@kornel@mastodon.social
@kornel@mastodon.social avatar

kornel

@kornel@mastodon.social

I compress pixels (https://ImageOptim.com https://pngquant.org https://gif.ski).
Web Standards & #RustLang

This profile is from a federated server and may be incomplete. Browse more on the original instance.

kornel, to random
@kornel@mastodon.social avatar

Chrome is shipping page transitions that work with regular non-JS-overloaded web pages.

Now I need to figure out how to use this without going full-PowerPoint.

https://developer.chrome.com/docs/web-platform/view-transitions#cross-document_view_transitions

kornel, to passkeys
@kornel@mastodon.social avatar

Authentication with U2F keys (AKA Fido or Yubikey) works well for me. It's phishing-proof and as easy as tapping a button. I'm not in a rush to try #Passkeys.

joelanman, to random
@joelanman@hachyderm.io avatar

A lot of people are finding LLMs more useful than Google search.

I think a huge part of that is that Google and web in general has massively deteriorated in terms of finding clear, concise answers. Even if you can find it, it's covered in ads, cookie popups and so on.

But LLMs will inevitably follow the same path - investors want their returns. And the complete non transparency of LLMs will make it even worse this time around. Is that really the best answer or is it sponsored?

kornel,
@kornel@mastodon.social avatar

@joelanman Ability to ask follow up questions is very valuable, and isn't served well by search.

kornel,
@kornel@mastodon.social avatar

@benjamineskola @joelanman

  1. Pour billions into LLMs
  2. ???
  3. Profit!

The algorithms are mostly public, and the training gets exponentially cheaper, so it's going to be a commodity. I don't expect "AI" companies to get their money back on LLMs.

kornel,
@kornel@mastodon.social avatar

@joelanman @benjamineskola it's genuinely useful that LLMs are good enough to clean up their own training data. Somehow we've solved garbage in - garbage out!

kornel,
@kornel@mastodon.social avatar

@benjamineskola Such cynical view overlooks the hard problems that LLMs have solved.

e.g. W3C dreamt about the Semantic Web for decades, and suddenly we have ability to convert human-facing information into machine-readable form, instead of unsuccessfully trying to implement it from the other direction.

kornel,
@kornel@mastodon.social avatar

@joelanman @benjamineskola The ML solutions are inherently imperfect, but their errors need to be weighed against issues you would have with other approaches.

For example, invisible metadata turned out to be systematically unreliable, because it gets less testing, so is likely to become stale or bitrot. Invisibility helps using it for spam.

So apart of rare cases of honest, correct, complete up-to-date semantic data, there are no less flawed ways of extracting information from the web.

kornel,
@kornel@mastodon.social avatar

@benjamineskola This pattern of hype, and people not being ready, is quite common in technologies, good and bad ones.
The Web has created enormous value, but also brought phishing and dangerous disinformation that many people are unprepared for. Smartphones have connected billions of people, but also let Apple make a fortune on facilitating payments to gambling apps for toddlers.

And don't worry, in "AI" everyone is hemorrhaging money, except NVidia selling them shovels ;)

kornel,
@kornel@mastodon.social avatar

@joelanman Programmatically, I've had success with Ollama+Mixtral. It is easy to run locally given enough VRAM, and worked great for categorization tasks, anti spam, and search indexing – it understands any language, jargon & slang, and picks up even subtle context-dependent clues that a bayes classifier never would.
Being able to run pretty advanced models locally makes me optimistic that it can be used as merely a tool, rather than being yet another VC-funded surveillance trap.

thisismissem, to random
@thisismissem@hachyderm.io avatar

So apparently MacOS 14 isn't available for devices from 2017 — so I'd need to buy a new mac, even though mine works perfectly fine, to use this app (MacOS 13 is still well supported by Apple)

Apparently Apple considers devices “vintage" when they're between 5 and 7 years old; so hardware that works perfectly well from 2017 becomes un-upgradable in 2023/24

I can't see any information on how long Apple will continue to support MacOS Ventura (13)
https://mastodon.social/@sindresorhus/112432845781407658

kornel,
@kornel@mastodon.social avatar

@artemist @thisismissem I suspect the rushed obsolescence is because Apple wants to drop all x86 CPU support and non-iPad-ified bootloaders ASAP.

joelanman, to random
@joelanman@hachyderm.io avatar

I have a database question - say you have an items table and a messages table, and messages can refer to items. If someone needs to delete an item, you can't because of the foreign reference in messages.

I know you could 'soft delete' - set a status column to deleted, but what if you really need to delete, for example because the content is illegal or the user has a legal right to delete it?

kornel,
@kornel@mastodon.social avatar

@joelanman you can change owner of the messages to a null or dummy user.

mcc, to random
@mcc@mastodon.social avatar

I wish Cargo (Rust) would let you mark a Feature as "internal" and then not expose it directly or mark it different or something when listing enabled Features in the CLI. This program has two Features, "speed-debug-visual" and "speed-debug-serial". They both rely on a third Feature, "speed-debug-internal", but speed-debug-internal does nothing but provide support guts for the other two Features and there's no reason to turn it on by itself, it's just there to be a dependency for the other two

kornel,
@kornel@mastodon.social avatar

@mcc There's informal convention that __-prefixed features are internal. Docs.rs and Lib.rs won't show them.

kornel,
@kornel@mastodon.social avatar

@mcc Single underscore works too, although crates generally use two underscores.

