Ignas Kiela

https://not.acu.lt/@ignaloidas

I do in fact exist

I'm an information sponge, so if you have some question that you think I might have an answer to, feel free to ask! Even if I won't have it off my head, I know how to look up things fast.

Public Key

npub1ceay677m38s396uvlpxkp20v8rpxtmx35z0sh7kuy3u5jxr87phsmfy7s5

NIP-05 Address

ignaloidas_at_not.acu.lt@momostr.pink

Profile Code

nprofile1qqsvv7jd00dcncgjawx0sntq48kr3sn9ang6p8ctltwzg72frpnlqmcprpmhxue69uhhyetvv9ujumt0d4hhxarj9ecxjmntvsjj7z

Publishing to

relay.momostr.pink

Author Public Key

npub1ceay677m38s396uvlpxkp20v8rpxtmx35z0sh7kuy3u5jxr87phsmfy7s5

Show more details

Last Notes

2026-03-26 21:51:13 UTC

- reply

the argument boils down to "is reproducing code speech or production"
which I guess could lead down to destruction of copyright for code if it falls into the speech pile and gov still wants to control it

2026-03-25 07:54:23 UTC

- reply

I'd hope they're moving over to NVMe over TCP these days

2026-02-12 13:19:39 UTC

- reply

depends on the human not having 100% trust for the LLM's results tho, like seemingly a whole bunch of curl's bug bounty reports were humans that just had absolute trust in LLM's results and even tho the talking didn't seem LLM, the arguments and reports did seem to fit a LLM

2026-01-14 11:10:12 UTC

- reply

oh wow, that's public sector type of salary.
Was maybe thinking about it, but not with that salary :|

2025-12-05 07:14:38 UTC

- reply

there was a blog post a couple days ago about using them for boating, where they seemed to be somewhat practical
but still feels like very much a specialized kind of thing

2025-08-20 08:59:57 UTC

- reply

you'd think that, but I've heard that stuff similar to this is not that uncommon of an occurance

2025-08-20 08:52:36 UTC

- reply

to be fair, they're pretty much legally mandated to do this, there's a whole industry of "how does this drug is actually used and what effects does it do to people" that is essentially looking through social media posts and shit. Look up pharmacovigilance, I used to work for a company doing that.

2025-08-18 06:16:20 UTC

- reply

> Tap water is always altered with chemicals due to how it must avoid microbe development inside the piping.
Not really? IDK if it's purely a Lithuania exclusive, but here it's common for tap water to be essentially straight from a boring well without any contact with air or processing besides removing iron salts because sometimes there's too much of those.

2025-08-12 12:47:16 UTC

- reply

I mean, arguably it already runs afoul of monopoly laws? But it would be way easier to argue if they still did this even when you can't pay for such games using them.

2025-08-12 12:42:05 UTC

- reply

"they can pay for it using our payment method" vs "they can't even pay for it using their payment method" is a fair bit of difference anyway. Mastercard may still push it, sure, but it ends up in courts, it will be a fair bit of difference in arguments.

2025-08-12 12:29:39 UTC

- reply

If mastercard still does, it gives ammunition that they aren't doing it just to "protect the brand", which might end up helping legal arguments.

2025-08-10 08:32:36 UTC

- reply

oooh, their philosophy *mostly* mirrors mine
just a bit of a shame in that there is no partial evaluator for clojure, if there was there would be an opportunity to find errors without a whole bunch of tests or runtime.
Honestly, it's annoying that there's very little work on more general partial evaluators - weval for wasm is the only one I know that's been worked on recently. Compilers are arguably just a special case of partial evaluation that just returns the partially evaluated code in a form that allows for faster execution.

2025-08-10 07:20:51 UTC

- reply

very true
I remeber talking how you could achive similar levels of safety as with statically typed languages by having the type-checks be dynamic *and* having partial evaluation make sure that all of the type checks aren't "compiled in". The response I got was "but that's not how type theory works", with little elaboration.

2025-07-31 06:45:00 UTC

- reply

It is a quality argument, but I don't think it's saying that it's bad quality *because* it's free software - rather that many "stallman enjoyers" tend to overvalue the fact that something is free software that they stop caring that it's bad quality.

2025-07-31 06:40:15 UTC

- reply

I don't think that's what the comment you originally were replying to was arguing for?

2025-07-31 06:20:48 UTC

- reply

while I do broadly agree, but the second cannot happen without the first. E.g. for your stuff to be usable it *has* to offer a migration path - even if that means that said migration path has to support proprietary operating systems. Putting principles over user needs breaks the migration chain for most users, and builds an ecosystem with no users.

