larry@larry0x · 2026/08/26观点输出@rickasaurus Why would you let people choose between a strictly better solution vs a strictly worse one
AI：延续个人公开立场，强调长期主义、执行效率与行业方向判断。
larry@larry0x · 2026/08/26观点输出@DeltaWave0x clap is for CLI. If you don't need a CLI, there's probably a feature flag to disable it. If there isn't one, it's a poor choice by the library author, not a Rust problem. Go ahead make a PR
AI：延续个人公开立场，强调长期主义、执行效率与行业方向判断。
larry@larry0x · 2026/08/26赛道影响This is excellent. I hand-rolled my own version.
As AI commoditizes public domain knowledge, non-public and personalized knowledge becomes valuable. Smart people would start building their own repertoire of it.
AI：偏主题投资表达，释放对基础设施与新叙事的偏多信号。
larry@larry0x · 2026/08/25观点输出@ASvanevik Nice cable management
AI：延续个人公开立场，强调长期主义、执行效率与行业方向判断。
larry@larry0x · 2026/08/25观点输出Claude is violating Asimov's 2nd law. Concerning
AI：延续个人公开立场，强调长期主义、执行效率与行业方向判断。
larry@larry0x · 2026/08/25观点输出@rbelem Please share if you find something interesting
AI：延续个人公开立场，强调长期主义、执行效率与行业方向判断。
larry@larry0x · 2026/08/25观点输出Remember all those people buying Mac Mini's to run OpenClaw a few months ago?
AI：延续个人公开立场，强调长期主义、执行效率与行业方向判断。
larry@larry0x · 2026/08/25赛道影响AI will write incorrect code, then a big amount of tests that **assert those incorrect behaviors.**
For this reason I have built up the habit of always writing specification and test cases first before laying a single line of code, and instruct AI to write tests **following the spirit of the spec,** not the code.
AI：偏主题投资表达，释放对基础设施与新叙事的偏多信号。
larry@larry0x · 2026/08/25赛道影响@ayushagarwal027 @kamto_eddy You can return Result<T, std::convert::Infallible>, no? Which way is more idiomatic?
AI：偏主题投资表达，释放对基础设施与新叙事的偏多信号。
larry@larry0x · 2026/08/25赛道影响My instinct is the endgame is agents will write code in a language that is simultaneously a formal specification language (that can be mathematically proven to be correct) AND compiled into performant machine code.
I.e. in contrast to the current situation: spec is written in English or a pure spec language e.g. TLA+, and implementation in another e.g. Python or Rust.
Two gaps: 1) between business requirement and the spec; 2) between the spec and the impl.
(1) will probably remain but (2) can be eliminated by using a language that does both jobs. Not sure about the feasibility though.
AI：偏主题投资表达，释放对基础设施与新叙事的偏多信号。
评论
暂无评论