It's not documented.

https://github.com/rust-lang/docs.rs/blob/37fa77264f3b08dff432c082cbd412361d89245c/src/db/types.rs#L17

Not in Cargo yet:

https://github.com/rust-lang/cargo/issues/10794

ekuber, to rust
@ekuber@hachyderm.io avatar

Request for feedback: how would you change this compiler error? Can you tell what's going on? What the problem is? Do you get a sense of how you might be able to solve it?

kornel,
@kornel@mastodon.social avatar

@ekuber Not having this problem would have been even better, so this is more of an issue with Cargo's UI not complaining about duplicates loud enough (that's an ecosystem-wide problem requiring users to complain), and Cargo's resolver not trying hard enough to deduplicate them.

kornel,
@kornel@mastodon.social avatar

@ekuber I'm not saying it should be disallowed. I'm saying it has bad UI. People get into such situation by accident, are unaware when it happens, and a simple cargo update can add dupes to a no-dupes project without informing the user, and without straightforward ability to undo that.

kornel, to random
@kornel@mastodon.social avatar

Pro has amazing hardware, but I can't get over the OS that feels like working with one-finger gloves on.

kornel, to random
@kornel@mastodon.social avatar

When I want my mind blown, I compare former world's largest supercomputers to the tech I have around.

ASCI Red, the world's fastest supercomputer until the year 2000, was 150m² large, consumed 850,000W of power …and was slower than an iPhone 14 Pro, which uses ~6W.

ASCI Red had over 1TB of RAM, which may still sound impressive, except that RAM was 2× slower than the iPhone's 1TB SSD.

https://upload.wikimedia.org/wikipedia/commons/9/97/Asci_red_-_tflop4m.jpeg

kornel,
@kornel@mastodon.social avatar

@happyborg It's funny that software bloat complaint is as old as software, only the numbers change.

https://cr.yp.to/bib/1995/wirth.pdf

I can't wait until every program bundles a 64GB LLM for its AI interface, making Electron apps feel tiny ;)

kornel, to random
@kornel@mastodon.social avatar

It's surreal how slowly time moves in the world of C compilers.

Today there are still active projects that are hesitant to move past C89, and C99 is still the "new" standard.

The C99 standard has been released before the first public Mac OS X and Windows XP. It's older Itanium and the x86-64 instruction set. It predates iPod, Game Cube, first ever Xbox, and Nokia 3310.

Entire platforms lived and died in the meantime, while C programmers still can't be sure if they can rely on the new C99.

kornel,
@kornel@mastodon.social avatar

@cdamian @skolima There's a difference between stability and stagnation. The downside for C is that it's impossible to fix anything in the language, because it will take 20+ years to become the baseline. I suspect C++ mis-evolution has scared C users into don't change anything ever!.
Both C and C++ are outliers. It's possible to be stable without stagnation: Java and C# evolved without jumping the shark. PHP cleaned up a ton of language warts. Go can ship quickly without breaking things.

kornel,
@kornel@mastodon.social avatar

@timthelion I can’t take your “objectively”intensified outcry seriously.
Java has added terser forms for some previous noisy boilerplate. The old syntax still works. It’s never been a terse language, and it has been conservative in its evolution, so “unreadable” is a gross exaggeration, especially that readability is inherently very subjective and dependent on experience.

heydon, to random
@heydon@front-end.social avatar

Currently apoplectic that the Speech Recognition API hears “fuck” and outputs ['f***']. You puritanical Christian dickheads are at it again aren't you?

kornel,
@kornel@mastodon.social avatar

@heydon I’ve had Android dictation transcribe “fax machine” as “f*** machine”.

ekuber, to rust
@ekuber@hachyderm.io avatar

Rust's unexpected super-power is just how flexible it is. It allows you to write very high level looking code on a low level language. That caused people to use it beyond its intended niche. But it is fundamentally a low level programming language. It will continue becoming easier to use (that's my personal goal!) but there are "obvious" changes that would make things easier at the cost of speed or correctness that cannot take.

kornel,
@kornel@mastodon.social avatar

@ekuber It would be interesting to have a "do-whatever-you-want-Rust" for prototyping, that has a GC/ARC, allows global state, aliased mut and other "dirty" things, but with support for gradual typing that tightens the requirements, so that you could gradually refactor code into Rust proper where it needs more performance or bugfixing.

kornel, to random
@kornel@mastodon.social avatar

The https://js-naked-day.org is easy for me: https://lib.rs has no JavaScript. It never had any.

annika, to random
@annika@xoxo.zone avatar

PHP tip: you can write 0 (zero) as 0_0 for added emotion.

if ($balance == 0_0) {  
 panic();  
}
kornel,
@kornel@mastodon.social avatar

@annika With varying screaming intensity:

0.0 | 0o0 | 0O0

  • All
  • Subscribed
  • Moderated
  • Favorites
  • •
  • anitta
  • kavyap
  • DreamBathrooms
  • ethstaker
  • magazineikmin
  • InstantRegret
  • modclub
  • Youngstown
  • everett
  • slotface
  • rosin
  • khanakhh
  • ngwrru68w68
  • PowerRangers
  • provamag3
  • thenastyranch
  • Durango
  • cubers
  • normalnudes
  • vwfavf
  • mdbf
  • GTA5RPClips
  • tacticalgear
  • tester
  • Leos
  • osvaldo12
  • cisconetworking
  • megavids
  • All magazines
    •