Sociopath Community

Buttered Toast said: 

I also did some querying for an optimal, single programming language.  However, I think it would be a better first stage to try and work out the best features for a particular paradigm first, but that might not be more practical than just trying to get the best out of all the possible worlds right away.

Claude Sonnet 3.7 said:

Building a Unified Language
To create a unified language incorporating these strengths:

Core Foundation:

Start with Rust's ownership model for memory safety
Add ML-family type inference for expressiveness with safety
Include pattern matching as a first-class construct


Modularity:

Support multiple paradigms through flexible syntax and semantics
Allow pure functional code with guarantees when needed
Provide OOP capabilities but discourage inheritance in favor of composition


Concurrency Model:

Adopt Erlang's actor model for distributed systems
Integrate with Rust's ownership for safe concurrent memory access
Include Go-style lightweight threads/coroutines


Developer Experience:

Focus on excellent error messages like Elm/Rust
Build a comprehensive package manager and build system
Design for gradual adoption and interoperability


Implementation Strategy:

Start with an LLVM-based compiler for performance
Design a VM for portability and dynamic features
Create progressive typing system that can be as strict or flexible as needed


Practical Considerations:

Include a foreign function interface for existing code
Provide tooling for gradual migration from other languages
Focus on compile-time guarantees while maintaining runtime flexibility



This unified language would balance the safety of Rust, the expressiveness of functional languages, the practicality of Python, and the concurrency of Go/Erlang while maintaining a coherent design philosophy rather than just being a collection of features.

 Operating systems are written in C. Hard to beat C for machine level programming. For front end stuff, hard to tell, there are so many now.

Searching for one programming tool to do all jobs is like telling a carpenter that he can only have one saw to cut wood. Cross cut, ripping, miter, chain, sawzaw. jig, hole. Not going to happen.

It's probably not worth picking apart that analogy and its incompatibility.  Convince me English isn't already kind of doing what I'm describing for humans.  ...Hmm, actually computers are already making having different languages sort of a relic of humanity, able to translate and remove the superfluous layers.  (I wouldn't argue that any of this is toward optimization, though.)

But I'll tell you the end of the story for computer languages: Wolfram Alpha.

Buttered Toast said: 

It's probably not worth picking apart that analogy and its incompatibility.  Convince me English isn't already kind of doing what I'm describing for humans.  ...Hmm, actually computers are already making having different languages sort of a relic of humanity, able to translate and remove the superfluous layers.  (I wouldn't argue that any of this is toward optimization, though.)

But I'll tell you the end of the story for computer languages: Wolfram Alpha.

 You have a point about english.

I just see two types of coding models, machine level and interface level.

I think i will post here everytime the AI fails me. Seems like a great idea.

I use GPT

So today, I asked it to write me some code where it extracts data from an HTML webpage into a c# class.
First attempt, it wanted me to download a fucking nuget package for HTML parsing (which I use on this forum btw), for something so simple. Laughable.
I told it to use regex.

So it generated some code, but after running it, it did not get all the results. And some results were wrong, because the data in some parts was missing in the HTML page and it didn't consider it. I told it that it failed and that sometimes data is missing, so it did corrections.

These corrections had the exact same result. Exactly the same erroneous result.

I could of combed through its code and find the error, but instead I wrote the thing myself and it took me less time then asking the AI.

Good said: 

I think i will post here everytime the AI fails me. Seems like a great idea.

I use GPT

 

First, why don't you do what you said and stop posting in my thread.

Second, you didn't watch the video, you're using the wrong AI. I really don't know why I'm helping you.

Third, posting your failures because you can't figure out how  to use an advanced AI tool. lol

I read this a few days ago and I think I still have the context. And let me say I make my money working with AI I'm not totally a casual here. Good is right, asking even Claude or GPT to work out a huge project will produce dogshit code. Once you start involving libraries or complicated work, the thing starts to fall apart. Even React code SUCKS. Is it good for bouncing ideas off of, or targeted questions where you need some bit of information? Yes absolutely, I use AI to bullshit all the time. I can navigate Java with AI even though I never learned it. But now I can use Maven. But I'm sure my Java blows.

People worry about purism a lot, but AI is just another tool. If you have a good tool, you use it. AI is a heated topic because people can sense it's going to put people out of work. Well, that's life...similar thing happened during the Industrial Revolution, but even worse because those people had less of a social safety net. If we don't perfect it and use it to achieve dominance, the Chinese will, etc etc.

On the topic of C and operating systems, interesting thing happening lately with the Linux kernel. Rust is more secure with memory and even Torvalds is saying they need to let the Rust team do their thing. But so many of the contributors always did C, they just want their C. They tried to stop Rust integration, even. But I don't see how Rust isn't the future 

Tryptamine said: 

 

but AI is just another tool. If you have a good tool, you use it. AI is a heated topic because people can sense it's going to put people out of work.

On the topic of C and operating systems, interesting thing happening lately with the Linux kernel. Rust is more secure with memory and even Torvalds is saying they need to let the Rust team do their thing. But so many of the contributors always did C, they just want their C. They tried to stop Rust integration, even. But I don't see how Rust isn't the future 

 Exactly. AI is an efficiency tool. A tool very similar to robotic welders that put many auto workers out of jobs. And coders that do not use AI will be left behind.

Looks like Rust is still experimental in the Linux kernel.

"If you are an end user, please note that there are currently no in-tree drivers/modules suitable or intended for production use, and that the Rust support is still in development/experimental, especially for certain kernel configurations."

next example, i tried to use the official GPT, for a first time in a few years, I usually use my own, for its Ghibli style filter.

I am sure I didn't use it right, but what GPT said made no sense and it didn't help me use it right at all. I gave up after a few tries, I don't have hours to spend on this.

Here are parts of the convo:

Somehow it forgot it can do it???

Then it said:

GPT said:

I wasn't able to process your request because it violates our content policies. Let me know if you'd like a different type of image transformation, and I'd be happy to help! 😊

 So I posted another pic:

And it again, forgot it can do it ?? in the same convo lol.

Here is the end of the convo:

It didn't do it again and just refused entirely this time, I got tired of nagging an AI to do its job, so I stopped trying.

Good said: 

 

I am sure I didn't use it right, but what GPT said made no sense and it didn't help me use it right at all. I gave up after a few tries, I don't have hours to spend on this.

More failures, huh?  I find being specific to the input and the output variables to be important. I've had great success with IA coding. You should really watch that video I posted.

LiYang said: 

Good said: 

I am sure I didn't use it right, but what GPT said made no sense and it didn't help me use it right at all. I gave up after a few tries, I don't have hours to spend on this.

More failures, huh?  I find being specific to the input and the output variables to be important. I've had great success with IA coding. You should really watch that video I posted.

As he has expressed, which has been my experience as well, accessing the functions without the use of a middleman tends to give better results. 

The services hosting them otherwise are too busy maintaining their image to lean in on progress.