zero_bias

It used to be shared, but vast amount of insidious attacks like Heartbleed forced web engine developers to force strict isolation between everything

Modern web development is more like computer alchemy

Thanks for clarifying, then the suggestion above does makes sense

M series chips are system-on-chip with RAM on the same wafer with CPU and GPU, so it impossible to outsource only part of the chip

Checked with vast Amon Tobin discography with all his aliases as his discography is a very complicated edge case for any recommendation algorithm: Stone Giants/Figueroa is listened by people who like Amon Tobin but the former is pure indie rock and folk and should not show Amon Tobin itself but more something like Current93. Don’t know how to solve this issue but it but it break any recommendation algorithm for last 20 years

But very nice ui, looks smooth and useful

That judgment is an essential skill of an experienced programmer, and it is required at every level of the big picture, from high level architecture decisions to the development of particular features: what should I polish and what needs to be developed fast? How exactly should I cut corners in the safest way?

So there are still only two camps

No, why it should prevent me from describing vast issues with current solutions?

So lastfm become relevant again because slop will not appear statistically in user scrobblings because of vast amount of "musicians" required to be profitable: if I listen to 1000 AI artists with one track produced and Linkin Park then my average will be Linkin Park

Cannot call lastfm algorithm advanced in any sense. Just opened Amon Tobin page: "similar artists: Kid Koala and DJ Kush", which is an impressively shallow understanding of the last 20 (!!) years of his life, and this happened with almost every artist on the platform, because the average sum of tastes of every listener does not exist in reality. E.g. in the case of Amon Tobin, Kid Koala is the average of similarities between early albums and recent releases, which is just not true, his music cannot be averaged throughout his career. I love my Web 2.0 youth, but the average similarity algorithm doesnt deserve praise. Its not better, its nostalgia and lack of faang-style unlimited greed which confused with better quality

Edit: of course spotify-style recommendations are much much worse, I just mean that lastfm doesnt have good algorithm either because artists are not consistent in releases. What is an average between electronic cult classic "The last resort" and every other Trentemoller album in strict indie rock style? This average does not exist

Actually it’s pretty explicitly stated, polymarket even have special docs section, "market maker guide"

That’s the question, yes.

The main issue is an ability to rebuild literally any part of the system from sources. A few changes here and there allow cheaters to bypass anticheat protection in a significant amount of ways

This won’t last for long too. Valve with SteamDeck and apparent future release of SteamMachine/SteamOS is preparing users for Linux gaming

First hit is always free

I’m author of relatively popular open source project (4.8k stars, 100k+ downloads/months), lived on donations for five years. I use and am eternally grateful for the following oss plans:

- Unlimited browserstack. This would cost thousands of dollars

- Free netlify hosting. Server side analytics is still $9/m, but anyway

These plans have one thing in common: they are not limited in time. Open source cannot be built on an unstable foundation.

The six-month anthropic offer is just ridiculous. Bland PR move, I can’t express how miserable this plan is. It just not for us

ISS and MIR combined are not a "large market". How many radiators they require? Probably a single space dc will demand a whole orders of magnitude more cooling

At least 50 cent didn’t make TikTok style music with single polished part and junky rest of the song. Nowadays his goals looks almost innocent

In rust you could use multiple allocators at the same time. Allocation failure handled by allocator, converting panic to some useful behavior. This logic is observable in WASM, as there are OOMs all the time, which handled transparently to application code

So I assume there is no real blockers as people in this tread assume, this is just not a conventional behavior, ad hoc, so we need to wait and well defined stable OOM handlers will appear

Why not? It’s a common equipment and it’s not count as "digital device forbidden in analog studio" as you connect synth directly to it, just to make sure that your front waves are in sync

It’s important for electronic music to have consistent and predictable pitch, otherwise djs on stage will have hard time to play (they loop a start of the song and play it together with tail loop of previous song), so Daft Punk need to intentionally choose fractional BPM as mastering engineers will not change pitch even slightly

Usually top electronic studios use external clock sync device which prevents that kind of issues, I’m sure that Daft Punk uses it too

It’s called WML/WAP

Instead of offloading batch computations to a proprietary cloud, it’s better to actually optimize the incredibly slow and unstable computational kernel.

In any case that’s not the happy path, Mathematica gets stuck in symbolic computations for ages. My FFT-based research in Mathematica slowed to a crawl, tens of minutes of waiting, even with 90% of the code compiled to binary. MATLAB finishes this task in milliseconds.

Universe could be probability based GoL simulation; basic Turing machine cannot handle that

Is this your pull request?

Unfortunately, this characterizes the entire project: "cool" examples with no practical utility. Meanwhile, the language itself is incredibly strange (functions via patterns are an example of strange language choice), extremely slow, and very unstable.

In short, it's developing in the wrong direction.

I switched from Mathematica to Matlab in my work; it was the best investment of time in the entire project

No, M series is a system on chip (SoC), that’s why it’s able to run local LLM models in a range impossible for other laptop brands: VRAM == RAM, unified shared memory at max speed for both CPU and GPU

I run qwen models on MBA M4 16 Gb and MBP M2 Max 32 Gb, MBA is able to handle models in accordance with its vram memory capacity (with external cooling), e.g. qwen3 embedding 8B (not 1B!) but inference is 4x-6x times slower than on mbp. I suspect weaker SoC

Anyway, Apple SoC in M series is a huge leverage thanks to shared memory: VRAM size == RAM size so if you buy M chip with 128+ Gb memory, you’re pretty much able to run SOTA models locally, and price is significantly lower than AI GPU cards

Safari on iOS support uBlock origin too

Python is ubiquitous in ML, often you have no choice but to use it

When primordial black holes formed, there was no matter that could clump around them, as the matter at that time had a very high temperature