HNHacker News
TopNewBestAskShowJobs

arghwhat

9,034 karma · joined January 12, 2017

no.
submissionscomments
arghwhat··on TurboDiffusion: 100–200× Acceleration for Video Diffusion Models
A very, very different mechanism that "just" displays the scene as the author explictly and manually drew it, and yet has to pull an ungodly amount of hacks to make that viable and fast enough, resulting in a far from realistic rendition...

This on the other hand happily pretends to match any kind of realism requested like a skilled painter would, with the tradeoff mainly being control and artistic errors.

arghwhat··on I'm returning my Framework 16
> I think this is much less general than you make it out to be and has an extremely strong dependency on how you use the thing and of your preference. It makes me think of the boot theory.

The boot theory is different: It is about buying something not crafted with proper materials which will quickly fall apart and cost more in the long run. However, unlike technology, there is no cowhide 2.0 coming out 6 months after, with all leather made the cowhide 1.0 way instantly dropping in value and fading into irrelevancy.

Low-end laptops tend to be build around older SKU's which are no longer of interest but function no worse than they did when launched, and mid-tier laptops tend to be made with current SKU's in the more reasonable binning categories.

At the same time, the replacement rate also means that whatever high-end laptop will soon have its rear handed to it by a mid-tier machine of the next chip generation. While it's outdated within a year or two you of course won't replace it that soon, but that upgrade might stretch your upgrade schedule 50% to justify it - from, say, 4 years to 6. More years of being generations behind.

Keeping a Framework around for longer also only makes sense when considering upgradability.

> A case in point: at work we have HP Elitebooks. The brand-new 2025 models I see people receive have worse screens and trackpads than my 2013 MBP

HP Elitebooks are expensive, high-end machines. The issues you see is because HP, Dell and Lenovo all cater to IT department buying strategies, and therefore all offer a bargain screen option as the machines are bought in bulk and mostly used docked. You'll find that they also have an mid-tier screen option (usually a color-accurate 1080p or 1440p panel), as well as a high-end option (say, 4k OLED touchscreen with wacom digitizer).

The trackpad is a different story entirely. If you're used to a Mac trackpad, things are a bit grim on the PC side. There's some nice ones coming out though.

> They also degrade from daily use: the screen hinge loosens so it moves if you look at it wrong

Everything degrades from daily use. My hinge loosened on my last Mac, the screen damaged itself because the 15" panel did not have the necessary rigidity to main the intended 0.5mm gap to the keyboard keys (confirmed as the second panel did the same), the shitty magsafe port overheated as using pogopins for high power transfer is a terrible idea, all I/O on one side died, and the battery had inflated at least once...

My Dell XPS 13 costing 1/3rd never had any problems, and when replaced it just felt a bit slow and had a somewhat aged battery.

Price is not an indicator of quality or expected reliability, nor is the brand itself a reliable indicator. Use common sense, take a look at the product and avoid the bottom of the bargain bin.

---

Do get a good mouse and keyboard though, it has a much more direct impact on your user experience and a mouse that costs twice as much isn't as much of an issue as a laptop costing twice as much. The keyboard and mouse also lasts longer if kept well.

arghwhat··on I'm returning my Framework 16
The pricing when I looked was similar. I went with a Lenovo last time because the Framework 16 hadn't quite matured, but premium anything is never going to make financial sense.

Buying and repairing a framework is never going to be cheaper than going through consumable trash laptops, and buying top of the line laptops and trying to use them longer is never going to be cheaper or better than buying medium grade laptops and upgrading more often.

What you're paying for right now is the customization capabilities and the ideology. Upgrading and customizing a single platform with a community, vs. a fixed one-off design that'll be lost next time you upgrade.

If Framework isn't already compelling to you at this time, then you're not the target audience. They might drop in price, but they'll never win a race to the bottom.

arghwhat··on Dependable C
Yes but someone might also have silly opinions from having no experience how production assembly actually looks, such as underestimating just how different working with that is to working in high-level languages like C and why such languages were quite revolutionary. :)

This should not be mistaken as appeal to authority, it is merely reasonable discrimination between those speaking from experience, and those forming opinions without experience.