2025-07-30 12:01:18 UTC

- reply

I've said this a bunch and I'll say it again: Free software movement is fundamentally a "purity" argument, while general ass users don't care about some "purity", they care for a solution to their problems.
Sure, that solution can follow the "pure" philosophy, but that fact alone will not win you any significant userbase besides a couple of FSF superfans (if that). You get users by solving their problems, better than others, and for consumer users it mostly doesn't matter whether your software is free or proprietary.

2025-07-14 11:49:36 UTC

- reply

btw, here's a local example on how to make an overlay, might be a good help on the "frontend" side https://github.com/geerlingguy/obs-task-list-overlay

2025-07-14 11:32:29 UTC

- reply

I think there's some native transparency support? internet says that body { background-color: rgba(0,0,0,0)!important } should work

2025-07-14 11:27:14 UTC

- reply

yeah, basically streamelements subscribes to twitch events (I don't recall exactly how), and then through presumably websockets sends appropriate notifications to open alert pages, one of which is hopefully the one in streamer's OBS browser source
if you want to build stuff like that into your streaming platform, I'd say that for MVP the basic browser source is enough, advanced stuff is a bonus (though still - it's important to provide hooks for others to trigger stuff from)

2025-07-14 11:22:49 UTC

- reply

maybe more advanced stuff, that very few people bother to set up, yes.
But the basics that most people use is essentially a browser source to some url like alerts.streamelements.com/basic/username and the website in said url gets a websocket connection and then on events just plays an animation and sound. It's not OBS specific or anything, just a webpage that plays alerts. If you got that working decently, you're at the interactivity level of an average twitch streamer - messing with advanced OBS stuff is something that only like 0.1% of streamers do, if that.

2025-07-14 11:09:18 UTC

- reply

I'm not sure what's your current architecture? I don't really understand what you're trying to do

2025-07-14 10:59:26 UTC

- reply

yeah
just look at streamelements and if you copy 80% of their most used features, you got everything that "regular streamers" want

2025-07-14 10:29:29 UTC

- reply

99.9% it's just a web source with transparency, only "big serious productions" go for something more than that, even "fully real-time collaborative stream overlay" stuff like pogly is just a web source

2025-07-08 14:21:39 UTC

- reply

my ID card has a digital certificate with which I can do digital signatures that are legally equivalent with written down ones.

2025-07-08 14:17:23 UTC

- reply

with digital signatures! And photos! Also, the chip has fingerprint info AFAIK

2025-07-08 14:13:11 UTC

- reply

oh, wait, there is already an open source card-side implementation https://jmrtd.org/

2025-07-08 14:12:00 UTC

- reply

yeah, freedom of choice is most likely not possible, but I think it might be viable for a dedicated individual to make sure the software the passport runs is open source.
Though, by FSF metrics, they might decide that the passport is purely hardware by virtue of the software on it being non-updateable

2025-07-08 12:47:22 UTC

- reply

I mean the standards are open, someone should write an open source implementation of them

2025-07-04 16:05:33 UTC

- reply

does it actually work, or does it just "feel" like it works
very different things, and without quantifiable measurements, it's hard to pin it down for sure.

2025-06-20 06:48:52 UTC

- reply

there is some talks about webrtc streaming, which (somehow) is way lighter on server side, and allows to stream to way more people without a lot of overhead, might be interesting

2025-06-19 19:48:41 UTC

- reply

I mean, you can have multiple windows and arrange them in Wayland, so I'm not sure what exactly do you mean?

2025-06-19 19:46:32 UTC

- reply

Restore has a protocol, you ask compositor to remember the window config, and ask to restore it on startup. Spawning new top level windows from a drag is possible, my Firefox is running Wayland-native, and I can drag a tab out of the toolbar to spawn a new window, and can drag tabs between windows and drag all tabs from one window into another window to leave only one open.
But I don't see much utility for determining the relative positions between windows (no good way to define it without baking in assumptions), or spawning a window at the cursor location.

2025-06-19 19:28:19 UTC

- reply

does that really need *absolute* positioning?
They don't want to "get stuck" to a single coordinate system - current way allows a lot of freedoms, like freely rotatable windows (admittedly a gimmick, but why not?), or e.g. VR Wayland compositors (trying to do that in X11 would lead to a lot of pain).

2025-06-18 09:03:32 UTC

- reply

remember that like 90% of devs barely know git beyond commit, pull, push, log, checkout - and if you only use that, it's barely acceptable to have this shit

2025-06-18 08:56:44 UTC

- reply

I've never even heard about anything this bad wtf
this is brainrot of the highest level

2025-06-12 10:07:19 UTC

- reply

I think the joke is that the movement ends up exactly the same as the regular horse, no?

2025-06-11 20:56:26 UTC

- reply

ah
I guess it's time for x11 to finaly die (thank goodness)

2025-06-11 20:51:54 UTC

- reply

I mean, not everything has to be there for a reason other than "I like it"

2025-06-02 04:33:42 UTC

- reply

I don't even think it's that really, us politics just has devolved into brainworms, and those brainworms spread into countries where political reporting is easily shared (that is, same language)

2025-06-02 04:29:53 UTC

- reply

so everybody who's politics happen in the same language in the US gets poisoned I guess
local languages as a defense mechanism from brainworms

2025-06-02 04:27:31 UTC

- reply

feels like just america is like that tbh, there's way more choice and nuance elsewhere

2025-05-28 01:13:08 UTC

- reply

IMO code-focused LLMs should have different tokenizers, to handle identifiers and syntactic stuff better, so that you could outright reject tokens that do not fit in position

2025-05-22 06:55:35 UTC

- reply

tbh in my very simple use of AI when coding, I find that it's just very shit at abstraction. It can extend stuff, sure, but offers near-zero capabilities for simplification (without spending a fuckton of money like OpenAI did with ARC). And there's only so much you can extend stuff without simplification before you get into a massive pile of spaghetti code.

2025-04-23 04:28:18 UTC

- reply

wasm would be a better shout, riscv seems dumb for such purpose

2025-03-23 22:37:57 UTC

- reply

I'd say the problem is that a lot of his other metrics have been built on top of that one metric, and it will eventually bring the other ones down together with it

2025-03-19 23:08:12 UTC

- reply

did he really?

2025-03-16 23:32:43 UTC

- reply

I see it more as a demonstration on why tesla's stance is so dumb, rather than a demonstration of an actual real life problem. Elon has been spouting shit about how cameras will be enough, and there's quite good evidence on how they aren't.

2025-03-13 15:57:57 UTC

- reply

conjugations are legit easier to learn
anglo mind is truly something

2025-03-02 02:33:44 UTC

- reply

o7 is alive and well in some places
though often in an emote form

2025-02-23 02:26:39 UTC

- reply

Honestly, I could see him being *that* concerned about his image, that appearing like he's the smartest guy in a company is more important than any competitive advantage patents would give.

2025-02-23 02:17:20 UTC

- reply

tbh it wouldn't surprise me - there is absolutely zero information on who actually does shit at Tesla or SpaceX, like absolute zero. Not even project managers or sth. speaking in PR releases or some shit. Any other company, they would share at least some of the main names from inside, where as with Elon's companies it's all Elon. Dude at seems to want to look like he does everyting at the damn companies.

2025-02-22 16:56:48 UTC

- reply

yeah, I feel like with rust, all of the worse code will end up getting exposed, and I guess some people just don't want that

2025-02-22 16:44:00 UTC

- reply

yeah, but managing you can abstract over
you can't abstract over "the only reason this doesn't leak in other drivers is because it leaks on shutting down" https://vt.social/@lina/113051677686279824

2025-02-22 16:37:14 UTC

- reply

> I don't have a citation but I remember reading last year complaints from C developers that they were being harassed by Rust maintainers to document their code better and stop making so many changes because it added a lot of difficulty to maintaining their Rust bindings and they didn't like the work of having to figure out how to safely use the APIs by having to read and understand the C code. There are various technical and historical details for why the C part operates like this.
Tbh that's just bad maintainership. Rust people try to write their interfaces in a way that you can't misuse them, and when trying to do that, they find code that's hard to impossible to not misuse, and the maintainers just don't want to fix their shit.

2025-02-19 20:53:07 UTC

- reply

Have you fucking read what's in those articles?
> As no significant British incursions into American territory had occurred by that time, and Britain's war strategy was largely defensive, the election proceeded without disruption.[3]
> While the Confederacy seemed to have survival potential in summer 1864, it was visibly collapsing by election day in November.
There's no end in sight for current war, unlike for the examples you picked

2025-02-19 20:43:43 UTC

- reply

with a fuck ton of risk reducing measures
it would be pretty fucking hard to reduce the risk of a drone strike to a polling center

2025-02-19 20:40:40 UTC

- reply

I think "bombs regularly falling on and near major cities" is wartime enough. If you cannot guarantee the safety of election goers, then you shouldn't run them.

