated on most platforms.
14:19 < fenn> i also get perry/phil metzger confused
14:19 < muurkha> docl: tell him Kragen Sitaker says hi
14:19 < muurkha> he might know me as Kragen Sittler
14:19 < docl> maaku: not that I'm aware of... last name spelled the same though I think
14:19 < fenn> i think there are actually two phil metzgers in crypto-land
14:19 < muurkha> really?
14:20 < muurkha> one perry and two phils?  that's a lot of metzgers
14:20 < fenn> i am foncuzed
14:20 < maaku> fenn: if you want a low-level cross-platform API to code against that most resembles what OpenGL was >10 years ago, Vulkan is the best option.
14:21 < fenn> "Dr. Philip Metzger is a planetary physicist with the Planetary Science faculty at the University of Central Florida"
14:21 < fenn> docl: which perry metzger are you referring to?
14:21 < maaku> But honestly I think you should just develop it in godot rather than wasting years of your life reinventing the wheel
14:21 < muurkha> I don't think programming in Vulkan resembles programming programming in OpenGL very much, especially if you're using the parts of OpenGL that aren't in OpenGL ES
14:21 < muurkha> but I'm the farthest thing from an expert, I'm a muurkha
14:21 < fenn> i don't really want opengl, i want some framework that handles scene graphs and loading shaders for me
14:21 < maaku> (I would actually say use Unity or Unreal, but if you're unwilling to sign up for Visual Studio than I assume those aren't options either)
14:22 < fenn> rihgt
14:22 < muurkha> yeah, I think Godot is a genuinely free alternative to unity and unreal for now
14:22 < fenn> one major goal of the game is to showcase open hardware and the whole open source ecosystem, so using unreal is hypocritical
14:22 < maaku> fenn: that would be godat. it handles all the scene graph and shaders and stuff
14:22 < muurkha> you could write the prototype in Godot and then rewrite it in something else once you have enough experience to know what matters
14:22 < fenn> yeah
14:23 < muurkha> OpenFrameworks is another cross-platform alternative that handles loading shaders for you.  not sure if you can get it to compile with mingw32 tho
14:23 < muurkha> it's a C++ library
14:23 < maaku> muurkha: there is no modern API that resembles OpenGL, because 3d hardware no longer resembles the SGI workstations OpenGL was developed on 30 years ago
14:23 < muurkha> 30?
14:23 < fenn> yep
14:23 < muurkha> I guess 35 is kind of like 30
14:24 < muurkha> those 5 years seemed really long at the time though
14:24 < maaku> muurkha: https://www.khronos.org/opengl/wiki/History_of_OpenGL#OpenGL_1.0_.281992.29
14:24 < muurkha> maaku: yeah but that's omitting IRIS GL
14:24 < fenn> "Silicon Graphics, Inc. (SGI) began developing OpenGL in 1991 and released it on June 30, 1992"
14:24 < maaku> fair enough
14:25 < fenn> did you actually have access to IRIS GL?
14:25 < maaku> but Vulkan is the replacement for "low level API to program the graphics hardware in a platform-independent way"
14:25 < maaku> which is what OpenGL used to be
14:25 < muurkha> yeah, I did, but I didn't take advantage of it
14:26 < muurkha> I only had access to IRIS GL after OpenGL came out though!
14:26 < muurkha> first on an Indigo
14:26 < fenn> my dad ran a computational chemistry lab at a university with lots of SGI hardware, which is where i learned computer stuff. at home we had a mac plus, the cute little box with the face on it that you still see in mac OS occasionally when things go horribly wrong
14:26 < muurkha> but the IRIS 1000 shipped 39 years ago and I think it already had "GL"
14:27 < maaku> fenn: the Mac doorstop
14:27 < docl> fenn: he's an old guard extropian/cryptography guy, you've probably seen him around
14:27 < fenn> there were several "toasters" which were sort of hardware accelerated cheap (for SGI anyway) thin clients, and then a couple "Octane" servers which were these big green boxes with with a sleek design
14:28 < muurkha> docl: yeah, that's the one I know
14:28 < muurkha> maaku: yes, that's true!
14:29 < fenn> all these things were networked together so you could run jobs on a server with X forwarding over the LAN
14:29 < maaku> fenn: also fyi "webassembly" is more like a replacement for JVM. it should never have had the "web" prefix
14:29 < fenn> meh
14:30 < maaku> it's a link-once, run-everywhere architecture that doesn't really have anything to do with browsers or html
14:32 < fenn> Indigo sounds familiar too
14:33 < fenn> everything ran IRIX but afaik all the software was using openGL by then
14:34 < lsneff> I have an interview with the starlink engineering team!
14:34 < fenn> run in circles, scream and shout!
14:35 < docl> congrats!
14:35 < lsneff> thank you 😁
14:36 < muurkha> lsneff: that's wonderful!  for what position?
14:37 < lsneff> I'm not sure yet—some yet to be announced project apparently
14:37 < fenn> i'm pretty concerned that starlink is openly being used as part of the ukraine-russia war, in military applications
14:38 < fenn> i feel this significantly raises the likelihood of ASAT weapons being used against the starlink constellation, which would result in some kessler syndrome scenario
14:38 < lsneff> Yeah, I do worry about that
14:38 < muurkha> maybe, so far nobody has tried using ASAT weapons against surveillance satellites and GNSS, which are used nonstop in military applications
14:39 < muurkha> and surveillance satellites are generally in LEO so they'd be easy to hit
14:39 < fenn> russia has a lot less to lose by ruining LEO
14:41 < muurkha> than Ukraine?
14:41 < fenn> than the US
14:42 < muurkha> yeah
14:42 < lsneff> Actually, starlink orbits are so low that the debris would deorbit in just a few years
14:42 < muurkha> but they aren't at war with the US
14:42 < fenn> at least from russia's point of view, Ukraine is a proxy of the US
14:42 < muurkha> lsneff: all of it?  I was wondering about that
14:42 < muurkha> if Russia gets desperate we'll have bigger concerns than LEO
14:42 < lsneff> That's my understanding, but I'm not 100% sure
14:42 < muurkha> or, well, slightly smaller in diameter
14:43 < lsneff> Debris would have more surface area per mass as well, so it would deorbit faster than the original sats as well
14:43 < fenn> lsneff when you blow up a satellite, bits and pieces can end up on much higher trajectories and impact other satellites in higher circular orbits, and then you have debris in a higher circular orbit. this happened on the recent india ASAT test for example
14:43 < lsneff> That's a good point
14:44 < muurkha> fenn: how do they circularize their orbit after getting higher?
14:44 < lsneff> Yeah, some would probably end up in higher orbits
14:44 < muurkha> I'm pretty weak on orbital dynamics
14:44 < fenn> ballistic coefficient of debris tends to be about the same as the original satellite, because it's relatively dense shards rather than a big hollow box
14:44 < fenn> muurkha: they don't circularize, they hit stuff that's in a circular orbit already, generating new debris in a circular orbit
14:44 < muurkha> oh, that makes more sense
14:45 < muurkha> but that means they only have a single orbit to hit stuff
14:45 < lsneff> Ah, hmm
14:45 < fenn> no, it's an elliptical orbit that gradually lowers apogee
14:46 < muurkha> oh, like if it's only slightly elliptical?
14:46 < fenn> the probability of any individual piece of debris intersecting a satellite at a high orbit is very low, but with thousands of satellites in LEO being blown up, that's a large number of chances
14:46 < muurkha> because if it was very significantly elliptical it would intersect the earth
14:46 < fenn> perigee is 500 km
14:46 < muurkha> since it intersects the impact point?
14:47 < muurkha> how low is the apogee if the perigee is 500 km MSL and the impact point is, like, 150 km?
14:47 < maaku> fenn: so long as the ASAT doesn't intercept with too much vertical velocity, any kessler syndrome should be limited to <500km and last only a few years
14:47 < fenn> the impact point isn't 150km
14:47 < fenn> starlink is in a 500 km circular orbit
14:47 < muurkha> I see, thanks
14:47 < maaku> but lots of unknowns here... it's quite possible that orbital resonances could kick some stuff into higher orbits
14:48 < muurkha> orbital resonances with... other bits of debris?
