[bwfan123]

So, you have a vibe-coded TUI which happens to work, and then, as a workaround you vibe-translate its engine to make it more performant. Where does that leave you ? Basically, fully dependent on AI to fix whatever breaks. Workaround on a workaround is the way I see it, and it aligns with the AI design mentality in general. For a variety of usecases, this might still be a win in terms of overall cost. But, for software that is intended to be built to last, I dont see this approach working out.

[galangalalgol]

That does seem to be the way many use it. I'd be very surprised if they didn't have to insert a rule to prevent opus from constantly asking why they didn't just use ncurses. I just asked sonnet for design options for a tui to onteract with llms to perform sw dev tasks. After describing the tui it immediately suggested ratatui and crossterm as the tech stack. I feel like they must have ignored even the advice of their own llm to come up with this solution.

Edit, the sonnet question shouldn't be taken as proof, it knows I'm a rust dev.

[simonw]

They decided on the tech stack for Claude Code over a year and a half ago.

[tough]

ratatui was still an excellent choice over a year and a half ago.

Nobody but a JS dev would choose react-on-terminal shenanigans over it

I could see CC getting a ratatui port eventually actually

[jeremyjh]

Bun was not ported to Rust for performance reasons and its not clear to me how anyone can think it was. The reason they've given is memory safety.

[vips7L]

It was ported for vibe reasons.

[galaxyLogic]

Plus 1) There's much more reusable open source code written in Rust than in Zig 2) There are more Rust developers than Zig developers.

[LtWorf]

I don't think there being more rust developers should matter if we're talking about a company that solved coding.

[galangalalgol]

It does, mostly because there is more recent rust in the training corpus. Rust that is almost entirely available under permissive licenses

[simonw]

I'm going to guess Anthropic are OK with "fully dependent on AI to fix whatever breaks".

[jubilanti]

> for sw that is intended to be built to last, I dont see this working out.

the era in which the tech industry built software to last was over long before LLMs, especially VC-backed startups.

[zdragnar]

Did it really exist? I remember buying games on discs as a kid that had bugs in them that prevented you from actually finishing the game at all.

I'm pretty sure "built to last" was only ever more true in specific contexts, and wasn't particularly more true than today.

[nozzlegear]

The era in which the [VC-backed, SV area] tech industry built software to last was over long before LLMs.