Programming thread

  • Want to keep track of this thread?
    Accounts can bookmark posts, watch threads for updates, and jump back to where you stopped reading.
    Create account
I know lambda expressions get used everywhere, but I don't like it and refuse it out of principle.
Agreed. I've always said that the biggest problem with Lisp is that its userbase consists of Lisp programmers. This sounds like a tautology, but I'm using "Lisp programmer" as shorthand for a smug asshole with a penchant for obscurantism and a superiority complex. They have this notion that using higher-order functions is big-brained behavior, so they make a point of conspicuously using them, even though the lambdas add a lot of syntactic noise and Lisp has had macros since the '60s. Whenever I write a higher-order function in Lisp, I always end its name in -fn, then define a macro with the bare name that expands into a call to the function with arguments wrapped in lambdas as appropriate. Its absurd that in Scheme you need to write something like
Code:
(map (lambda (x y)
       (+ (* 2 x) y))
     xs ys)
when you could be writing something like
Code:
(map (x xs y ys)
  (+ (* 2 x) y))
(or maybe
Code:
(map ((x xs) (y ys))
  (+ (* 2 x) y))
if you want to mimic the syntax of let and do), but if you could do that, it would be the macro definition that uses a higher-order function and not you, so you couldn't feel smug about it.
 
Its absurd that in Scheme you need to write something like
1. Clearly states that you are passing a function (that takes 2 arguments) as argument, and then 2 lists.
when you could be writing something like
2, For me it looks convoluted. You mix inner function args with lists that you pass. At first glance why are x and y used in function but no xs and ys? Seems like higher cognitive load.
 

It's crazy. The "what the fuck" title I actually don't think is clickbait. Its just the natural reaction I think most people will have to hearing what is in this video.

To not spoil it outright I'll hide what the video is about in here.

Right now they are doing studies on human brain cells powering computer chips.

They made human brain cells on a chip learn to play a modified version of doom in this study, and did the same thing with pong in the past.

Apparently the goal is to eventually use them for things like robotic arms. They said these are much easier to train than normal llm's also.

Personally I find this incredibly fucked up. And at best it's ethicality is questionable. And really the implications for where this could go in the future are actually horrific.

They say they don't even fully understand how this actually works.

But the most horrifying thing of all. Is it uses vibe coded python.
 
Last edited:
It's crazy. The "what the fuck" title I actually don't think is clickbait. Its just the natural reaction I think most people will have to hearing what is in this video.
It's been going for some time now. The issue seems to be keeping those neurons alive, as you need to feed them and keep metabolic waste away. So any scaling will be hard.
On their site they say that they can keep them alive up to 6 months, so not exactly long time. Especially if you remember that human neurons often last a lifetime.

I feel like it will go the way that printed meat went. Just not feasible, as cells cannot support themselves, so you need all the supporting "animal". Or waste time/energy/complexity on workarounds that do not scale.

So in summary probably a investor scam, like most of new tech is today. Seems impressive but might have limited real world use.
 
The issue seems to be keeping those neurons alive, as you need to feed them and keep metabolic waste away. So any scaling will be hard.
A way to alleviate some of the pain would be to switch to some nervous cell that's less finicky than human, like some kind of extremeophile flatworm neuron or something. Probably easier to "overclock" these and absolutely more ethical.
 
We may finally see a future where JavaScript is no longer mandatory for programming in browsers, outside of every browser and its source code disappearing off the face off the earth this is one of the better things that could happen to "the web".

No idea why it took this long to even consider this but better late than never I guess, also has a polyfill thing to try it out if anyone cares but we probably won't be able to use this until years from now (if ever).

 
We may finally see a future where JavaScript is no longer mandatory for programming in browsers, outside of every browser and its source code disappearing off the face off the earth this is one of the better things that could happen to "the web".

No idea why it took this long to even consider this but better late than never I guess, also has a polyfill thing to try it out if anyone cares but we probably won't be able to use this until years from now (if ever).
I just wish the options were better than JavaScript or the gay neo-JVM that is WASM. I don't think a proliferation of VM-style intermediate-representation-as-portable languages is great either. The problem is that webcucks are so buck-broken by JS at this point that even the steaming pile of garbage that is WASM looks preferable. Meanwhile, I'll be over here making websites that don't use either.
 
Has anyone here ever touched HTMX or is it a complete meme? I've never done front end stuff before and I really do not want to have to tangle with React, Vue, or whatever the other massive JS beasts are.
 
I just wish the options were better than JavaScript or the gay neo-JVM that is WASM. I don't think a proliferation of VM-style intermediate-representation-as-portable languages is great either. The problem is that webcucks are so buck-broken by JS at this point that even the steaming pile of garbage that is WASM looks preferable. Meanwhile, I'll be over here making websites that don't use either.
What alternative is there to bytecode? It's a low trust environment, you can't just run pure assembly code some website has just sent down to you. Pure HTML isn't really any better and doesn't really solve the problem of what the modern Internet has become (program distribution platform), HTML sucks and should not be the interface to presenting things to the screen. WASM has problems (64 bit mode runs slower than 32 bit mode on 64 bit machines) but a bytecode engine is what should have been there instead of Javascript from the very beginning.

Web browsers shouldn't be what they are today, it should be a bytecode engine + a simplified UI API (and probably graphics API access for people who want more control) instead of the absolute trash that is HTML + CSS, there is nothing cool or good about hand writing XML.
 
Web browsers shouldn't be what they are today, it should be a bytecode engine + a simplified UI API (and probably graphics API access for people who want more control) instead of the absolute trash that is HTML + CSS, there is nothing cool or good about hand writing XML.
Generally agree, however I'm fine with HTML as part of the model. There should be a hundred different viewers for HTML documents, not a rigid W3C defined display engine that MUST look the same for EVERY viewer on EVERY device out there. That rigidity is why it is easy (and lazy) to deploy "apps" anywhere, but its limitations also give us the modern JS nightmare when those apps require more than mere display.

In a world of SSL, E2EE, SSO and other auth, we really should be allowed to trust what comes over the web at a lower level.
 
rigid W3C defined display engine that MUST look the same for EVERY viewer on EVERY device out there.
It would be cool if the HTML standard got culled down to a minimum of tags, no CSS, no JS. And people would have their own CSS in the browser. It would put the focus on delivering just the important content in an easily understandable format.
It would obviously make most modern "apps" unusable, which is a bonus in my eyes.
 
It would be cool if the HTML standard got culled down to a minimum of tags, no CSS, no JS. And people would have their own CSS in the browser
boy would you love the gemini protocol
only markdown-like text + images
own css in browser
no js
only kind of input is text input to allow search engines

ETA: an image is only downloaded and viewable after you click on a link to it, only then it is embedded into the page
 
Last edited:
Back
Top Bottom