14:48 < maaku> oh i should have read the full scrollback sorry
14:48 < maaku> muurkha: yeah
14:50 < fenn> i'm not sure about the velocity distribution of the debris cloud when a satellite hits a stationary (relative to the earth) projectile
14:51 < muurkha> is that really a thing that can happen?  as I said I'm not an expert on orbital dynamics but it seems like even if two bits of debris are in the same orbit their mutual gravitation is going to be insignificant on human timescales
14:51 < fenn> clearly most of it goes along roughly the original trajectory of the satellite
14:51 < muurkha> like, 0.1 kg vs. the 6 000 000 000 000 000 000 000 000 kg of Earth
14:52 < muurkha> over a distance of the same order of magnitude
14:52 < muurkha> (due to gravity repel)
14:52 < fenn> muurkha: no, gravitational effects between satellites are negligible
14:53 < muurkha> okay, good
14:53 < fenn> muurkha: it's possible that the moon could affect things in higher orbits, like GEO
14:53 < muurkha> sure
14:53 < fenn> it seems unlikely that debris from LEO would make it to GEO
14:55 < fenn> if a hydrazine tank exploded, it would have to impart 3.6 km/s to the pieces flying apart
14:55 < fenn> but the speed of sound in liquid hydrazine is only 2 km/s
14:56 < muurkha> in an impact you'll get pieces going past escape velocity I intuit
14:56 < muurkha> since even the relative velocity between two LEO objects in opposite direction is higher than escape velocity
14:57 < fenn> well, gaussian distributions are a nice intuition but not always a perfect match for reality
14:58 < maaku> OK I admit I’m confused. When the ASAT test happened there was some debris that has higher perigee and apogee than the test itself. I’m not sure the mechanism that caused that
14:58 < muurkha> yeah, I don't have any idea what the velocity distribution will be
14:58 < maaku> I thought it was orbital resonance, but you’re right that effect should be negligible on human timescales.
14:59 < muurkha> maaku: maybe the debris happened to be outgassing at apogee?  also I (we?) were confusing perigee and apogee earlier
14:59 < fenn> maaku: can you link to some data showing higher perigee?
15:01 < fenn> http://celestrak.com/NORAD/elements/table.php?GROUP=2019-006&FORMAT=json says 427km perigee, whereas the impact happened at 285km (?)
15:01 < muurkha> speaking of people learning to design things through simulation, I'd probably have a better grasp on this stuff if I'd played Kerbal Space Progra
15:01 < muurkha> m
15:02 < fenn> definitely
15:04 < muurkha> it would be a lot easier for the Russians to blow up the comparatively small number of US imagery satellites, and probably produce a larger operational advantage
15:04 < fenn> ah nm kalamsat is something else
15:04 < fenn> they both start with 2019-006
15:05 < darsie> muurkha: There are tons of KSP videos, too.
15:05 < fenn> it's more fun to just play the game though
15:06 < darsie> KSP1 can be quite cheap these days.
15:06 < muurkha> darsie: yeah, but watching them takes as much time as playing the game and teaches you less
15:06 < muurkha> and also it's less fun yes
15:06 < darsie> When it's on special.
15:06 < muurkha> proprietary tho
15:06 < darsie> We're waiting for KSP23
15:06 < darsie> We're waiting for KSP2.
15:06 < darsie> Go for orbiter. It's more realistic, too.
15:06 < muurkha> foss?
15:06 < fenn> sorta
15:07 < darsie> not sure if it's FOSS, already.
15:07 < darsie> GMAT is for mission planning. Trajectory planning. Pretty involved, though. Very accurate.
15:08 < darsie> GMAT is used for real missions.
15:09 < muurkha> I forget what we used at Satellogic but I wasn't doing it
15:09 < darsie> There is/was a free (as in free beer) demo. Very old. Could ask in #kspofficial @ espernet where it can be found.
15:09 < fenn> orbiter was released as open source in 2019, but not sure if this includes everything
15:09 < darsie> free KSP demo
15:10 < fenn> erm, 2021
15:10 < fenn> but orbiter is boring and for flight simulator weenies
15:10 < fenn> you have to already be fanatical about aerospace to even be able to play orbiter, whereas KSP installs this fanaticism into you
15:12 < darsie> https://xkcd.com/1356/ Orbital Mechanics
15:12 < lsneff> ksp was my life in highschool
15:12 < lsneff> I think I had between 300 and 400 hours in it
15:13 < darsie> lsneff: Did you complete the tech tree and fully upgrade all buildings?
15:13 < lsneff> Most of my tenure was before ksp had a career mode
15:14 < darsie> Ah. I got it after that.
15:14 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has left #hplusroadmap []
15:14 < fenn> ah, a tenured professor of KSP :D
15:14  * fenn genuflects
15:15 < fenn> i landed on the Mun and rolled around
15:15  * lsneff bows
15:15 < lsneff> I was really into SSTOs
15:15 < darsie> I did a single stage Mun sample return mission, IIRC.
