HNHacker News
TopNewBestAskShowJobs

dgroshev

1,778 karma · joined September 30, 2014

submissionscomments
dgroshev··on $96 3D-printed rocket that recalculates its mid-air trajectory using a $5 sensor
It's fairly important to distinguish what kind of drones are we talking about [1]. Iran's using Group 3 drones.

The GP is confusing Iran's neighbours not being ready to counter group 3 drones with the drones being inevitably effective. These drones are by necessity large and slow, because they need a lot of energy and aerodynamic efficiency to get their range. That means that they are vulnerable to cheap counters, which Ukraine is demonstrating very convincingly: even though Russia is now launching 800+-drone raids, the vast majority is shot down.

Even when those drones do get through, they are extremely inefficient. It's not just that they can't carry a heavy or sophisticated payload (more complex warheads are more effective, but way more expensive), the extremely high attrition ratio forces the enemy to try to target way too many drones per aimpoint. Instead of serving a few hundred aimpoints, the 800-strong raid is forced to concentrate on just a few, otherwise most aimpoints will get no hits whatsoever.

But also the only reason 800-strong raids can even be launched is Ukraine lacking the capability to interdict the launches. 800 group 3 drones have an enormous logistics and manufacturing tail, which a Western force would have no problem destroying way before the raid can be launched. For example, Iran in its current state can't launch such raids. So in practice Iran's neighbours would need to intercept only a handful of drones, which is hardly an insurmountable challenge.

[1]: https://mwi.westpoint.edu/wp-content/uploads/2024/05/FL1.jpg

dgroshev··on $96 3D-printed rocket that recalculates its mid-air trajectory using a $5 sensor
I'm just not sure what to even say when you're both so assertive and completely wrong. Please stop relying on twitter/reddit to inform your takes.

The war in Ukraine is being fought with all tiers of systems, ranging from Zircons and PAC-3 on the high end to booby traps on the low end. All of them are essential, and shortcomings on any of the tiers is ruthlessly exploited by the other side. Saying that it's only the small drones that matter betrays over-reliance on the gory FPV kill footage.

"QA, safety, blah blah blah" get implemented on every level as soon as it's feasible. You can just look at photos from Yelabuga and see how their assembly lines are not fundamentally different from Raytheon's. Ukraine is standardising their drone manufacturing. This is inevitable, because faulty munitions lead to

- killed friendly soldiers if the munitions explode pre-launch

- wasted logistic resources if they don't launch

- wasted time and targeting opportunity or friendly units not getting fire support when they fail after launch

The cost of faults is severe and much higher than just the cost of the munition itself.

It seems that you're misinformed about the real cost of modern FPVs used in Ukraine. Reports of sub-$1000 drones are years out of date and heavily relied on salvaged munitions, but there are only so many RPG warheads you can get for "free". Current FPVs are heavier, more capable, and cost a few thousand dollars. Further, it's reported that it takes dozens of FPVs to kill a single "hedgehog tank", which brings the total cost of one kill to a rough parity with "classic", "expensive" systems like the Javelin, except Javelins can be carried by a mobile squad, and launching FPVs requires a dedicated immobile unit with a long logistical tail.

Don't mistake forces not being ready to counter low-tier threats immediately with the threats being impossible to counter. Group 3 drones are very effectively countered in Ukraine, to the extent that it takes hundreds to deliver maybe a few TLAMs worth of payload to the target. There are mature systems being rolled out right now across western armies, from various gun-based solutions to APKWS. Group 2 drones are decimated with cheap anti-air drones. Group 1 drones are being handled with APSes, which work pretty well even in urban environments, as Israel has (very unfortunately) demonstrated lately.

dgroshev··on $96 3D-printed rocket that recalculates its mid-air trajectory using a $5 sensor
That's entirely possible, but doing so reliably and safely is difficult and expensive enough that for a very long time Ukrainians were accepting the risk instead.

The risk appetite countries in existential conflicts have is quite different from what we're used to. For example, there are plenty of videos of Ukrainian soldiers angle grinding cluster munitions open to extract submunitions to put on drones, but that's not a strategy that western armies can rely on.

dgroshev··on $96 3D-printed rocket that recalculates its mid-air trajectory using a $5 sensor
> TFA is literally about a $96 rocket

It's a firework-grade rocket with no payload that can't even ignite reliably.

To imitate even a TBM or a MBRM, you need similar kinematics, even if you're running without a payload. Maybe your solid rocket motor would be a touch smaller because you're not delivering hundreds of kilograms of explosives, but it still has to be large because of the rocket equation. With a large motor you're looking at a lot of damage if it explodes at the launch point, so you need quality casting. You can't really save much money on the motor.