If one believes those with experience has poorly informed opinions, they're always free to gain experience and associated perspective. They will then either have the fundamentals to properly push their viewpoint, or end up better understanding and aligning with the common viewpoint.

arghwhat··on Dependable C
Java also targets an abstract machine model (JVM) - such statement really doesn't mean much.

Assembly is not about corresponding to exactly which gates open when in the CPU. It's just the human writable form of whatever the CPU ingests, whereas C is an early take on a language reasonable capable of expressing higher level ideas with less low-level noise.

I seriously doubt anyone who has written projects in assembly would make such comparisons...

arghwhat··on Google confirms Android attacks; no fix for most Samsung users
That's still one OS. Customization is mostly userspace "system" apps that they swap out and maintain, but reused across all their phones with some small variation. Hardware enablement will differ between models, but that's just the cost of doing business.

Can be a pain to move the whole suite to a new major (porting all their inhouse apps, getting all the hardware enablement from vendors updated to match, ...), but we're not dealing with a major upgrade here.

A security patch is "just" a matter of taking the last release, applying the diff, build, qa, release. No customization.

arghwhat··on Accepting US car standards would risk European lives
(In the context of the discussion about these vehicles in the EU)

In the EU, neither would any American pickup truck: If registered as a normal class B vehicle, the total gross vehicle weight would be limited to 3500 kg (7700 lbs), and it would at most be permitted to tow 3500 kg (7700 lbs) with full independent trailer brakes, 750 kg (1650 lbs) without. You can add roughly 1000 kg if you tow a semitrailer, but getting the vehicle certified with a fifth wheel would probably be infeasible.

It doesn't make sense as a class C truck here (special driver's license, tachograph requirements for commercial use). It's way less nimble than our Scania/Volvo trucks (their turning radii are way tighter, and and have much smaller footprint for a given capacity), and is obviously a lot less capable than a vehicle that can be build from small utility up to the ~100k lbs range.

At the same time, if a farmer is outside the scope of a regular personal vehicle, they're most likely going to use their go-to tractor (e.g., Lamborghini, John Deere) which can haul anything anywhere, otherwise if they really need to haul they'll be reaching for a Scania/Volvo.

(It is common to register smaller, 7500 kg class C vehicles, but that's usually stuff like large Mercedes Sprinter vans, often built up as specialized service vehicles - think sewer inspection and repair.)

In the context of the US: It might seem like the best choice given the common options there, but I think the issue is with the options and perceived utility. It's the same with large trucks: The common ones in the EU are much more powerful, rated to haul more, are more comfortable, safer, have much smaller footprint for the given load and turns on a dime compared to US options.

arghwhat··on Helldivers 2 devs slash install size from 154GB to 23GB
Not sure if this is what they did, but you can just put all the things you need together sequentially into a single file and rely on the filesystem to allocate contiguous blocks where possible (using the appropriate size hints to help). It's trivial unpack at loading time without any performance impact.

A filesystem is by itself just one big "file" acting like a file archive.

arghwhat··on Accepting US car standards would risk European lives
You misunderstand - almost every tradesman here drives their work vehicle home and drives errands in it. Use of the company vehicle for commuting is considered a standard perk of these trades for regular employees - free fuel (fuel is way more expensive here), they can some days drive straight to the first customer (saves time), and might save them from getting a car (maybe the spouse has a microcar for their commute, otherwise biking and public transport are common).

Source: I live here and see it every day. Family, friends and customers are doing it, plus many eons ago I too was a tradesman driving home every day in the company work van.

(Heck, many companies wouldn't even have a place to park all their company cars at once, many such smaller companies run out of regular residential buildings with no dedicated parking.)

arghwhat··on Accepting US car standards would risk European lives
If you really want an open bed, the pickup configuration of any fiat ducato, toyota dyna, mercedes vito or sprinter, etc., will work and have much more space. All 3 sides fold down, and you can even get power tilt or a small crane if you want. The dyna is like a scaled up kei-car.

There's plenty of variation as they're all custom, and as they are work vehicles there should be plenty of cheap used ones on the market. The bed is also just a plate bolted to a steel frame so you can do whatever you want with it easily - adding custom boxes underneath, built-in ramps, changing the floor, whatever. They're also available with tall roofs with openable soft cover.