15:15 < fenn> then i tried to do first stage recovery and lost interest when i discovered that KSP just destroys things that are too far apart
15:16 < fenn> never got into modding
15:16 < darsie> Now there are mods that allow you to do it.
15:16 < lsneff> there was a multiplayer mod haha
15:16 < darsie> ckan makes that easy.
15:17 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has joined #hplusroadmap
15:17 < darsie> It's a mod package manager dealing wiht dependencies.
15:17 < darsie> Like debian's apt.
15:18 < lsneff> I'm really looking forward to ksp2, though I don't know how I'll find the time to play
15:18 < muurkha> nice
15:18 < fenn> you'd think that all of those mods could be scraped together into a halfway decent KSP clone
15:19 < lsneff> There are def enough mod assets for a whole new game
15:20 < darsie> I tried RSS/RO/RP1 again, recently, but something was frustrating again. Maybe fairings.
15:21 < muurkha> RSS/RO/RP1?
15:25 < darsie> real solar system (Earth, Moon, ...), realism overhaul, realistic progression 1 (tech tree).
15:25 < fenn> another educational game that everybody should play is "autonauts" which is about simple programmable robots and bootstrapping
15:25 < muurkha> oh, are those mods for KSP?
15:25 < darsie> yes
15:25 < darsie> With lots of dependencies.
15:25 < muurkha> everybody, fenn, or just people who don't know about programming?
15:25 < fenn> the free to play demo is about the same game mechanics as the published final version https://denki.itch.io/autonauts
15:26 < fenn> dunno muurkha 
15:26 < fenn> i enjoyed it even if i didn't learn any profound subtleties about the universe
15:27 < fenn> i think it would make a good IQ test
15:27 < fenn> "here kid/drone-candidate/employee, play this for the next week" and see how far they get
15:28 < fenn> if they score low they become a janitor, if they score high they become a manager, if they're extremely high performing you put them in charge of civilization
15:29 < fenn> unfortunately we don't live in a technocracy
15:30 < fenn> i guess it's as much about management as it is about programming
15:30 < fenn> the game is too easy though, and there's no "hard mode" at all, so you have to invent your own challenges, and most people don't do that
15:31 < fenn> and the programming language is severely lacking in higher level concepts, and you can't really make your own abstractions at all
15:32 < fenn> certainly someone should make a better version of autonauts, but i haven't seen anything else like it
15:34 < muurkha> haha
15:35 < fenn> they used to do aptitude tests but this has become illegal or something
15:35 < muurkha> not always
15:36 < muurkha> you just have to show it's a BFOQ in court if you get sued
15:36 < fenn> but the best aptitude tests are generic and therefore not a BFOQ
15:37 < muurkha> you might be able to make an argument that strong general intelligence is a BFOQ
15:37 < muurkha> a police department successfully defended the use of IQ tests for hiring in the US a few years ago
15:37 < muurkha> they didn't want to hire high-IQ cops who would quit as soon as they found a better job
15:39 < fenn> i'm talking about aptitude tests like given by a guidance counsellor in school, which determine your course of study and personal development, not necessarily a hiring filter
15:39 < fenn> counselor*
15:40 < fenn> hm. "Counsellor is the preferred spelling everywhere outside the U.S."
15:40 < muurkha> IBM used to give those for hiring, that's how my friend Ann got a job writing operating systems
15:41 < muurkha> and I think guidance counsell?ors can still do that without having to defend it in court
15:42 < fenn> i'd like to believe that everyone would be happier in a society where everyone is good at the job they end up in, but i don't actually know if it's true
15:42 < fenn> performing right at the edge of your capability level
15:42 < fenn> i certainly never got any data based career recommendations
15:43 < fenn> there's got to be more to it than just IQ, which is why i'm talking about aptitude tests
15:43 < fenn> short term memory, ability to avoid distractions, extroversion, etc.
15:44 < fenn> it just seems like the whole concept has fallen into disfavor
15:44 < fenn> "who are YOU to tell me what to do?! that's RACIST!"
