131–140 of 145 posts
I would judge commits by what it does not by who wrote it.
your AI coder is worse than a junior developer, because junior devs may write bad code but generally they won't write code that they don't understand. AI on the other hand has no clue what it is writing.
I don’t use public repos very often but I had toyed with the idea of just creating a git user specifically for an agent to use for this purpose so it would not be my user account, is this not standard practice already? Kinda seems obvious to me, I mean so people can tell which parts of my public project were commits managed by an agent.
I do this so that AI can only have limited GitHub permissions. It can't merge, doesn't have admin rights, etc. This after I started catching it commit directly to upstream main without PRs among other things.
Earlier quoted context omitted.
I don't know. For as long as I can remember, game dev has had the reputation of being the most sweat-shoppish of all the software engineering disciplines. I have a hard time believing that game devs aren't also going to find themselves being crushed under the CTO imperative to "use AI or else" like the rest of us.
Ok I should’ve said indie/solo game dev
you could have chosen indy/solo dev in general. solo game dev in my understanding is very hard to make a living in.
Good move, and a good reminder of how much of an echo chamber Hacker News is on AI matters. In here, and big tech at large, it's touted like the unavoidable future that either you adapt or you die. LLMs are always a few months away from the (u|dys)topia of never having to write code ever again. Elsewhere, especially in fields where craft and artistry are valued (i.e. game development), AI is synonym of wanting to cut…
"AI is synonym of wanting to cut corners, poor quality, and to put it simply, slop" A craftsman knows how to use his tools. You can with AI produce very complete, polished, maintainable and tested, secure, performant high quality code. It does take planning and lots of work on your part, but there is a high payoff. So many people just dump a one paragraph brainfart into a prompt and then label the AI "slop". Slop in…
less hard work than writing code myself? a higher payoff than the satisfaction of having written code myself?
i want to be a coder, not a prompt manager. (not sure i want to call that engineer)
Incredibly dumb and nonenforceable policy. What matters is human review and correctness. You're never going to be able to prove that a contributor didn't ask an LLM to help them make some changes, or review/optimize changes that were made. Capable people who like to get stuff done will use LLMs, review their work carefully, and never disclose it. And you'll never be able to tell. People who generated slop PRs won't e…
the policy allows me to reject the things i know are done with AI and it allows me to punish (ban) devs who lie to me when i find out. without a policy i have no argument.
Earlier quoted context omitted.
Will they? Will someone have enough time, skill and dedication to maintain it? I don’t think using AI will by itself make a big enough difference, it’s still a lot of work to maintain a project
> I don’t think using AI will by itself make a big enough difference, it’s still a lot of work to maintain a project I think you are wrong. The "a lot of work maintaining a project" would be reduced, specially issues investigation, code improvement, security issues detection and fixes. SDL isn't a that relevant project, but "ban AI-written commit" - which reading the issue, sounds more like ban "AI usage" - is counte…
Many, many things use SDL. It's one of those bottom pieces in the Jenga tower of infrastructure dependency[0].
Not maintained by some random person (that would be Sean Barret's STB library) but still, it seems irrelevant only because it's already ubiquitous.
Earlier quoted context omitted.
1) Gabe's a front man. He doesn't run Valve. 2) New Zealand is a favorite place for Western apparatchiks to build their bunkers. They don't move there out of a love for Kiwi culture and desire to integrate with the locals. Much like their interest in Wyoming/Montana also; they see a place they like, and they go take it over and drive out/murder whoever was there before.
Gabe may be a the front man, but he's still like the benevolent dictator for life of Valve. Kind of like how Linux Torvalds is the BDFL of Linux
Earlier quoted context omitted.
I do this so that AI can only have limited GitHub permissions. It can't merge, doesn't have admin rights, etc. This after I started catching it commit directly to upstream main without PRs among other things.
that makes a lot of sense. unfortunately github doesn't allow multiple accounts per person. at least it didn't last time i checked. i hope they change their policy for AI agents though.
"You must be a human to create an Account. Accounts registered by "bots" or other automated methods are not permitted. We do permit machine accounts: A machine account is an Account set up by an individual human who accepts the Terms on behalf of the Account, provides a valid email address, and is responsible for its actions. A machine account is used exclusively for performing automated tasks. Multiple users may direct the actions of a machine account, but the owner of the Account is ultimately responsible for the machine's actions. You may maintain no more than one free machine account in addition to your free Personal Account. One person or legal entity may maintain no more than one free Account (if you choose to control a machine account as well, that's fine, but it can only be used for running a machine)."
https://docs.github.com/en/site-policy/github-terms/github-t...
Earlier quoted context omitted.
Because to decide if it's crap, you still have to read it.And because AI respect coding guidelines, you have to actually understand what the code does to detect crap. Also the sheer number is unmanageable.
Oh no, reading the code, so before AI era noone was reading the code? Unless you have automated checks done by AI to red flag AI submissions... what else can you do? Ask 100 times not to submit AI code or click a checkbox or add this really serious terms and conditions paragraph?
I'll be even more truthful about what I felt this past few months: AI emboldened bad or inexperienced devs to push a huge amount of code, way more than what they were able to, and of higher quality that they used to produce. Still, if the overseer is bad, the code won't be great (LLM are inconsistent when it comes to code quality, sometimes they find nice tricks but don't generalize them, and you have to steer them or to modify the code yourself, and sometimes they take baffling decisions, especially when the code is OOP heavy).
Those devs use to be way less visible (and I'm not saying only in the open source world, in enterprise it was the same), or grew out of their inexperience. Nowadays they produce a lot of code, often good enough, sometimes very dumb, but almost never of the quality you want to find in open source, and they are _very_ tiring to speak to because they do not understand what they have produced, and don't seem to understand plain English.
Don't get me wrong: I like the tool. It improved my life in a lot of ways, the main way it did was cure my imposter syndrome by making those less visible developers more visible. I like the fact that this tier of developers exists, especially in the corporate world. Still I think a way to filter their contribution to avoid reviewers burning out,at least for now, is necessary.
Earlier quoted context omitted.
Oh no, reading the code, so before AI era noone was reading the code? Unless you have automated checks done by AI to red flag AI submissions... what else can you do? Ask 100 times not to submit AI code or click a checkbox or add this really serious terms and conditions paragraph?
Don't be daft, the issue is the number of PR having to be reviewed, not to have to review PRs. I'll be even more truthful about what I felt this past few months: AI emboldened bad or inexperienced devs to push a huge amount of code, way more than what they were able to, and of higher quality that they used to produce. Still, if the overseer is bad, the code won't be great (LLM are inconsistent when it comes to code q…