Hacker News new | past | comments | ask | show | jobs | submit | mattmanser's comments login

The unfortunate truth is that the one true way these days is React, and if you're doing anything else you're in for a world of pain.

The actual truth is that you don’t need any of these frameworks at all. Vanilla JS & HTML has come a long way, and LLMs like Claude can build Vanilla JS with no trouble.

So AI does stifle adoption of new tech then?

Both things can be true. I have also been frustrated trying to get LLMs to work with SwiftUI, given that it’s slightly newer, changing frequently, and has lower adoption relative to something like React.

Honestly this is a short term problem. Eventually the models will have larger context windows and will be able to easily browse brand new documentation or example code.


No such thing as one true way with React and the pain is just as bad if not sometimes worse.

Do you choose createreact app / webpack / vite / next.js ?

Do you choose built in react state/contexts, redux, recoil, zustand, jotai state libraries?

Do you do data tables yourself or use the behemoths TanStack, react-table-libray, MUI x-data-grid

What about HTTP requests? Fetch api? axios? rtk-quey?

What about forms and validation? KISS simple, do it yourself? react-hook-form + zod, tanstack form?

Now how do you integrate them all in to your app, managing each edge case for each dependency with their own opinionated way of doing things not quite neatly working with your other dependency but can brute force it to work with some effort and hacks?

I find things like doing basic vanilla JS, or even things like of Svelte or Elm far less pain than React projects i've worked on. Svelte/Elm did require reading the docs not relying on a LLM.


Looking at the react devs in my company, they already live in a world of pain and confusion. So many simple, silly things turn complicated because of it.

But I thought it was a joke, or do people really choose their tech stack based on LLM support?

And now you've just got lots of different complications, like that seawater is very corrosive.

> And now you've just got lots of different complications, like that seawater is very corrosive.

Is it as corrosive as lightning? All space elevators will encounter that problem pretty regularly.

We've been running cables under the ocean for over a century now, there's ways to address seawater corrosion and intrusion.


I find it difficult to imagine a scenario where building a space elevator is possible but sea water corrosion is an insurmountable problem.

If you're building the base of the elevator in water, your whole operation is basically operating in water.

The effort to do that is huge.

It's basically a big over complication.


I've been doing this a long time and seen a few different apps use config in database. There's different levels of config you're talking about here, but general app config should generally not go in a db.

No-one ever changes the bloody things and it's just an extra thing to go wrong. If it only loads on startup, it achieves nothing over a bog standard config file. If it loads every request you've just incurred a 5% overhead on every call.

And it ALWAYS ends up filled with crap that doesn't work anymore. Because unlike config files, no-one clear it up.

Worse still is when people haven't made it injectable and then it means unit tests rely on a real database, or it blocks getting a proper CI/CD pipeline working.

I end up having to pick the damn thing out of the app.

Use a config file like everyone else that's probably built into the framework you're using.

To be honest, most of the time I've seen it has been when people who clearly did not know their language/framework who wrote the app.

I'm not saying it's you, but that's been my honest experience of config in the db, it's generally been a serious code smell that the whole app will be bad.


There's differences to what kind of configuration you'd want to have in a config file (or environment variables, or some other "system level" management tooling) versus a feature flagging system.

In my experience, feature flagging is more application-level than system-level. What I mean by that is, feature flagging is for stuff like: roll this feature out to 10% of users, or to users in North America, or to users who have opted into beta features; enable this feature and report conversion metrics (aka A/B testing); enable this experimental speedup for 15 minutes so we can measure the performance increase. It's stuff that you want to change at runtime, through centralized tooling with e.g. auditing and alerting, without restarting all of your application servers. It's a bit different than config for like "what's the database host and user", stuff that you don't want to change after initialization (generally).

Regarding the article though, early on your deployment pipeline should be fast enough that updating a hardcoded JSON file and redeploying is just as easy as updating a feature flag, so I agree it's not something to invest in if you're still trying to get your first 1000 users.


For some kind of software, another call to the DB is the best way to add bog-standard functionality without adding complexity and failure modes.

Granted, not for all software. And there's something to be said about a config file that you can just replace at deployment. But that's something that varies a lot from one environment to another.


Someone else has written this exact code on the internet, OpenAI stole it, and now chatgpt is regurgitating it. Just like it can regurgitate whole articles.

You need to stop being wow'd by human intelligence masquerading as AI!


Where? Searching for "Use the Double Reflection Algorithm (from Wang et al.)" only returns this post.


Unlikely. I wrote this myself. Of course it is in a private github repo.


you are so confused it’s not even funny


In the UK it costs £12 and takes 5 minutes. It costs between £300-600 per year for an accountant to file your accounts, and £12 for your confirmation statement.

And did when we were part of the EU.

Cheaper and quicker than America.

You don't incorporate in the EU, you incorporate in one of the 27 different countries.

All have wildly different requirements.


UK was vastly different from the mainland EU. You're right that the EU is not singular, but once we start talking of Germany, the Netherlands, France, etc. - we quickly hit regulations that bear no resemblance to a free market and some of which are incompatible with IT business whatsoever.


(These costs rocketed last year. Incorporation is now £50, and an electronic confirmation statement is £34.)


You also have all sorts of ancillary fees like the Information Commissioner’s annual charge.


This feels like a sentiment I'd have supported in 2019, wished we're still true in 2022, and is completely outdated in 2025.

I'm a full stack who phases in and out of feontend. I'm now all in on React + MaterialUI. Even though I think they're a bit shit.

I loved jQuery. I loved doing `$(...raw html)`. But everything it did, the browsers do natively now.

You just don't really need it anymore.


I think that's an oft repeated over statement of reality.

I've enjoyed a good 60+ hours in it, it's good. But not great. I didn't pay for it, not sure I'd have been as happy would I have paid for it.

It's still very shallow. As a programmer you can see all the really basic algos underlying all the generation.

There are like 10 planet types, with a couple of extra wacky ones. All the stations are virtually identical inside. There are like 8 points of identical interest. And bizarrely they're everywhere. There are 3 alien races. You land on a planet and within seconds spaceships are over flying you. You enter a new system and 10 seconds later a fleet of "random" carriers will turn up within your sightline.

You arrived at a new planet and the creatures are all exactly the same as you've seen before, just more legs, or 4 wings instead of two.

The thing the really bugs me are the sentinels. I get the in game explanation, but it's boring that every planet has them. And the few that don't are weird ones.

Everything is different and yet, apart from a very few minor things, it's all the same.

And the randomness gets grating. Want a cool ship? Have fun trying to grind that out. Want a cool gun? Ditto. You have to resort to going to particular coords supplied by the internet, not in game play.

For me, never before has space felt so small. Elite dangerous suffers from similar shallowness problems, but at least it felt epic.


>over statement of reality.

You don't have to enjoy the game, that's totally fine. But my comment is pretty reality-based.

The game completely bombed on launch and was dragged through the mud (mostly deservedly) in every review. Now has very positive reviews on Steam, people still talk about it, and people still buy it a decade later. It's sold enough copies to support the staff at Hello Games for that time.

If that isn't turning a shitty launch into a success, I have no idea what is.


I've played the game on and off since launch. I think the biggest difference in the game is expectations going in.

A ton has been added to the game but I still play it for the same reason, it's a good ambient exploration game where you get to fly around space but I also still stop for the same reason, it's shallow and repetitive. That was my experience at launch and that was my experience last time I played.

Based on hours played (and I enjoyed them) it was well worth the money but I don't think it's really improved that much over time. There's definitely more to do but it's not much deeper.


I feel same way. Played on Game Pass, not sure I would have valued it if I paid for it. It is very grating that you see same set piece, be it pirates or ruins again and again. I have seen this one before. Oh and to get best output I should just save scum...

And then randomness leads to issues like finding very nice looking end game planet. To make your own base. And even there the option to place it so that certain things refresh.... So game world is not even consistent...


The "sentinels" or whatever is what brought me out of it immediately. There are a billion planets to explore, but someone always got there before you.


Here in the UK I was seeing the recruiter emails begin to return to more normal levels in Sept/Oct. Albeit with lower salaries than 2022/23.

But then it all abruptly stopped and disappeared again for a few months.

It's now started again but very slowly.


It was from a time where Google's ethos was still "Don't be evil" and generally speaking the naked greed triggered by AdSense hadn't infected the rest of the company yet.

So I think a lot of nostalgia is not just for the reader but also for the company Google used to be.

I wonder if in 20 years time there will be the next generation of programmers sneering over vapour-ware Google products while middle managers still buy them products because "no-one ever got fired for buying Google".


It's a really interesting comment, but I would also have loved the context of what you're trying to do?


Without going into specifics, trying to classify a body of freeform text (or rather, many such bodies) along a bunch of dimensions. Imagine a fiction story that you're trying to answer questions about such as "what is the genre?" and "is the protagonist male or female, adult or child?", etc, where I'm instructing the LLM to return values out of a specific range only (aka, enums). OpenAI's new-ish structured outputs feature enforces answers to a user-specified JSON schema, which is awesome, but I'm finding that the actual answers it comes up with are sub-par. There are all kinds of hacks out there like creating additional fields to let it reason out the answers before constraining them to an enum elsewhere, but I have yet to make it work well for me. There are other approaches I haven't yet tried that require much more effort or cost (more expensive models, fine-tuning custom models). It feels like it should work, so the random poking and prodding continues...


Interesting, thanks for sharing.

By coincidence I had been wondering how hard something like that would be to do the other day.


Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: