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

Rsync 3.4.3 has hundreds of Claude commits

mastodon.gamedev.place

Post reply on HN
Print view View original

51–60 of 86 posts

Re: Rsync 3.4.3 has hundreds of Claude commits

#52

Post by sph »

bhaak wrote: post #45
vbtechguy wrote: post #41

What's the difference between experience a human made bug versus an AI made bug in software?

A human preserves more context and might remember what they did and when pointing out a new bug, they often have an idea what's wrong.

A human also learns from their mistakes and grows their skillset.

I cringe any time I read loaded questions like GP's. Have they ever met a human in their life?

Re: Rsync 3.4.3 has hundreds of Claude commits

#53

Post by sph »

bhaak wrote: post #44
CursedSilicon wrote: post #5

I saw an exceptionally long and thoughtful post on Mastodon from "Space Hobo" https://teh.entar.net/@spacehobo that definitely deserves reprinting here ----- I actually worked at the same place as Andrew Tridgell, over a quarter-century ago. I got to know a few of the OzLabs folks during their immediate post-IBM years, and always had the highest respect for them in that way where you feel acute impostor syndrome when…

> If TRIDGE of all people can't handle #LLMs without a slopocalypse, no one can. > That means you. That means someone you admire who is intelligent and careful and considerate. Not even someone whose opinions on technology you respect a great deal. I disagree. The amount of commits is not from somebody who is carefully reviewing the new code and considering the changes done. It's from somebody who thinks they are in…

This is a common fallacy: that vibecoding is not that bad if one carefully reviews the output. It's true in a vacuum, but what happens when you're late and stressed out and can't be bothered with doing a proper job.

Humans are lazy, and the mistakes of being lazy when vibe coding are orders of magnitude larger than being lazy when you have to do the damn thing yourself. In fact in the latter case, laziness is a feature.

If the AI-powered software world depends on humans not being lazy, we're all fucked.

Re: Rsync 3.4.3 has hundreds of Claude commits

#55

Post by My_Name »

bhaak wrote: post #46
My_Name wrote: post #24

This is a problem of insufficient checking happening in-between a PR being made, and it being committed. Imagine you have a low quality coder in your coders, they produce a lot of code, but while some of it is fine, some of it is... dubious. That is no different from an AI and the way you deal with it is the same. You check the PR before committing it. To allow PRs from them (or anyone really) to get merged without p…

The commits were all from the original inventor of rsync. Not a low quality newbie coder.

"To allow PRs from them (or anyone really) to get merged without proper checking for bugs etc is just sloppy repo management."

I stand by my post.

Re: Rsync 3.4.3 has hundreds of Claude commits

#56

Post by bhaak »

My_Name wrote: post #55
bhaak wrote: post #46

Earlier quoted context omitted.

The commits were all from the original inventor of rsync. Not a low quality newbie coder.

"To allow PRs from them (or anyone really) to get merged without proper checking for bugs etc is just sloppy repo management." I stand by my post.

Your post makes no sense unless you speak about project management in general.

The commits in question are no pull requests.

Re: Rsync 3.4.3 has hundreds of Claude commits

#57

Post by 3form »

kelnos wrote: post #30
3form wrote: post #20

Earlier quoted context omitted.

I don't think using AI as such is the core problem here. It's the type of use. Vibe coding, brain off coding and blind trust are the issue, and an issue everywhere, just enterprises were never really about quality in the first place. But eventually, they too will generate more crap than they can handle.

I mean, we have no idea how tridge is using claude. I would easily give him the benefit of the doubt that he's not vibe-coding, is involved in the change (not just turning on auto-accept), and reviews the output before committing. But it seems like everyone's immediate hot take here and on Mastodon is to assume the worst and shit on him.

To be precise, I meant the open source as a whole, as this is what the parent poster mentioned. I don't know about Tridge, I would review the changes first to see what happened there.

For rsync in general, I would say that the important value is the trust in it not breaking my data, more so than other projects. That trust can be broken in different ways, AI or not, and the means are of a secondary concern. I hope this gets sorted out soon.

Re: Rsync 3.4.3 has hundreds of Claude commits

#58

Post by zulux »

sph wrote: post #48

I reckon we will soon see a growing movement of maintainers forking popular open-source projects to the point before vibecoding was introduced to the development process. I can definitely see myself supporting this. Vibecoding promotes the uncontrollable growth of features, and thus bugs, when the vast majority of software benefits from stability. It should be possible to be done with development, barring security pa…

Counterpoint: AI makes refactoring so much easier. Many of our monolithic code bases have gotten smaller and more organized because we were too lazy to refactor.
Post reply on HN