15:45 < fenn> but like, maybe your life would actually be improved if you were a horse trainer instead of a middle manager
15:45 < fenn> ok maybe those jobs are too similar
15:45 < muurkha> yeah
15:46 < fenn> a hot tub installer instead of a middle manager
15:46 < muurkha> I didn't get *capability* based career recommendations but I did get *interest* based career recommendations
15:46 < muurkha> we had a system installed on our VAX called the Guidance Information System
15:47 < muurkha> it had a database of jobs and what personality and interests types they benefited from
15:47 < muurkha> *personality types and interests
15:47 < muurkha> so I told it what my interests were and I forget what it recommended
15:48 < fenn> well, garbage in garbage out
15:49 < fenn> i don't believe in surveys as a valid psychological approach
15:49 < fenn> a person will answer what they believe to be the highest status answer, or what they think the test administrator wants to hear
15:50 < fenn> nobody really understands that they're neurotic and disagreeable
15:50 < fenn> and by how much
15:56 < muurkha> I have the openpsychometrics percentiles written down somewhere.  most of them were pretty extreme
15:57 -!- Jay_Dugger [~jwd@47-185-224-111.dlls.tx.frontiernet.net] has joined #hplusroadmap
16:25 -!- srk [~sorki@user/srk] has quit [Ping timeout: 260 seconds]
16:26 < kanzure> docl: pmetzger was here once upon a time
16:32 < kanzure> until i scared him off.
16:32 < muurkha> how did you do that?
16:34 < kanzure> don't remember, probably told him to roll over and die
16:34 < kanzure> as one does
16:35 < muurkha> not much chance he'll come back in that case
16:38 -!- srk [~sorki@user/srk] has joined #hplusroadmap
16:47 -!- xaete[m] [~xaetematr@2001:470:69fc:105::a438] has quit [Ping timeout: 240 seconds]
16:49 -!- xaete[m] [~xaetematr@2001:470:69fc:105::a438] has joined #hplusroadmap
16:54 < fenn> http://gnusha.org/logs/2010-07-27.log seems to have been some sort of misunderstanding
16:55 < fenn> can't imagine anyone would hold a grudge over that
16:57 < kanzure> wait was that me?
16:57 < fenn> you were controlling gnusha yes
16:59 < kanzure> apologized in 2016 but didn't work
17:01 < fenn> well at least xentrac came back :)
17:01 < muurkha> one might not hold a grudge but one might also carry a low prior probability that returning will be worthwhile
17:02 -!- Codaraxis [~Codaraxis@user/codaraxis] has joined #hplusroadmap
17:02 < muurkha> coming back took a lot of persistence on my part (and nsh's)
17:02 < fenn> well it might not be worthwhile
17:02 < fenn> we don't really accomplish a lot
17:03 < fenn> as a channel anyway
17:03 < fenn> i suspect pmetzger is jacked into the same mailing lists kanzure gets all his links from anyway
17:03 < muurkha> there was a period of several years where kanzure would tell anyone who asked that I was involved in some unspecified evildoings
17:03 < jrayhawk> i am amused how wrong he was and how much he kept going even after you told him off about it
17:04 < muurkha> 02015 to 02017 I think
17:04 < muurkha> so I figured that keeping a distance was a safer policy
17:04 -!- aedla [~aedla@0f71-0000-0000-0000-5f00-87d8-07d0-2001.dyn.estpak.ee] has quit [Quit: Client closed]
17:04 < fenn> i never understood your rationale for leaving
17:05 < muurkha> I never did figure out whether the supposed evildoings were that I was dating a 19-year-old or that I volunteered in the slums
17:05 < muurkha> or what
17:05 < muurkha> and now kanzure says he can't remember
17:06 < muurkha> dunno, global writability sounds like a reasonable solution to me.  or you could just share a password
17:06 < muurkha> fenn: does that clear it up?
17:07 < jrayhawk> there is literally one sane approach to the problem, and that is a default ACL. everything else is a dumb hack.
17:07 < muurkha> chmod g+s and then umask 002 is an easier solution
17:07 < jrayhawk> g+s does not propagate safely to subdirectories
17:08 < muurkha> umask 002 is fine now that every Linux distro ha sper-user groups (and did in 02010 too)
17:08 < muurkha> when doesn't g+s propagate to subdirectories, jrayhawk?
17:09 < jrayhawk> when it is not the primary gid
17:09 < muurkha> you mean that if you create a subdirectory it inherits your process primary gid instead of the directory's gid?
17:10 < jrayhawk> yes
17:10 < muurkha> not on my machine
17:10 < muurkha> I just tested it
17:11 < muurkha> maybe it's a bug in an ancient version of Linux (or some other Unix) that's been fixed for decades?
17:11 < jrayhawk> interesting; which distribution is this? I wonder if there's a linux-specific sysctl for this
17:11 < muurkha> Debian
17:11 < muurkha> try it: mkdir foo; cd foo; chmod g+s .; sg sudo -c mkdir bar; ls -la
17:12 < muurkha> uh
17:12 < muurkha> no, that's right
17:13 < muurkha> not quite: mkdir foo; cd foo; chmod g+s .; sg cdrom -c 'mkdir bar'; ls -la
17:13 < muurkha> for whatever reason sg was prompting me for a password for the sudo group
17:15 < jrayhawk> default debian is not propagating the writability
17:16 < muurkha> no, the writability depends on your umask
17:17 < muurkha> probably if you say `umask` it will say `0022`
17:17 -!- Jay_Dugger [~jwd@47-185-224-111.dlls.tx.frontiernet.net] has quit [Ping timeout: 260 seconds]
17:17 < muurkha> it can be a bit of a pain to get everyone's umask fixed, usually requires a few go-rounds
17:18 < fenn> maybe this is a dumb question, but what controls this logic? is it the kernel?
17:18 < jrayhawk> yes, we could do a dumb thing, or we could use a default ACL
17:18 < muurkha> if they have half-understood blocks of text they've pasted into their .bashrc or something
17:19 < muurkha> yeah, umask and filesystem permissions and ownership are a kernel features
17:19 < muurkha> I don't think they're even in the filesystem code except in the sense that the filesystem code is responsible for saving and restoring them
17:20 < muurkha> I don't think there's a significant difference between using group writability and using ACLs in this case except that ACLs are harder to debug
17:21 < jrayhawk> no, POSIX has no other concept of a subdir-propagating default
17:21 < muurkha> are you saying Linux's behavior of propagating the group ownership and sgid flag to subdirectories is not specified by POSIX?
17:22 < muurkha> also ACLs are not supported by as many filesystems (but who cares if the one you're using does support them?)
17:22 < jrayhawk> the writability is the problem here
17:22 < muurkha> right, you solve that by setting umask 002
17:23 < jrayhawk> there is no way to provide a subdir-specific umask
17:23 < muurkha> yeah, but you don't need to; you make sure everyone has umask 002 set all the time
17:23 < muurkha> it's unusual for people to go around changing their umask, they just inherit it from pam_umask
17:24 < muurkha> now that every Linux distribution has per-user groups there's no reason to ever set your umask to 022.  007 might be a reasonable default for some people some of the time but that also doesn't cause problems
17:25 < jrayhawk> oh my god
17:27 < jrayhawk> april 1st was four weeks ago
17:28 < jrayhawk> just put inode-specific policy in the inode for god's sake
17:29 < jrayhawk> the closest thing to get to sane policy for what you're describing is to write some BPF filter for anyone trying to do anything in a given directory with the wrong umask
17:29 -!- srk [~sorki@user/srk] has quit [Ping timeout: 240 seconds]
17:29 < muurkha> hmm, that sounds like it would be even harder than ACLs to debug when it breaks
17:30 < muurkha> there's a Unix feature that does what kanzure wanted and has since the 01970s, it was just error-prone until the introduction of per-user groups in the 02000s.  you might as well just use it
17:31 < muurkha> every backup program will back it up correctly, every Unix filesystem supports it, you can debug it with ls if you get it wrong
17:32 < muurkha> it's not as expressive as ACLs, it's true
17:35 < muurkha> but that isn't needed in this case
17:35 < muurkha> (to be fair, ls -l *does* tell you there's an extended ACL on the file, but it doesn't tell you what it says)
17:37 < jrayhawk> what
17:38 < jrayhawk> to be fair, ls -l *doesn't* tell you there is a umask policy on the inode, let alone what that policy is
17:39 < muurkha> there's no such thing as "a umask policy on the inode" because umask is an attribute of processes, not inodes
17:39 < jrayhawk> CORRECT
17:40 < muurkha> ls -l *does* tell you the permissions on the file so you can tell why you can't write it (even over ftp, relevant to the original use, which is not true of getfacl)
17:40 < muurkha> it doesn't tell you how they got set wrong, though it does tell you who set them wrong
