apatheticonion

6 ms·

1.
by apatheticonion 6d ago
Performance isn't really the selling point for me. It's the fact that runtime errors are essentially impossible, making writing, reviewing, testing code very easy. It also means LLMs can spin it out with the compiler, and if it co
2.
by apatheticonion 6d ago
I do have fond feelings towards winforms and wpf from years ago, but I can't use it to write cross platform applications. There really isn't anything other than web that covers that use case, sadly. C# still has runtime exceptions
3.
by apatheticonion 6d ago
From a software design standpoint, it certainly takes some adjustment going from OOP to the compositional architecture and structural trait system used by Rust, but it's not that big a shift. The biggest downsides are the poor standard
4.
by apatheticonion 6d ago
I know, haha. It's a bit of an absolute statement but as far as all of the features I look for in the context of creating scalable and maintainable software (especially in the age of agent-assisted coding) - Rust has been the most prod
5.
by apatheticonion 6d ago
Yes, that contributes to the "if it actually worked" sentiment. It doesn't have to be slow, if the browser exposed C-like ABI for DOM access and web APIs like LocalStorage, the FileSystem API, ServiceWorker, etc - Rust bindin
6.
by apatheticonion 6d ago
I've been writing Rust professionally for the last 5 years (where my first decade of professional experience started in frontend, then moved down the stack TypeScript/Node, Go, C# and so on). From the perspective of high level app
7.
by apatheticonion 10d ago
I'd love to. Want to join?
8.
by apatheticonion 10d ago
> The market for people who want to install Linux on a Mac laptop is very tiny. I suspect there would be an element of induced demand here. IMO the MacBook range are the only laptops on the market worth buying - but not having support fo
9.
by apatheticonion 10d ago
I'd imagine it's for IP protection. There's a lot of documentation required to write drivers for hardware; for the GPU, they'd need to share the instruction set of the GPU along with the specifications of the controller.
10.
by apatheticonion 10d ago
Right!? Linux can be configured to idle at less than 500mb of ram - by contrast, with some aggressive shenanigans, MacOS can idle at 2-3gb. When running inference, I would love to use all of my hardware please Apple.
11.
by apatheticonion 10d ago
On the topic of browsers, I feel your pain. There were a good few years where they were shipping new and useful features, but they plateaued and started removing features. I can't remember the last time I was excited about browser feat
12.
by apatheticonion 11d ago
Yes, that's what I'm saying. 200t/s is more than enough to 5 - 10x my productivity and the intelligence of current open weight models covers 90% of both my guided and software factory workflows
13.
by apatheticonion 13d ago
Too new, but my old M1 MacBook Pro runs Asahi and, while I love it and everything the team has done, it's not really daily drivable for work (due to a lack of external display support and other issues).
14.
by apatheticonion 13d ago
I just want an API that takes these crazy small / cost effective open weight models and charges peanuts for access. Think, DeepSeek Flash (before the price hikes) prices. If I can run this on a 32gb card while they have a datacenter wi
15.
by apatheticonion 15d ago
Not true. Apple, for instance, does not provide low level device documentation sufficient for creating open source drivers - for instance - publishing documentation on the Apple GPU and coprocessor would have reduced the development effort
16.
by apatheticonion 15d ago
I know it's unethical, but when I read this, I can't help but hope a pseudo-cleanroom/ai-laundered Linux GPU driver appears for MacBooks. What I wouldn't give to run Linux on my Apple hardware.
17.
by apatheticonion 15d ago
It's rarely sunshine, but there are a lot of rainbows!
18.
by apatheticonion 20d ago
Agreed. I've been "guide coding" in my editor (Zed / VSCode) for a while now and it's really enjoyable. DeepSeek v4 flash has been dirt cheap and so fast that my development loop is; - small prompt - review - small
19.
by apatheticonion 21d ago
Unfortunately they decided not to proceed despite glowing feedback in the other rounds
20.
by apatheticonion 22d ago
Nice! Linux Please. My M5 MacBook Pro, while amazing, is basically relegated to being a thin client with amazing battery life that exclusively SSH's into my desktop PC for work. I could spawn a Linux VM / use Docker/Podman lo
21.
by apatheticonion 22d ago
I am on that team and REALLY want it to be the case - but GNU/LInux devices are usually a significantly worse hardware experience than their walled garden cousins. For example, there is literally no laptop on Earth as nice as the MacBo
22.
by apatheticonion 23d ago
Even then, it tends to code itself into a corner. Code tends to be ridged and difficult to expand. One shot benchmarks that have exhaustive prompts are impressive, but benchmarks that emulate real software development (iterative changes ove
23.
by apatheticonion 23d ago
I had a job interview at Canva where they had an "AI coding round". They wanted to see how I used AI in my workflow. Because I was using DeepSeek flash, the interviewer was confused at how fast I was getting edits - he'd only
24.
by apatheticonion 23d ago
FR the iPhone is such a great missed opportunity. Imagine if someone developed a Linux based mobile OS that could run on it. Imagine if someone figured out that the USB-C port could do displayport out. Then imagine plugging that Linux iPhon
25.
by apatheticonion 23d ago
I see a large emphasis placed on headless agentic/vibe coding, what I don't see people talking about is how great guided coding is. I have over a 15 YoE writing software and guided coding sessions - that is, using an editor like Z
26.
by apatheticonion 24d ago
I would like this to be extended to software. If I buy an Android phone, I should be able to maintain/update the operating system. If I buy a MacBook, I should be able to maintain or change the operating system. If I buy a TV, smart wa
27.
by apatheticonion 25d ago
Exactly this. Using an LLM to study the decompiled output then another LLM to implement an emulator/engine/etc That sounds similar to how LLMs have been trained on GPL code and yet output is not (yet) considered derivative work
28.
by apatheticonion 26d ago
DS is dead to me after the pricing changes. Qwen 3.8 has replaced it entirely for me
29.
by apatheticonion 27d ago
Does this mean someone could develop a switch emulator using an AI and Nintendo couldn't win a lawsuit against them? Does it enable decompilation remasters of classic games? It feels like AI is a cleanroom laundromat
30.
by apatheticonion 1mo ago
I've been using DeepSeek exclusively since I've been playing with LLMs, largely due to the price. I joined a company that is an Anthropic shop and I am genuinely shocked. Sonnet 5 is a little better on long horizon tasks and headl

More ›