HN user

jjice

6,981 karma
Posts7
Comments1,840
View on HN
Making 1 hour ago

Captures my feelings as well. I don't get the same joy I used to because I end up using an LLM because it'll be faster. I need to come to terms with speed not taking precedence over fun. The act of writing is something I need to learn to refocus on again without the thought of the "inefficiency" of doing so. I want to feel like I did when I was 13 and writing code in the middle of the night.

Beej's writing always impresses me.

Maybe, but I feel like those are all just overwhelming trend/clickbait nonsense videos/articles. E-readers are awesome, but I bet the amount of people that have truly replaced their social media habits with ebooks out of nowhere is near zero.

Also, for anyone that wants to cut doom scrolling, just delete your accounts and delete the apps for the services you're not using in a way you like. In many cases, it's that simple.

Love e-readers though. I've carried one with me (when I had a reason to expect at least 30 minutes of downtime) for a few years now and it's a lot more convenient than the paperbacks I used to lug around.

Transcribe.cpp 3 days ago

I believe you're correct as of the last few iPhone generations. iOS 27 (with newer models of device required) have even better transcription coming as well, all on device.

Oh man I'd love a nice full stack framework in Rust! Django, Laravel, and Rails are very neat. I used to be a micro framework kind of guy, but having everything integrated is more and more appealing over the last few years.

This would take a long time to get feature complete with the core of those big full stack frameworks, but I'm rooting for it! Getting to use the Rust type system with a full stack web framework sounds incredible.

Definitely interesting seeing it grow there. I'm now curious to see if it's overall market usage growing for LLMs (with more OpenAI and Anthopic usage in a similar growth rate), or if it's more heavily weighted towards open models specifically.

With the price to performance of the latest open models, I think most cases for integration into applications would get a best bang for the buck from an open model.

Correct me if I'm wrong, but OpenRouter doesn't tell the whole story.

If I'm using OpenAI, Anthropic, or Google models, I'm probably using their API directly, so OpenRouter won't have those stats to compare to.

All that said, it is very exciting to see open model usage grow via OpenRouter.

No prize is definitely the way to go. My university hosted a 24 hour, in person hackathon every spring. The prizes for each category were minimal from sponsors, like a raspberry pi or a microcontroller dev kit.

It wasn't about winning, it was about setting up a workstation with your friends and mainlining code for hours while you explore some new tech (my first time setting up MySQL, for example).

Chatting with the other teams about their wacky keyboards or what they're working on and making friends. Lots of good times.

I use the Tab Wrangler extension and have it nuke everything after like 8 hours or so. If I need it again, I can search for it in that extension. I've probably gone back to search for a tab two or three times in the last five years. That 8 hour window is almost always plenty for anything I actually care about.

GPT OSS was post Llama 3 and pretty strong at the time. But yeah this is the first seriously competitive non-Chinese open model in a good bit now.

If you work in a legacy PHP codebase, it's inevitable that you'll see $_GET in the middle of what you thought could be isolated business logic. The coupling that was done in my experience old PHP codebases is awful.

I guess I can't blame PHP because it comes down to the devs to enforce decent boundaries, but I think PHP makes it easier than others to do so.

PHP is so much better now, but if you're writing PHP, there's a very good chance you're not working with that new, clean PHP.

GPT-5.6 13 days ago

Simon did start the pelicans on bicycles as an SVG, but I think it's more of a fun goofy thing to see how the model performs at. I don't think it has a direct correlation to a model's performance though.

But then you have to deal with manual memory management. Not a bad thing, but it's the big win you get from GC languages and Rust.

Without accusing anyone of anything, I do think that this coming from the head of Zig, who gets a lot of negative publicity from the Bun rewrite (unjustifiably, it's a wonderful language) makes it harder for me to take this without wondering if there's some animosity that's really the main complaint.

Two, I actually don't have any personal criticisms of Jarred

The author says two things that really popped out to me that you could say are "professional" and not "personal" criticisms, but I think they're still rude and contrast this statement.

Jarred was already writing slop well before he had access to LLMs

The grapevine was large and healthy and full of juicy grapes, and all those grapes contained the juice of the same message: Jarred was a stinky manager. Poor communication, unrealistic expectations, low empathy, no experience

Now, both of these may be true. I don't have any evidence though, so I don't know how to take it.

All this to say, I'll take both of these posts without a ton of salt when it comes to the non-technical parts.

Oxide and Friends has a good mix of light banter and interesting technical talk across the stack, leaning lower. They have a lot of great commentary on a range of things. Their LLM discussions are particularly interesting to me because they're very reasonable.

I remembering hearing from a former Cisco employee once that in the mid 00s this would happen to them and the knock off router manuals were literal photocopies of the Cisco ones with very half-assed attempts to block of Cisco.

I have it build self contained tools all the time and I've never had an issue getting it to let me download the standalone HTML page. Usually just a "Download" button on the artifact on the page UI screen (just confirmed with recent tool I had it build). Assuming you mean the claude.ai website interface.

For another perspective, it's worked very well for me and my friends. An incredibly powerful model that does some impressive planning and development.

Is it that much better than Opus? It's definitely up there, but Opus is my cheaper workhorse for the kind of work I do.

He is a natural born teacher. The kind of person whose excitement and love to share is contagious. Best of luck to him!

I saw some old code of mine that used PLY a few weeks ago and god damn is that a fine library. Fast, to the point, and a beautiful API.

Vite+ Beta 20 days ago

Cargo itself doesn't pull the dependencies, but yes to Rust's standard library being much more lean than Go. Bring your own HTTP, text templating, and such, but core data structures are provided.

Go gives you a bunch of goodies in the standard library.

Rust provides things like your build system, testing, and package management all together, which is what I assume OP meant.

It's been about five years since I've used Flex and Bison, but if I recall I just didn't check in the generated files and had a Makefile that built everything all together.

If I'm not misremembering that case then, it sounds like this should've never been an issue (well, as long as this is after basic version control and make). Curious if I'm missing something.