Then, you need a TEL. Because the motor is large, the launcher has to be comparable to the real thing. You probably don't want to have two different vehicles, so you keep the same vehicle; it needs to be armed, driven around, and set for launch. Not that different from the real thing.

So you've done all of that, and then you realise that your empty warheads are too light and the missiles (or warheads, if you split) don't interact with the atmosphere in the same way as non-decoy missiles do. What's worse, modern radars are perfectly capable of noticing that and discriminating the decoys. All of that effort, and you didn't win anything. Might as well add the payload.

The US and the UK spent vast amount of money chasing exactly your line of reasoning with nuclear warhead decoys. Chevaline is a culmination of the effort, and it's retired for 30 years. In the end, relying on decoys doesn't really work, they are too expensive.

Fancier CPUs change very little of this calculation, because compute is a very little part of the cost to begin with.

dgroshev··on $96 3D-printed rocket that recalculates its mid-air trajectory using a $5 sensor
Ukraine does bother with all of that when they can afford it. I'd even say that FPV drones are the main exception, and only because Ukraine was so pressed for immediate results and had stockpiles to repurpose. There are only so many old RPG warheads you can reuse with a detonator made of live wires, and maiming your own launch crews because someone made a tiny wrong movement arming their thirtieth drone of the day under artillery fire gets old fast.

Hezbollah is a terrorist organisation, they don't need their contraptions to work reliably. GMLRS serves an entirely different purpose to rockets made of repurposed telephone poles, and is much more useful for a military force.

Also, don't forget the distances. Ukraine is fighting a war in their own country, with direct ground lines of communication to the frontline. On the other hand, you can fit three Ukraines just between Guam and Taiwan.

dgroshev··on Kuwaiti F/A-18's Triple Friendly Fire Shootdown Gets Stranger by the Day
None of those articles elaborates on how that particular target was selected, which is the claim the GP made. There's a difference between "Pentagon uses AI for targeting" and "AI selected a school as a legitimate target".

Ted Bundy drove a car. Cars kill people. Doesn't mean that Ted Bundy was killing people by running them over.

dgroshev··on $96 3D-printed rocket that recalculates its mid-air trajectory using a $5 sensor
100 rockets for $10k is not happening. The price floor is not dictated by the electronics (which did get cheaper), it's dictated by the rest of the system: propulsion, warheads, arming and safety, QA, traceability, climate and shelf life stability.

Take a look at Raytheon's manufacturing line: https://www.youtube.com/watch?v=kCCkVAHSzrc That's what it takes to have missiles that are nearly guaranteed to perform to specification every time. You can stockpile the packaged missiles in a non-climate-controlled shed for years, replenish them at sea while being showered with salt water, subject them to shock of a nearby blast while in a VLS, and they will still launch, go up to Mach 13, and catch an incoming ballistic missile nearly every time.

Sure, Iran's ballistic missiles are simpler than SM-3, but they are still subject to most of the constraints. They still need perfectly cast large size solid rocket motors that don't crack after being stored for a year, they need warheads that only go off when they are supposed to, they still need to trace every part for QA, etc. There's a vast gap, largely invisible to amateurs, between garage prototypes and stockpiled AURs.

dgroshev··on Okmain: How to pick an OK main colour of an image
Picking the best colour is a difficult problem, I don't think there's a good general answer short of ML magic. I assembled an adversarial collection of images:

https://github.com/si14/okmain/blob/main/test_images/IMG_134...

https://github.com/si14/okmain/blob/main/test_images/pendant...

https://github.com/si14/okmain/blob/main/test_images/pendant...

https://github.com/si14/okmain/blob/main/test_images/red_moo...

https://github.com/si14/okmain/blob/main/test_images/supremu...

For every heuristic, I can think of an image that breaks it. On the other hand, I just wanted to do better than the 1x1 trick, and I think the library clears that bar.

dgroshev··on Okmain: How to pick an OK main colour of an image
Good idea, I'll add a CLI tool over the weekend.
dgroshev··on Okmain: How to pick an OK main colour of an image
Author here: the library just accepts RGB8 bitmaps, probably coming either from Rust's image crate [1] or Python's Pillow [2], which are both mature and widely used. Dealing with codecs is way out of scope.

As for loading into memory at once: I suppose I could integrate with something like libvips and stream strips out of the decoded image without holding the entire bitmap, but that'd require substantially more glue and complexity. The current approach works fine for extracting dominant colours once to save in a database.

You're right that pre-resizing the images makes everything faster, but keep in mind that k-means still requires a pretty nontrivial amount of computation.

