Hacker Newsnew | past | comments | ask | show | jobs | submit | pixelesque's commentslogin

Google argued that duplicating largely (I know JpegXL does support a bit more, but from most users' perspectives, they're largely right) what AVIF provided while being written in an unsafe language was not what they wanted in terms in increasing the attack surface.

And it really was the right move at the time, imo. JXL however now has better implementations and better momentum in the wider ecosystem and not just yet another image format that gets put into chrome and de facto becomes a standard.



Hahaha perfect! Can't believe I never heard this story before.

I can confirm. I found multiple problems in the "official" cjxl encoder back in 2023 contrary to the webp2 (cwp2) implementation where I could not find any bug or error.

If the encoder have obvious problems it is not a big deal, but it doesn't bode well for the decoder.


CVE-2023-0645 in libjxl that year too, and several since

It's also a horrible API. Will start working on the rust lib then. Hopefully it's better, because I really want to use it.

Forcing other companies to override them is a way to prove momentum but it's not a good way to prove momentum.

> duplicating largely what AVIF provided

That's not a great bar since both of them showed up around the same time. And importantly JXL hits many use cases that AVIF doesn't.

> while being written in an unsafe language

They put little emphasis on that part when they were rejecting JXL. If they wanted to call for a safer implementation they could have done that.


A couple of thoughts:

That graph is number of questions asked being posted: very often the question already exists (although obviously with technology and frameworks changing over time, things aren't constant, and answers can be out of date at some point), so you don't need to post the question.

Also: Would LLMs be as good for answers if they hadn't been trained on scraping StackOverflow in the first place?


I very much disagree with the first point that there had somehow been a critical mass of questions achieved to explain the slowdown. With the change in tech, outdated answer, and increasing population of people writing code, I just don't buy it.

Hyper-cope.

But on your 2nd question,

>Would LLMs be as good for answers [...]

Yes.


LA to Dubai is just over 16 hours flight time, so either way, he's not comparing realistic flight times anyway.

Roughly that figure (45%) was used to get to Mach 2.0 at 60,000 feet, about 45 minutes after takeoff from LHR (normally over the Bristol channel) to JFK.

Takeoff and climb / accel to Mach 1.7 was done with re-heat (afterburners), which did use a lot of fuel. After that, normal power (no re-heat) was used to get to Mach 2.0 and cruising (supercruise).


There's experimental/nightly support for things like: `push_within_capacity()` which is a more manual way (user-space code would have to handle the return code and then increase the capacity manually if required) of trying to handle that situation.

And of course the kernel - which doesn't even use Rust's Vec but has its own entire allocator library because it is the kernel - likewise provides

https://rust.docs.kernel.org/next/kernel/alloc/kvec/struct.V...

Vec::push_within_capacity is a nice API to confront the reality of running out of memory. "Clever" ideas that don't actually work are obviously ineffective once we see this API. We need to do something with this T, we can't just say "Somebody else should ensure I have room to store it" because it's too late now. Here's your T back, there was no more space.


It's very annoying, but you can turn it off (snowflake icon, although then you get a yellow background), and IMO the article content is worth reading.

I tried that then discarded it, because it doesn't immediately hide the snowflakes that are already on screen.

The article is good but the choice of distracting snowflakes or radioactive piss burning your retina is not a welcome one.


It takes over 10 seconds to turn off fully. I had to go back and try it after your comment, because I thought that button just turned the page yellow, which was worse than the blue.

I ended up using reader mode to read the page. The whole site design undermined the point being made. One of the first things mentioned is not to be distracting. Yet they went out of their way to make their own site distracting. "Do as I say, not as I do."


The snowflake icon that disappears off-screen the instant you scroll down past the obnoxiously large header image to read the actual content?

I literally didn't notice that the snowflake icon turned it off:

   1. I scrolled through the article getting more and more frustrated with the snow
   2. I scrolled all the way back to the top and saw the snowflake icon
   3. I clicked the snowflake, saw the hideous yellow, said WTF and clicked again to go back to blue
   4. **I never noticed** that the snowflake *does* stop the snow, but *only* stops *new* snow, so the existing snow continues to fall across the screen
   5. I clicked several other things, then came here to complain and saw this thread

  > although then you get a yellow background
Yes, and this is arguably worse. I ended up using Immersive Reader mode in Edge.

> SQLite has concurrent writes now

Just to clarify: Unless I've missed something, this is only with WAL mode and concurrent reads at the same time as writes, I don't think it can handle multiple concurrent writes at the same time?


As I understand it, there can be concurrent writes as long as they don't touch the same data (the same file system pages, to be exact). Also, the actual COMMIT part is still serialized and you need to begin your transactions with BEGIN CONCURRENT. If two transactions do conflict, the later one will be forced to ROLLBACK although you can still try again. It is up to the application to do this.

See also https://www.sqlite.org/src/doc/begin-concurrent/doc/begin_co...

This type of limitation is exactly why I would recommend "normal" server-based databases like Postgres or MySQL for the vast majority of web backends.


Also just a note: BEGIN CONCURRENT is not in mainline SQLite releases. You need to build your own from a branch. Not a huge deal but just something to note.

I think only Turso — SQLite rewritten in Rust — supports that.

That tabletmag.com article is a ridiculous work of fiction.

personally, the editor of tablet elevating a claim that immigrants are third world trash put me off that magazine permanently

Ted Turner once said something egregiously stupid and wrong and ever since then I discounted everything CNN did or reported.

Guessing it’s what he said in 2002? But the Tablet EIC made these remarks last week. Not really equivalent, although you’re entitled to your media choices.

I was pointing out the bad reasoning you’ve employed.

It's bad reasoning to say that I don't trust the editorial decisions of someone who cosigns that immigrants are trash?

From Wikipedia:

A referendum to dissolve parliament and give the prime minister power to make law was submitted to voters, and it passed with 99 per cent approval, 2,043,300 votes to 1300 votes against.[83] According to historian Mark Gasiorowski, "There were separate polling stations for yes and no votes, producing sharp criticism of Mosaddeq" and that the "controversial referendum...gave the CIA's precoup propaganda campaign to show up Mosaddeq as an anti-democratic dictator an easy target".[84]

A person has to be very gullible to believe 99% of the vote went one way in a fair election involving 2+ million people.


Thank you for that convincing insight.

I had a Thinkpad T480s that was absolutely perfect with Linux (Mint), although very underpowered, but that was due to Intel CPU.

This year I got a T14s Gen6 AMD as a replacement, and it's essentially unusable on Debian-based distros (Ubuntu, Mint), but works fine with Fedora and with Windows.

On Ubuntu and Mint, X just locks up every 80 seconds or so, and I have to hard-reboot it (or switch ttys and restart X). Nothing in syslog, nothing in dmesg, nothing in X.org.log to show what might be going on.


I have the T14s Gen3 AMD and everything just works with CachyOS and Fedora. In general I tend to use as recent kernel as possible with AMD. They do update their drivers a lot in every Linux version.

It is.

Some modern films are still filmed with anamorphic lenses because the director / DP like that, and so we in the VFX industry have to deal with plate footage that way, and so have to deal with non-square pixels in the software handling the images (to de-squash the image, even though the digital camera sensor pixels that recorded the image from the lens were square) in order to display correctly (i.e. so that round circular things still look round, and are not squashed).

Even to the degree that full CG element renders (i.e. rendered to EXR with a pathtracing renderer) should really use anisotropic pixel filter widths to look correct.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: