troupo

14 ms·

241.
by troupo 3mo ago
> You will never get the same quality of fundamental material If you're looking for fundamental material, the last thing you want is learning a very (not slightly) outdated and weird API. > Vulkan and DX12 are currently flawed
242.
by troupo 3mo ago
How do you think things appear in the browser if they don't use web technologies?
243.
by troupo 3mo ago
Last time I tried it (quite some time ago) it didn't have anything built-in for file uploads: https://github.com/bartblast/hologram/issues/228 I've tried to keep an eye on its development, but I don
244.
by troupo 3mo ago
The only thing that's stopping me from trying this currently is file uploads. Otherwise it's a really cool project
245.
by troupo 3mo ago
> When did it become normal to talk about open source like this? Why does something being open source absolve it of any sins?
246.
by troupo 3mo ago
TVs are good when used appropriately. We watched Macbeth for the literature class.
247.
by troupo 3mo ago
I doubt US tech understands. For one, they are deeply in bed with the current administration (for any definition of current, past, and future). For another, they fight tooth and nail any and all sovereignity of data even within their own wa
248.
by troupo 3mo ago
Many many words, no meaning. Because you keep vibe-reading what I write. Here's what I literally started with: "I always have to correct its hallucinations during the day. Why would I ever let it run unsupervised overnight?"
249.
by troupo 3mo ago
Yes. "People don't use alternatives not because they literally aren't allowed to, but because other things matter more"
250.
by troupo 3mo ago
John Oliver did a report on militarization of police 11 years ago . It's only become worse since: https://youtu.be/KUdHIatS36A?is=C86HI8463VZnqjXU
251.
by troupo 3mo ago
Because there's no legal way to use Claude Code on subscription otherwise
252.
by troupo 3mo ago
A TUI outputting nothing but text shouldn't consume 70 GB RAM and yet there they were just half a year ago: https://x.com/jarredsumner/status/2026497606575398987 They couldn't even fix flickering in thei
253.
by troupo 3mo ago
You started with "scanning logs for errors" with LLMs which is one of the most stupid things could do with LLMs on a "serious SASS product with hundreds of thousands of users" (aka logs easily in gigabytes to hundreds o
254.
by troupo 3mo ago
> The lack of long context is the main reason that I still end up using Anthropic. Anthropic's "long context" is mostly marketing bullshit. It semi-reliably holds about the same amount of info, and then starts suffering th
255.
by troupo 3mo ago
> Is the high expense coming from cache misses? Yup. They can't keep your workload in cache forever, or they would run out of cache for users. > I wonder would starting new sessions and having to re-read the contexts and results
256.
by troupo 3mo ago
> and then just wait for the next prompt A single session running for multiple hours is prohibitively expensive, as per Anthropic. Regardless of whether it just waits for a prompt or does something.
257.
by troupo 3mo ago
> I regularly have my code reviewed by professional devs > I also read my code regularly So you're literally doing what I am talking about.
258.
by troupo 3mo ago
Demagoguery. Your "great product" (where is it? show me these great products) will have wildly ineficient code and things like "read the entire DB into memory" regardless of the prompt . > plenty of people have figur
259.
by troupo 3mo ago
> You see, there's your problem right there. You're vibe coding, which by definition literally means you're unwilling to look at the generated code. That's not what successful ai assisted software developers are doing
260.
by troupo 3mo ago
> Imagine for example looking for http status code 500 in an api log over the past hour. 1. On a blog that no one visits maybe? 2. It's called a grep 3. For bigger projects it's called sentry
261.
by troupo 3mo ago
Running tests that the agent created for its own hallucinations? Optimize using another hallucinating agent? Which tests and optimizations do you propose to run after a night of supervised work when one of main things that all agents keep d
262.
by troupo 3mo ago
On all projects I've run any of the models they: - infinitely duplicate any and all code, helpers and components - infinitely duplicate CSS (because they duplicate components) - continuously write code like "read the entire db int
263.
by troupo 3mo ago
> scanning logs for errors famously a good job for a tool that takes 10-50k logs to run out of context and forget what it's doing.
264.
by troupo 3mo ago
I always have to correct its hallucinations during the day . Why would I ever let it run unsupervised overnight?
265.
by troupo 3mo ago
> take up to hours at a time to run, so Claude is « monitoring » tasks most of the tim How is Claude monitoring them for hours? Claude runs out of context and extremely long sessions are prohibitively expensive even according to Anthropi
266.
by troupo 3mo ago
Not Google. Microsoft, of all companies started this project: https://open-ui.org/ when Google was busy breaking the web web components, hardware APis etc. Google was very, very, very late to the project and of course imm
267.
by troupo 3mo ago
> Are there any other Operating Systems than iOS, Android or Android flavors? I remember a gov.uk team presentation. They had a usecase of someone using a PS Vita to access a government assistance program because that was the only device
268.
by troupo 3mo ago
> for the machine, it'd make it easy to add in new tests because the function+tests are in the same context window Given how many useless unit tests Codex just loves to write, no, thank you. Thankfully, its training data completely
269.
by troupo 3mo ago
Basically anything you see out of a window, anything in the street more than a few meters away from a camera, almost any overhead shot, any wide shot, any panorama or landscape, blood, liquids, glass, any semblance of stunt work, most signs
270.
by troupo 3mo ago
While many of the things tou say are true, movies like PHM and other "predominantly practical" movies have so many effects that they put Marvel to shame. I highly recommend the 4-part essay series "'No CGI' is just

More ›