But as others suggest, used closed vans are also cheap and quite spacious, and on the big end you have the usual choice of a long-body sprinter which could probably fit 3 motorcycles inside with space to spare, with a much lower ramp height needed to get them in/out. Look around - it might not be as sexy, but there's definitely something that fits your need.

arghwhat··on Accepting US car standards would risk European lives
The Dodge RAM that was being discussed is according to the numbers i see 2020 mm wide without mirrors. The F150 SuperCab (representing the most common US pickup truck) is 2030 mm wide without mirrors. The F-450 Super Duty SRW is only 2032 wide without mirrors - it's just the DRW configuration that adds extra wheels and super wide wheel wells on the back.

A standard mercedes sprinter in van configuration is 2020 mm wide without mirrors, which is as wide as the RAM and just 10mm narrower than the f150. I suspect the sprinter has wider mirrors, but I don't have the F-150 numbers to compare to so I'll leave that unanswered. Pickup configurations of the sprinter go much wider (and have extended mirrors to fit) - a common compact pickup bed configuration has a 2030mm internal bed width for example.

Note that the F-450 Super Duty is not applicable to the discussion as it won't work in the EU: A standard vehicle (class B) has an upper weight limit of 3500 kg. The F-450 Super Duty would have to be registered as an actual truck (class C), which requires a different drivers license and the use of a tachograph to track all driving and adherence to resting period rules. We don't use those vehicle classes unless strictly necessary.

arghwhat··on Accepting US car standards would risk European lives
Absolutely true but beside the current point of whether they are more or less in the way when parked in residential areas than our normal commercial vehicles.
arghwhat··on Accepting US car standards would risk European lives
I do empathize with those picking the vehicle not on practicality but cool factor - considering how common and accepted gadget cravings are in other areas, I would find it unfair to attack that aspect. I'm currently using ~5GB out of my laptops 64GB of RAM, pretty sure I could start a small fire with my flashlight, and my motorcycle has off-road suspension in a country where the most demanding obstacle is a curb. Other things would objectively fit my needs better while costing less, but be less fun - and fun can be hard to find these days.

As you say, they are absolutely terrible for work use as well - Japanese kei trucks famously have larger beds than some common US pickup trucks, and the size of the custom beds we use in the EU makes the US ones look like absolute kids toys - but that too I wouldn't mind too much if they were just forced to be safe and with decent emissions so the idiocy mainly affected the driver and their wallets.

I'm not too impressed with your vehicle only getting twice the gas mileage though. I'd expect more than that. :P

arghwhat··on Accepting US car standards would risk European lives
I can't speak for where you live of course, but they park in pedestrian areas where I live.

It's the norm in many businesses for employees to drive their work vehicle home and park it where they live, so they're everywhere. Not as many as regular passenger cars of course, but you'll see them on any residential road. Gardeners, plumbers, electricians, delivery services, this is the norm for all of them (a perk of sorts). Even big name-brand logistics companies, as it's common for the drivers to be independent contractors owning the van themselves so home is the only place to park.

They are also used for errands. They're legal for private use proportionate to the amount of VAT paid irrespective of registration type here, so you'll see them pick up/drop off kids, do groceries, recycle bottles, etc. in such vehicles too. Pretty sure that would be just as legal where you are given familiar EU rules.

arghwhat··on Accepting US car standards would risk European lives
Our regular local European vehicles are often larger, they're just safer. So no, nothing specific to the use of imported vehicles.

For example, a Mercedes Sprinter in the standard long box configuration (as is used by local grocery delivery services, plumbers and the likes where I live) is 7.4 meters long , way longer than even the longest American pickup trucks (for some of them, several meters longer!), and is just as wide as them.

In custom box or pickup bed configuration (used by e.g., gardeners), these vehicles get wider (and sharper).

arghwhat··on Accepting US car standards would risk European lives
To be fair, parking illegally and/or disrespectfully is not a problem with the vehicle type but with the driver and lack of local enforcement. People also block footpaths, roads and parking spots in Polos and similar smaller vehicles, and plenty of workers cause issues with their regular european cans and pickup trucks. A favorite of mine being small roads with perpendicular parking spots, with an extended Mercedes Sprinter parked so that both footpath and road is restricted.
arghwhat··on Accepting US car standards would risk European lives
Modern US trucks are an absolute atrocity. I am the demographic that thinks they look cool and might one day have bought one should I end up with more money than I knew what to do with if I hadn't learned that they're death traps.