17:41 < jrayhawk> FTP users can't do anything about the wrongness either way
17:41 < muurkha> they can actually; they can get the file, delete it, and put it again
17:41 < muurkha> assuming they have write access to the directory
17:41 < jrayhawk> they can't do that with subdirectories
17:42 < jrayhawk> which are the thing at issue
17:43 < muurkha> true, they can't.  the best they can do is move the subdirectory out of the way and recreate it.  but at least they can see what is broken if it is broken
17:44 -!- srk [~sorki@user/srk] has joined #hplusroadmap
17:46 < jrayhawk> well, anyway, kudos to kanzure for doing the smart thing and obviating this discussion
17:46 < jrayhawk> the last time, that is
17:47 < muurkha> heh
17:50 < jrayhawk> FWIW the specific transparency issue you're complaining about is nonexistent on Linux; I don't know about other unices, though
17:51 < jrayhawk> default:group::rwx
17:51 < jrayhawk> will literally just set the normal permission bits
17:54 < muurkha> but if you accidentally only set group::rwx instead it looks the same in ls -l and has the same behavior when you're creating files (except that it overrides your umask, which is a potential security hole)
17:56 < muurkha> oh, no, just on the directory itself
17:56 < muurkha> you really do need the default ACL to get any inheritance, not just to subdirectories
17:58 < muurkha> which is, you know, a saner design than the original Unix design.  just, less supported by things like backup tools and harder to debug
17:58 < fenn> well i'm glad we resolved that
17:58 < fenn> back to your ordinary scheduled lurking
17:58 < jrayhawk> that's what i do best
18:08 < jrayhawk> fenn: if you have manpages-dev installed, 'man 2 umask' is the complete and concise documentation of Linux's implementation, including ACL interaction
18:11 < fenn> nice
18:12 < jrayhawk> strace is a little bit deceptive in that it displays umask as if it were an argument to the relevant syscalls (mkdir, etc.), so don't expect it to be accurate in detail
18:17 -!- test_ [flooded@gateway/vpn/protonvpn/flood/x-43489060] has joined #hplusroadmap
18:20 -!- _flood [flooded@gateway/vpn/protonvpn/flood/x-43489060] has quit [Ping timeout: 240 seconds]
18:22 < muurkha> that's a useful feature of strace, I didn't know that
18:23 < muurkha> maybe it's a new feature?
18:39 < jrayhawk> OTOH reading 2010 jrayhawk say '17:08 < jrayhawk> I also like forcing people to use git for object storage.' makes me wish i could reach back in time to slap myself upside the head
18:39 < jrayhawk> wait, no, that was before github employees trashed git
18:40 < jrayhawk> nevermind
18:40 < jrayhawk> still dumb, but not slap-upside-the-head-dumb
18:47 < muurkha> git for object storage is pretty reasonable up to a point
18:48 < muurkha> like, 10MB objects are fine, 100MB objects are very iffy, 1GB objects are not good at all
18:49 < jrayhawk> as of 2017 it is completely unacceptable for that particular context
18:49 < jrayhawk> https://www.spinics.net/lists/git/msg298544.html
18:51 < jrayhawk> er, 2016
18:51 < muurkha> interesting, is that still the case?
18:51 < jrayhawk> it has gotten worse
18:52 < muurkha> I've never used git in a situation where this would be a problem
18:55 -!- Codaraxis [~Codaraxis@user/codaraxis] has quit [Quit: Leaving]
18:57 < jrayhawk> technically you can work around the HEAD issue by using the legacy refs/HEAD path, but 'packed-refs' and 'shallow' are broken under the present lockfile logic
18:57 < jrayhawk> which is why you can't do shallow checkouts of e.g. the diyhpl.us wiki
18:59 -!- yashgaroth [~ffffffff@2601:5c4:c780:6aa0::93] has quit [Quit: Leaving]
19:28 -!- xaete [~user12345@2607:9880:2458:58:7ae4:45dd:dfa3:2399] has joined #hplusroadmap
20:33 -!- darsie [~darsie@84-113-55-200.cable.dynamic.surfer.at] has quit [Ping timeout: 276 seconds]
22:09 -!- Molly_Lucy [~Molly_Luc@user/Molly-Lucy/x-8688804] has quit [Read error: Connection reset by peer]
23:07 -!- xaete [~user12345@2607:9880:2458:58:7ae4:45dd:dfa3:2399] has quit [Ping timeout: 240 seconds]
--- Log closed Thu Apr 28 00:00:07 2022