lolinder

15 ms·

151.
by lolinder 1y ago
> If a Deno library uses a Node API or an npm library somewhere, it can be entirely encapsulated, so you might not even notice until you see it in a stack trace. That doesn’t seem very intrusive? Node API yes, NPM library no. If you add
152.
by lolinder 1y ago
An equivalent argument to yours could be made to defend the introduction of async/await to a language that has previously not had it (edit: like Rust): if you don't like async/await, just don't use it! What does it hurt
153.
by lolinder 1y ago
It misses the frontend/backend symmetry and has too large a coupling to Microsoft and Windows in my head. I know that these days it's supposed to be cross platform, but every time I've tried to figure out how to install it I
154.
by lolinder 1y ago
For me Go doesn't even come close. It's way too restrictive in what I can and can't do with it. I might feel differently if I worked with a large number of people who I didn't trust, but on small to medium teams composed
155.
by lolinder 1y ago
Yep. Honestly, the pivot to Node/NPM compatibility was the moment I lost interest in Deno. I know why they did it, and as you say from a financial perspective it makes complete sense, but they had the chance to be a fresh start to the
156.
by lolinder 1y ago
Because out of all the languages that stand a chance of being adopted at most workplaces, TypeScript is in my opinion the single most enjoyable to code in. It has an extremely expressive type system that gets out of your way and allows you
157.
by lolinder 1y ago
In the same spirit if "it depends", there are other options that may work for people with different Linux/Windows balance points: * Wine is surprisingly good these days for a lot of software. If you only have an app or two th
158.
by lolinder 1y ago
In some cases, sure, but not here—neither option had more risk associated with it than the other, it was just an optimization problem. The first answer that the model gave to my wife was just wrong about the math, with no room for subjectiv
159.
by lolinder 1y ago
Yeah, I first thought of Hedy too. It doesn't keep them within the ecosystem but it does address the gradual intro problem and the localization problem.
160.
by lolinder 1y ago
That probably would work most of the time, but that's also an example of the phenomenon that TFA is talking about: you can't safely just use these tools without becoming an expert at least in the tool. The way they're curre
161.
by lolinder 1y ago
Given the topic I'm unfortunately not comfortable sharing the details in a public space like this, but the answer that it gave was not just a misinterpretation of the question, it was actually entirely wrong on the merits of its own in
162.
by lolinder 1y ago
I considered it, but given the personal nature of the topic I want to keep the details off the public internet. I realize that makes my observation less specific than it could be, but I think the result should be relatively unsurprising to
163.
by lolinder 1y ago
An example of the prompt engineering phenomenon: my wife and I were recently discussing a financial decision. I'd offered my arguments in favor of one choice and she was mostly persuaded but decided to check in with ChatGPT to help rea
164.
by lolinder 1y ago
To have a good mental model for modern AI agents you have to understand both the LLM and the other stuff that's built up around it. OP is correct about the behavior of LLMs, and that is valuable information to keep in mind. Then you la
165.
by lolinder 1y ago
Ah, that makes sense. My brain inserted a "to" before "it".
166.
by lolinder 1y ago
Yeah, definitely don't do turn off auth for things that matter, if only because I'm not perfectly disciplined at keeping my guests off my main network. I'm talking about tiny services that do one thing—ollama, tts/stt, a
167.
by lolinder 1y ago
You're right, but since when has long-term thinking been a feature of the average American business?
168.
by lolinder 1y ago
> it’s a waste of resources for seniors to be doing the work that the juniors and 2s can do. Hunting down infrequent or low priority bugs, fixing small layout issues, etc. These are tasks that in the current environment often get pushed
169.
by lolinder 1y ago
Any serious approach to security begins with a reasonable and clearly defined threat model, and my threat model for my home network doesn't currently include a team of superheros targeting my file backups in an effort to save the world
170.
by lolinder 1y ago
A stochastic pattern to generate language-like text in the early 1400s is a lot more interesting than gibberish.
171.
by lolinder 1y ago
I'm not running a business, I'm running a home. The threat models are totally different and I adjust my security posture accordingly. Besides, I don't bother with auth for simple services, not stuff that actually hosts data
172.
by lolinder 1y ago
> Since my PKMS is hosted online to manage notes across devices, I have multiple layers of security to ensure my notes are kept private. {Screenshot of a login form} The biggest life hack I can recommend for a self hoster is to set up a
173.
by lolinder 1y ago
Statistical analyses such as this one consistently find patterns that are consistent with a proper language and would be unlikely to have emerged from someone who was just putting gibberish on the page. To get the kinds of patterns these tu
174.
by lolinder 1y ago
I couldn't care less about Anthropic or MCP—as I noted, I'm a critic of MCP—but pedants bug me quite a bit especially when they're wrong. > The API is fed into the LLM as context, the response is part of the output. So y
175.
by lolinder 1y ago
> But that's not what MCP does. It is a tool created by anthropic ( 2nd most used LLM) to provide portabiliry and vendor neutrality between different LLMs. It's like terraform for LLMs. Given that your only contributions in thi
176.
by lolinder 1y ago
I mean, if you're paying for the subscription but never use it that's effectively a donation, right? OP explains their motivation for not accepting small donations in another comment: https://news.ycombinator.com/i
177.
by lolinder 1y ago
It's a pretty bad spec that has a lot of artifacts of having been generated by an LLM. But at least it exists!
178.
by lolinder 1y ago
I'm saying that as long as websites use the username/password model alongside passkeys with no way to turn off the former, you're just as at risk of phishing with passkeys as I am with domain-bound autofill. Either one of us
179.
by lolinder 1y ago
I'm in a similar position to the author—running NixOS on a Framework 13 for about a year—and while I agree with some of their points I think I disagree with others. > However, this means that your projects end up populated with Nix
180.
by lolinder 1y ago
This is a really shallow dismissal, and I say that as someone who is outspokenly critical of MCP [0]. As you yourself say, the context is the text that is fed as input at runtime to an LLM. This text could just always come from the user a

More ›