2025-02-19 20:30:23 UTC

- reply

neat, UK was an autocracy during WW1 then I guess https://en.wikipedia.org/wiki/War-time_electoral_pact
like, I don't see any problem in not doing an election during wartime if there's no pressure from the citizens - the only pressure right now is coming from outside

2025-02-19 20:25:25 UTC

- reply

like 60% of people said that they shouldn't have elections during war, so it's not like there's much pressure from the citizens

2025-02-19 10:56:30 UTC

- reply

oh
maybe viable without changing os that much then

2025-02-19 10:46:15 UTC

- reply

my evil idea for game graphics: run the game loop render everything at ~1000FPS at low res, and then do the whole optical flow and temporal super sampling stuff but in reverse. You'd get proper (not fake!) motion blur and stuff with this approach, but I don't think current operating systems can truly keep up with such update rate.

2025-02-06 20:04:48 UTC

- reply

eeehh, not sure, FIPS is still joked around as "insecure mode" quite widely
sure it's better and not obviously broken
but it's not good either

2025-02-06 20:02:51 UTC

- reply

maybe
it's still nicer here though
fedi resists change, and because of that, it grows slowly

2025-02-05 15:51:30 UTC

- reply

https://not-acu-lt.s3.nl-ams.scw.cloud/null/10750668-084f-44bb-beb4-6a0421fa59c2.png

2025-01-28 12:25:32 UTC

- reply

reverting to 1980s watching culture ig
https://not-acu-lt.s3.nl-ams.scw.cloud/null/f01ad2b9-89c6-462f-993e-dee82ef10a2f.png

2025-01-22 23:12:27 UTC

- reply

they find people who can
this paper is about imaging stuff with mri, some fun notes about performance https://www.bmj.com/content/319/7225/1596

2025-01-22 23:05:39 UTC

- reply

there's just JS, and there is DOM soup
few people make proper DOM soup that works well with screen readers

2025-01-22 07:27:30 UTC

- reply

> if someone wanted to, they could point at this statement and state that the death of each spermatozoid and egg cell is the killing of a person
I doubt it would devolve that far, but if it did, there would be a lot of onion tier headlines

2025-01-22 06:49:13 UTC

- reply

incredible, how did they define it?
god, it's gonna be incredibly funny watching this clown show, it's just the fact that it will likely end up influencing me in some way that is concerning

2025-01-20 19:57:22 UTC

- reply

oh, for sure
but the state of politics in USA from my view has essentially devolved into a grudge match where each side is trying to make the other one hurt as much as possible without any consideration for others, and such moves start making sense.

2025-01-20 19:41:34 UTC

- reply

I mean, it's not really for future crimes, it's for future litigation
american political climate seems incredibly stupid in general though, even compared with some quite stupid stuff here

2025-01-12 19:24:53 UTC

- reply

IPFS in general is quite shit. It's essentially reinventing torrents+DHT, just with a shittier protocol. IPNS (or whatever the stable naming thing of theirs is called) is more interesting, but seemed hopelessly tied to the rest of the shit to be useful.

2025-01-12 18:54:13 UTC

- reply

Eh, not sure if that's necessarily true.
Natural fires are way rarer, because you need an ignition source (lightning I guess), and dry enough conditions for said ignition source to actually ignite and propagate.
With humans, both the ignition sources are way more common, and the ignition sources are way more potent, able to start fires without needing for it to be that dry.
I feel like that's the biggest reason for more fires - more and more potent ignition sources.

2025-01-12 18:41:33 UTC

- reply

I mean there's fairly few cases that fire "just happens", most of it is human-caused one way or another. Electrical lines sparking or arcing is human-caused, but isn't meaningfuly an intentional fire.

2025-01-10 04:45:32 UTC

- reply

lol'ing at a "decentralised currency" having vendor lock-in

2024-12-18 19:37:56 UTC

- reply

I mean just this alone would prevent choking in osu! and the like, in games where you need to perform for prolonged periods with the stress gradually rising

2024-12-18 19:34:46 UTC

- reply

huh
most e-sport doping seems to be stimulants, but I guess reducing stress is one way to do that as well
I can imagine quite a few games where this would help

2024-12-17 16:25:49 UTC

- reply

any reason for yices2 as a SMT solver of choice?

2024-12-10 08:19:41 UTC

- reply

"my distant relative was tortured by some organization, therefore it's fine to work for essentially the same organization just on a different side" is certainly a take
I would say this is mean to him if somebody showed me this as satire.

2024-11-09 20:50:47 UTC

- reply

sun is not commonly identified as an aircraft though

2024-10-15 22:39:27 UTC

- reply

circuits are still not required
you can do arbitrary ratio splitters, so that 1/41th of all input items output on one belt while the rest output at another. Just quite a lot harder than circuits.

2024-10-11 20:54:49 UTC

- reply

hmm, but there's 7 total layers where a connection could be, top layer, 3 horizontal undergrounds, and 3 vertical undergrounds. Not all at the same time but still.
I guess you mean the top layer as poly, but then consider long-handed inserters - I feel like that's way more fit for a poly layer.

2024-10-11 12:46:47 UTC

- reply

basically every belt tier has a separate underground level
https://not-acu-lt.s3.nl-ams.scw.cloud/null/7f08feee-f035-4f14-ad81-646ff343a270.png

2024-10-11 12:44:09 UTC

- reply

look up belt weaving
and then vertical + horizontal underground belts don't interact.

2024-10-11 06:54:40 UTC

- reply

it's more than one layer
it's 7 layers, but with a bunch of restrictions

2024-10-11 06:35:53 UTC

- reply

makes me wonder now if any EDA software uses SAT solvers to find optimal solutions to complex sub-problems

2024-10-10 20:43:37 UTC

- reply

not really though? It still goes back the call path as usual, and is handled by the code that catches errors, it doesn't jump into another context.

2024-10-10 20:41:30 UTC

- reply

essentially compiler carries a "stack trace" object through function calls, and when a function returns an error, it logs that into said object before returning. A mix between error values and regular exceptions
https://ziglang.org/documentation/master/#toc-Implementation-Details

2024-10-10 20:29:51 UTC

- reply

what about zig's error return traces
(ik not teeechnically exceptions, but essentially does the same job)

2024-09-18 08:53:27 UTC

- reply

sometimes I'm sad that I don't get dreams, and then I look at your posts and the sadness disappears

2024-09-11 22:46:41 UTC

- reply

not unusual, there are some playstyles where it's very easy to fall down into a pit of standardization
if you don't want that, imo the best ways are either to embrace spaghet (just throw those belts wherever, never fix them up) or go with a design so deranged, that there's fun in finding optimizations for it

2024-09-10 18:38:18 UTC

- reply

I mean logic gates with belts and trains are pretty straightforward
Here's belts, it's quite self-explanatory https://www.youtube.com/watch?v=Ab15R1qSw_A
Trains are a bit different, but you can reason about them in logic gates terms too https://www.reddit.com/r/factorio/comments/e6jl7b/trains_are_turing_complete_i_think/
Inserters would probably end up as weird machines, but even then, you're loosing a lot of interesting choices in the game

2024-09-10 18:30:23 UTC

- reply

I'm excited to say that this would remove almost all logistics options in the game
Only bots can't build a computer, trains and belts certainly can, and inserter mechanics are complex enough that it's highly likely that it's possible, just nobody tried that yet.

2024-09-09 22:14:55 UTC

- reply

so about the frequency...
there's a window where you're "supposed" to run it at, but the numbers in the spec are defined in this delightful way where it technically permits any frequency
https://not-acu-lt.s3.nl-ams.scw.cloud/null/3fd1e031-2073-4b2d-a3d5-346463ef3660.png

2024-09-09 22:10:20 UTC

- reply

I did say that it's an infohazard. The specification is tough to find these days, thankfully.

2024-09-09 20:00:57 UTC

- reply

PD 1.0 is one of those specifications that's been intentionally erased as much as possible from public consciousness because of it's infohazards, just like Wireless USB
https://not-acu-lt.s3.nl-ams.scw.cloud/null/1c9ea5e8-bab5-482d-b3ce-cf725028aede.png

2024-09-09 19:14:24 UTC

- reply

Of which one?
PD 1.0 is certainly incredibly wall-punch worthy

2024-09-06 22:32:47 UTC

- reply

look, at least this is more than one character per variable

2024-08-22 19:39:03 UTC

- reply

> (since the details of the TBT3 wire protocol are not public at all)
they mostly are, in the USB4 spec, section 13, Interoperability with ThunderboltTM 3 (TBT3) Systems
Intel released them specifically for compatibility. It's not the original spec variant, but it seems to be just enough to run TBT3 once you have the rest of USB4 up and running