Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Automatic Programming

antirez.com

Post reply on HN
Print view View original

251–257 of 257 posts

Re: Automatic Programming

#251

Post by jmalicki »

theshrike79 wrote: post #249
manmal wrote: post #200

Earlier quoted context omitted.

My experience with both Opus and GPT-codex is that they both just forget to implement big chunks of specs, unless you give them the means to self-validate their spec conformance. I’m finding myself sometimes spending more time coming up with tooling to enable this, than the actual work.

I just use beads or Github issues. Plan/spec first, split it into issues. Then reset context and implement each task one by one. Nothing gets forgotten.

A key part of this is making sure bite sized issues reference any related holistic concerns like code quality, testing, documentation style, commit strategy, git workflow etc.

In my experience you have to refer to the relevant docs on things explicitly in every single issue for it to work well.

Re: Automatic Programming

#252

Post by jmalicki »

lII1lIlI11ll wrote: post #182
dugmartin wrote: post #21

I have 30+ years of industry experience and I've been leaning heavily into spec driven development at work and it is a game changer. I love programming and now I get to program at one level higher: the spec. I spend hours on a spec, working with Claude Code to first generate and iterate on all the requirements, going over the requirements using self-reviews in Claude first using Opus 4.5 and then CoPilot using GPT-5.…

How does the resulting code look like though? I found that while can spit out barely working C++ code fast, I then have to spend 10x time prodding it to refactor the code to look at least somewhat acceptable. No matter how much I tell it that it is a "professional experienced 10x developer versed in modern C++, a second coming of Stroustrup" in per-project or global config files it still keeps spewing the same crap b…

Instead of asking for it to write good code, why not write a coding standards doc that makes it explicit and reference it in you plans?

Re: Automatic Programming

#253

Post by lII1lIlI11ll »

jmalicki wrote: post #252
lII1lIlI11ll wrote: post #182

Earlier quoted context omitted.

How does the resulting code look like though? I found that while can spit out barely working C++ code fast, I then have to spend 10x time prodding it to refactor the code to look at least somewhat acceptable. No matter how much I tell it that it is a "professional experienced 10x developer versed in modern C++, a second coming of Stroustrup" in per-project or global config files it still keeps spewing the same crap b…

Instead of asking for it to write good code, why not write a coding standards doc that makes it explicit and reference it in you plans?

Because:

1. LLMs tend to happily ignore those unless constantly prodded.

2. By the nature of my work I tend to spend a lot of time doing small modifications and fixes in other people's code for research and demo purposes. I can't just write a comprehensive code style documentation for every single library I touch, hence my attempts to improve the resulting code quality of Cursor agents with general global rules without much success.

Re: Automatic Programming

#254

Post by motoboi »

theshrike79 wrote: post #250
motoboi wrote: post #93

We maybe witnessing the last generation of master software artisans like antirez. This is beautiful to see, their mastery harnessing the power of the intelligent machine tools to design, understand and build. This is like seeing a master of image & light like michelangelo receiving a camera, photoshop and a printer. It's an exponential elevation of the art. But to become a master like michelangelo one had to dedicate…

I don't see anyone lamenting the generation who single-handedly built games in machine code for the early home computers. No art was lost, we just evolved and started using more advanced languages that didn't require as much dedication to work in.

Yeah. The guy who carefully arranged assembly instructions to match the memory magnetic drum movement latency perfectly so that when one instruction finished executing the drum was exactly over the next instruction to be read and executed.

Those guys? Masters. Their craft? Reinvented, rearranged, some parts of it abstracted away by a more intelligent machine.

Current masters don’t know about memory drums and their read latency, nor about assembly instruction set execution time, but they are still crafting interaction between machines and the physical world.

So the next generation will be master of what? Which skills will they possess?

Re: Automatic Programming

#255

Post by thedevilslawyer »

sarchertech wrote: post #186
thedevilslawyer wrote: post #158

Earlier quoted context omitted.

Absolutely. As any logical person should be.

This is obviously false on the face of it. Let’s say I have a patent, song, or a book that that I receive large royalty payments for. It would obviously not be logical for me be in favor of abolishing something that’s beneficial to me. Declaring that your side has a monopoly on logic is rarely helpful.

Either by democracy (more consumers than produces), or ethically (thoughts and intellect are not property), it's logical. I guess it is not logical for someone who makes money with it today.

Re: Automatic Programming

#256

Post by 4corners4sides »

This one touches on the metaphysical union of all prior minds and the implicit “natural” element to “artificial” intelligence. Yes, indeed sometimes reading LLM written text you cannot help but wince and cringe at the “contrastive framing” writing style (it’s not X, it’s Y) but these models did not arise from some kind man-hating void in space. They arose by learning from everything that can possibly be learned from that humans have made available digitally. Antirez says that “Pre-training is, actually, our collective gift that allows many individuals to do things they could otherwise never do, like if we are now linked in a collective mind, in a certain way”. This is borderline cultish but I can’t help but agree with it.

Re: Automatic Programming

#257

Post by rellfy »

kevmo314 wrote: post #231
rellfy wrote: post #109

I arrived at a very similar conclusion since trying Claude Code with Opus 4.5 (a huge paradigm shift in terms of tech and tools). I've been calling it "zen coding", where you treat the codebase like a zen garden. You maintain a mental map of the codebase, spec everything before prompting for the implementation, and review every diff line by line. The AI is a tool to implement the system design, not the system designe…

> review every diff line by line If you're this meticulous is it really any faster than writing code manually? I have found that in cases where I do care about the line-by-line it's actually slower to run it through Claude. It's only where I want to shovel it out that it's faster.

Yes, I definitely think it's much faster than writing it manually. For a few weeks now, >95% of the code I've authored wasn't written manually.

Sometimes you only care about the high level aspect of it. The requirements and the high-level specification. But writing the implementation code can take hours if you're unfamiliar with a specific library, API or framework.

"review every diff line by line" is maybe not the best way to have described it, I essentially I meant that I review the AI's code as if it were a PR written by a team member, so I'd still care about alignment with the rest of the codebase, overall quality, reasonable performance, etc.

Post reply on HN