[1]: https://crates.io/crates/image

[2]: https://pypi.org/project/pillow/

dgroshev··on Okmain: How to pick an OK main colour of an image
Random sampling makes a lot of intuitive sense, but unfortunately doesn't work well. I just answered over at lobsters: https://lobste.rs/s/t43mh5/okmain_you_have_image_you_want_co...

I should probably add this nuance to the post itself.

Edit: added a footnote

dgroshev··on Kuwaiti F/A-18's Triple Friendly Fire Shootdown Gets Stranger by the Day
> targeted by an LLM

I don't think it's a known fact at this point.

dgroshev··on How the Sriracha guys screwed over their supplier
Personally I don't like that they do that, not that they are pointing out that they do that.
dgroshev··on The changing goalposts of AGI and timelines
Just like everyone forgot about this https://www.wired.com/story/openai-staff-walk-protest-sam-al...
dgroshev··on FreeCAD
It's probably impossible for FreeCAD to catch up with the industry-standard CAD systems (SOLIDWORKS, NX, Fusion) unless they somehow pour a stupendous amount of money into their geometry kernel [1].

All major CAD systems use mature geometry kernels like Parasolid [2]. Parasolid was developed for 40 years and is still in active development. This is the piece of code that enables CAD systems to do things like computing an intersection of a G3 smooth fillet with embossed text, handling all corner cases.

FreeCAD runs on OpenCASCADE [3], which is both less sophisticated today and is slower to gain new features than Parasolid, being seemingly maintained by one person [4]. FreeCAD's geometry is hard limited by what OpenCASCADE can do.

This is the main difference from Blender. Blender ultimately operates on vertices, which doesn't require nearly the same level of inherent complexity. Blender isn't bottlenecked in what it can do like FreeCAD is.

[1]: https://en.wikipedia.org/wiki/Geometric_modeling_kernel

[2]: https://en.wikipedia.org/wiki/Parasolid

[3]: https://en.wikipedia.org/wiki/Open_Cascade_Technology

[4]: https://github.com/Open-Cascade-SAS/OCCT/commits/master/

dgroshev··on FreeCAD
I find SOLIDWORKS for Makers [1] a great middle ground between bait-and-switch "free" Fusion and the real, very expensive, deal. SW is one of industry standards, its interface is much better than FreeCAD's, and it's more powerful than both FreeCAD and Fusion. For example, both FreeCAD and Fusion struggle with G2/G3 smoothness [2] where SW doesn't even blink. Fusion doesn't allow to pattern features on sketch points (it's gated behind an expensive add-on [3]) when it's a built-in feature in SW.

[1]: https://www.solidworks.com/solution/solidworks-makers

[2]: https://www.printables.com/model/1490911-g0-g3-corners-visua...

[3]: https://www.autodesk.com/uk/products/fusion-360/design-exten...

dgroshev··on TIL: Apple Broke Time Machine Again on Tahoe
I tried moving to NFS, but the level of complexity of NFS auth is just comical. I gave up after trying to set up a Kerberos server on the Synology that I was trying to access. It's too much.

Using unauthenticated NFS, even on a local network, is too dodgy imo.

dgroshev··on Parametric CAD in Rust
Solidworks is $48/year for hobbyists though, and Fusion is free.
dgroshev··on Parametric CAD in Rust
To add to this, constructive solid geometry systems like OpenSCAD seem to be practically incapable of doing anything related to surfacing, even if it's just G2 fillets or G2 continuity (much less more advanced work with splines and patches).

I think it's partially a positive feedback loop: the "hacker" tools (OpenSCAD) influence the aesthetic (blocky/square brackets, like in the post), which in turn influences the tools. When creating a simple fillet or a chamfer is an annoying exercise in constructive geometry rather than five mouse clicks, there'll be fewer fillets, chamfers, and flowing surfaces, even when they make mechanical or manufacturing sense; but then that becomes "the look", and people don't even think about fillets.

dgroshev··on Some notes on starting to use Django
In my experience, django-ninja is a great alternative to DRF, precisely because it's so thin, unlike DRF.
dgroshev··on Some notes on starting to use Django
"Cold-blooded software" is a great term for that https://dubroy.com/blog/cold-blooded-software/ (HN: https://news.ycombinator.com/item?id=46488261 )
dgroshev··on UK House of Lords Votes to Extend Age Verification to VPNs
It takes just a few seconds to see that it's a random backbencher who is not in the government. We have a whole range of MPs, and some of them sometimes talk about things they have no idea about. The website you're citing is little more than propaganda, since it explicitly makes it seem like the MP has any connection to the government.
dgroshev··on Turso is an in-process SQL database, compatible with SQLite
It's pretty easy to get some hard data, it's an open source project. I went to https://github.com/tursodatabase/turso/pulls?q=is%3Apr+is%3A... and looked at the last few PRs:

https://github.com/tursodatabase/turso/pull/4824/files "some performance improvements", no new tests

https://github.com/tursodatabase/turso/pull/4820/ "fix wal checkpoint", one basic test

https://github.com/tursodatabase/turso/pull/4815/ "Optimizer: fix bugs, improve cost model", a lot of nontrivial logic, no new tests

https://github.com/tursodatabase/turso/pull/4814 "WAL auto truncation: increase epoch to prevent stale pages reuse", there's a new test with a comment "It is slightly fragile and can be removed if it will be unclear how to maintain it"

https://github.com/tursodatabase/turso/pull/4806/ "Busy snapshot bugfix" with two new tests with the same comments as 4814 (I guess they didn't fix the bug in one go?)

https://github.com/tursodatabase/turso/pull/4802/ "fix/translate: revert change that allowed index cursor with stale position to be read", fixes a data-corrupting bug, there's a regression test, good (although the original bug sounds like it should've been caught by a suite like the one SQLite has)

That's just a couple days worth of PRs.

This style of development does not inspire confidence. They develop features, sure. But I want my database to be rock-solid and completely covered by tests, not just move fast and break things. It's not FUD to just look at how they approach PRs.

dgroshev··on A Year of 3D Printing
CF filled filaments aren't a scam, they just have different tradeoffs: more rigid, nice surface finish, and less warping (all useful qualities!), but also lower layer adhesion, lower tensile strength, and less toughness. It's not just anecdotal, Igor Gaspar collected a lot of hard data on that.
dgroshev··on A Year of 3D Printing
There are different kinds of carbon fibres. It'd be great if it wasn't just Prusa disclosing which type they are using or offering studies on their impact, but in the meantime we can just use CF Prusament to be sure: https://www.reddit.com/r/3Dprinting/s/nQ5zUwnrWZ
dgroshev··on A Year of 3D Printing
That's largely a myth.

This review [1] cites the absolute highest amount of emitted styrene in the studies they are reviewing to be 113 μg/min. Using [2] for simplicity with styrene's molar mass (104.15 g/mol), we get to a printer creating at most 0.024 ppm of styrene per minute per m3 of unchanged air. For comparison, the "work exposure limit (WEL) for styrene is currently 100 parts per million (ppm) averaged over an 8-hour day" [4].

In other words, as long as you have some air exchange in the room, you'd be orders of magnitude away from the safe work exposure limit on styrene.

It also makes sense, considering that it's a microscopic amount of molten plastic, whereas injection moulding factories work with vats of the stuff.

[1]: https://www.sciencedirect.com/science/article/pii/S135223102...

[2]: https://teesing.com/en/tools/ppm-mg3-converter

[3]: https://en.wikipedia.org/wiki/Styrene

[4]: https://www.hse.gov.uk/plastics/faqs.htm

dgroshev··on A Year of 3D Printing
I'd encourage people doing engineering/functional parts to also try ASA and PC(-CF). Both are pretty easy to print on enclosed printer like Prusa Core One, and they offer unique qualities that are impossible to achieve with PLA or PETG.

Prusament PC Blend is insanely strong and stiff, I saw a 3mm PC bracket bending a high quality metal wood screw into an S-shape without breaking. PC-CF is much easier to print, looks great, and is stiffer still, even if a bit less strong. ASA looks great and is tougher than PC. Both creep less than PLA and PETG. Both shrug off 100C under load.

dgroshev··on A Year of 3D Printing
Less creep, slightly better at absorbing shocks without breaking, better failure behaviour (PLA can suddenly shatter leaving sharp edges, PETG tends to deform elastically first).
dgroshev··on A Year of 3D Printing
Unfortunately, even annealed HT-PLA-GF still creeps quite a bit. I find this to be the main problem using PLA as an engineering filament. For many parts it doesn't matter, of course.
dgroshev··on Predicting OpenAI's ad strategy
I wish we collectively had a better understanding of those ad tech shifts being fundamentally zero sum.

If we treat marketing as a black box, where are the benefits from supposedly more efficient marketing? Ad budgets are the same. People have the same amount of disposable income (or less, really). So on the both sides of the box converting ad budgets into paying customers the sum is the same. Quasi-monopolistic ad networks (Google and Meta) just hoovered up the money that previously went into other ad spaces, like local papers. Now OpenAI is going to fight for the same pie.

Ad tech is a market failure.

← PreviousPage 5 of 15Next →