The tall grill means impact to pedestrians, bicyclists and motorcycles is basically instant death as their head - the only thing above the grill - gets whiplashed onto the rigid tip of the hood. On a normal vehicle you get your legs swiped and rotate your whole body onto an intentionally flexible area of the hood for a much gentler impact.

The visibility from the driver seat is not only much worse than our actual semis, but also worse than actual tanks. You could have half a kindergarten and a small vehicle in front of your car without knowing.

As for the tax, eh - tbf these vehicles are mostly used for business purposes by sole proprietors and the likes, and while they're stupid vehicles they do still do the job. A fully decked Iveco Daily or Mercedes Sprinter is also expensive with little registration tax. Registration tax is a weird (and arguably stupid) system, this isn't really an outlier in that regard.

I roll my eyes more when I see a sports car attempted registered as a van.

arghwhat··on Isn't WSL2 just a VM?
Contributing to display servers and following kernel KMS/drm stuff, but #wayland on OFTC or #sway on libera chat are both very helpful.

Also https://wayland.app to see the current Wayland protocols (the Wayland core protocol is mainly some common primitives, most stuff is across the other protocols). For example, the sync object stuff is in https://wayland.app/protocols/linux-drm-syncobj-v1 (in many cases handled by your toolkit or WSI of choice).

arghwhat··on Isn't WSL2 just a VM?
> For example, a Linux distro running on WSL can (and will) use GPU partitioning (aka PCI/GPU passthrough) and a special implementation of DirectX

That is still just a normal VM, but it's nice that it's automated.

> enabling the installed video card to accelerate graphics within X and/or Wayland.

nit: X and/or Wayland is not involved in application rendering at all - its applications themselves that use the GPU and its acceleration directly.

Wayland and/or X is only involved when the apps are all done rendering[0], and the display servers own rendering is the comparatively simple task of stitching windows together[1], and sometimes not even that.

0: You can send buffers early over Wayland if you also send a sync fence, but this is just forwarded as a render dependency that the GPU scheduler will patiently wait for.

1: well also dealing with stuff like color transforms which can be complex to understand, but are computationally cheap and for fullscreen content possibly entirely free.

arghwhat··on Claude Opus 4.5
Sure you do, the architecture is known. An LLM will never be appropriate to use for exact input transforms and will never be able to guarantee accurate results - the input pipeline yields abstract ideas as text embedding vectors, not a stream of bytes - but just like a human it might have the skill to limp through the task with some accuracy.

While your base64 attempts likely went well, that it "could consistently encode and decode even fairly long base64 sequences" is just an anecdoate. I had the same model freak out in an empty chat, transcribing the word "hi" to a full YouTube "remember to like and subscribe" epilogue - precision and determinism are the parameters you give up when making such a thing.

(It is around this time that the models learnt to use tools autonomously in a response, such as running small code snippets which would solve the problem perfectly well, but even now it is much more consistent to tell it to do that, and for very long outputs the likelihood that it'll be able to recite the result correctly drops.)

arghwhat··on Claude Opus 4.5
I could also decode it by hand, but doing so is stupid and will be unreliable. Same with an LLM - the network is not geared for precision.
arghwhat··on Claude Opus 4.5
I disagree, the answer you get is dictated by the question you ask. Ask stupid, get stupid. Present the problem better, get a better answer. These tools are trained to be highly compliant, so you get what you ask.

Same happens with regular people - a smart person doing something stupid because they weren't overly critical and judgingof your request - and these tools have much more limited thinking/reasoning than a normal person would have, even if they seem to have a lot more "knowledge".

arghwhat··on Claude Opus 4.5
If you're asking an LLM to compute something "off the top of its head", you're using it wrong. Ask it to write the code to perform the computation and it'll do better.

Same with asking a person to solve something in their head vs. giving them an editor and a random python interpreter, or whatever it is normal people use to solve problems.

arghwhat··on Build desktop applications using Go and Web Technologies
> When we profiled it, we found that the GC essentially gave up under load

Hmm, the Go GC is really quite capable, so I wonder what kind of pathological load it was being presented with. Even then, when the GC "fails" it means elevated CPU load from collection.

