sshine

19 ms·

331.
by sshine 1y ago
I’ve selected for interesting work primarily by working for R&D departments, for sketchy startups, and in niche domains. The cost has been that I don’t have deep domain knowledge that could lead to making a startup myself. And it’s the
332.
by sshine 1y ago
Looking at the careers of mathematicians who turned software developers after putting minimal effort into CS courses, they all make CRUD work for insurance sector consulting companies. The pay is fine, but the work is absolutely dreadfully
333.
by sshine 1y ago
It is incredibly difficult to assess the quality of projects in depth. I tried doing this for Open-VSX extensions for handling justfile: https://open-vsx.org/?search=justfile&sortBy=relevance&sortO... 2.4k nefro
334.
by sshine 1y ago
> this “next generation search” doesn’t cite sources The training done on the content does not provide citable references with current models. The agentic search and summary done post-training does. A lot of the heavy traffic is for tr
335.
by sshine 1y ago
> Deep Research from OpenAI, is way better in doing product recommendations than Google Interesting, that’s not my experience and I’d be the first to replace Google if I could. I’ll have to try again.
336.
by sshine 1y ago
> If there was a paid-only search engine with dubious ethics practices that was overwhelming my site with traffic in order resell search trained off of (among other things) my personally generated content, I would absolute block it. Le
337.
by sshine 1y ago
This worked for me until iCloud started cache clearing all my files aggressively so my vault would take ten minutes to open on iPhone. Every few days. When I tried to copy my vault off iCloud, the copy failed and two years of notes were per
338.
by sshine 1y ago
Well, thanks for the flashback. One can get too obsessed about performance. Productivity with AI-generated Rust should be on par with anything since the verbosity is only measured in tok/s.
339.
by sshine 1y ago
> As for Rust, I don't have a use case at work where it is a win over managed compiled languages Very performant, very easy to distribute cross-compiled static binaries, very portable to web, embedded, anywhere. The main downside
340.
by sshine 1y ago
I've vibe-coded TypeScript and Rust with great success. I've vibe-coded Haskell to great disappointment. I just can't accept how verbose and ugly it gets. Which, for some reason, I'm totally okay with when coding Rust an
341.
by sshine 1y ago
I've been vibe-coding for weeks in Rust, and it works great. I've been vibe-coding for a few days in Haskell, and I don't like the result. Maybe I am just accustomed to being ok with verbose Rust, while Haskell comes with a g
342.
by sshine 1y ago
> nobody opens claude and thinks, "you know what? let me use the shitty version" Sure I do! I will consistently pick the fastest and cheapest model that will do the job. Sonnet > Opus when coding Haiku > Sonnet when fus
343.
by sshine 1y ago
I should have said my ($x, $y, $z) = @$_[0]; but I was making a joke. If $_[0] is an array reference, @$_[0] becomes a list. The joke was a reference to most wrong Perl programs being valid for something. Hence the meme: "93% of
344.
by sshine 1y ago
I wonder how widespread YAML 1.1 is. If you assume that YAML 1.2 is the default, you don't need that nasty %YAML header. This doesn't translate to arbitrary, open environments, but you can make that choice in closed environments.
345.
by sshine 1y ago
Yeah, you probably want my ($x, $y, $z) = @shift; :-D
346.
by sshine 1y ago
If he were buying liquor at a gas station, I wonder how many employees would mistake him for a child and insist on age verifying him. Clearly drawing this much on your face is the act of a child.
347.
by sshine 1y ago
VPS’es aren’t great for running VPNs from because their IP addresses are so obviously not residential.
348.
by sshine 1y ago
It seems like people who use AI for coding need to reinvent a lot of the same basic principles of software engineering before they gradually propagate into the mainstream agentic frameworks. Coding agents come with a lot of good behavior bu
349.
by sshine 1y ago
It should be. But it’s not incorrect to say The tree grown tall is the one I planted The horse raced past had somewhere to go But when you don’t end the subject syntagma in an obvious way, you make it difficult to read. It’s easie
350.
by sshine 1y ago
Time flies like an arrow, fruit flies like a banana.
351.
by sshine 1y ago
It’s a spectrum. I care when it doesn’t just work. I hardly look when it does.
352.
by sshine 1y ago
The LLM + agentic framework supplements many core dev skills including sporadic habits of tirelessly testing things, spec-driven development (thinking what you’re gonna make before you do), and debugging. Some devs may have all of these, so
353.
by sshine 1y ago
I’ve been coding a git-powered issue tracker (like git-bug) that’s been on my wishlist for years. With agentic programming, I need issue tracking that is accessible from CLI, and in three evenings I’ve made what would have taken two weeks.
354.
by sshine 1y ago
Because support requires understanding. If you can maintain long-term context for the LLM, then you have an approximation of actionable institutional knowledge. But context is text documentation, and that stuff rots just as much, if not mor
355.
by sshine 1y ago
The founder of IKEA, Ingvar Kamprad, passed away in 2018. IKEA is currently owned by a series of foundations. On account of Ingvar Kamprad being dead, they're not really in the same class. Before Ingvar Kamprad passed away, his estimat
356.
by sshine 1y ago
> widening of the criteria used to diagnose Seeing some autism influencers on social media, I'm reminded that every childhood friend I ever had was autistic.
357.
by sshine 1y ago
So what you're saying is vaccines might cause autism? ;-)
358.
by sshine 1y ago
Yes, we get to stay perpetually guilty for never even getting close to any “extreme” (actual) boycott of all technology.
359.
by sshine 1y ago
It was an interesting experiment working with very little clue of the generated code. I could learn about react and understand the large-scale incongruences / mismatching choices the LLM made for me. But I already have one reactive fra
360.
by sshine 1y ago
I worked with low-latency trading with C# in the hot path; they had started with C++ and eventually migrated to C# and just kept tight control of allocations. Later they’d go back to using C++ for training deep neural nets.

More ›