The main thing I can think of would be the application "leaking" by having unintentional references (or worse, actually leaking through cgo bindings), or trashing the allocator to cause temporary spikes in between cleanups.

However, while I don't think Go was actually to blame here, I would never use native UI bindings to a language that isn't 1:1 compatible with the original design and memory management principles, as such bindings get disproportionaly large and complex. It just sets you up for a bad time.

arghwhat··on Android and iPhone users can now share files, starting with the Pixel 10
It requires WiFi active monitor mode, which is a standard chipset feature. Nothing related to custom silicon, secure enclave, hardware acceleration or other such shenanigans being brought up in the current conversation, and nothing that most android phones wouldn't fully support.
arghwhat··on Android and iPhone users can now share files, starting with the Pixel 10
No, not at all. Someone even implemented AirDrop in Python before[1]. In fact, nothing ever needs such special hardware. It's a decision of the implementer if they'd like to get fancy and rely on such hardware in their implementation to change its security profile, but the iPhone at the other end or any Apple infrastructure would be none the wiser - they just see that they're getting appropriately signed or encrypted, and neither knows nor cares how that came to be. Use of a hardware security module would just make the process more tamper resistant but would not otherwise change the outcome.

1. https://github.com/seemoo-lab/opendrop

arghwhat··on Android and iPhone users can now share files, starting with the Pixel 10
That hardware is completely unrelated to such a simple feature. Something like AirDrop will only use fairly trivial crypto, which most likely ciphers with full acceleration available but even without it would work fine with plenty of performance headroom.

Neither Apple nor Google is doing anything revolutionary with their silicon for such a standard compute task. It's really mostly minor tuning to get a more optimal part instead of an off-the-shelf chip catering to other uses too, with die area and power consumption "wasted" in your setup.

arghwhat··on Nearly all UK drivers say headlights are too bright
I think the "horizontal cutoff style" you're referring to is the American DOT beam pattern (two horizontal lines at different height) as opposed to the European ECE beam pattern (which goes up towards the edge of the road to illuminate signs and such). I would assume this follows the normal regional model variations, although Tesla seems to have notoriously misaligned headlights - some suggest it's a bug causing it to reset occasionally?

The screens are a good point, but nothing about LED headlights require a single bright point. That's done for style and cost saving - old reflector headlights had to be big, so small looks "modern", and small means less material.

Cramming it all into one tiny spot just means cooking the LEDs, which are much better suited for either larger lens assemblies or multiple smaller lens assemblies to distribute the load, both of which increase the size of the light source and massively decrease the blinding and glare it causes. You could easily cut the glare to, idk, a quarter by just changing the geometry a bit while maintaining the same light output.

arghwhat··on Pebble, Rebble, and a path forward
The board falsely accused someone of a crime. The board members must take responsibility of that decision, which involves their interactions being evidence in claims of said crime. That responsibility is part of what it means to lead something.

It's a little gross it had come to that, but ugh. Sure rebble did good for the community, but that past tense is important. Now they're trying to do bad, and that cannot be justified. Accusations require a defense, so here we are.

The Pebble community is not and never was Rebble. I briefly used Cobble as an open source project, and used their app store mirror that they decided to host to download the same old apps and watch faces I used while Pebble lived, but I was not using a Rebble watch. The open source apps I want were never Rebble apps. No one expected Rebble to be anything than infrastructure life-support, but as no one thought the dream of new life in Pebble was never going to happen, it was the light we swarmed.

Now the life support task has ended with an appreciation for their efforts, and Rebble starts acting like a company running a smearing campaign trying and make up IP ownership and justification for royalties.

Rebble no longer represents any of the community, just a misguided and greedy board. What the community wanted now exists, so that is where we have gone.

arghwhat··on GPT-5.1: A smarter, more conversational ChatGPT
They are only the new Google when they are told to process and summarize web searches. When using trained knowledge they're about as reliable as a smart but stubborn uncle.

Pretty much only search-specific modes (perplexity, deep research toggles) do that right now...

← PreviousPage 5 of 34Next →