Do you realize the review process does not magically go away? You can still reject code for any valid reasons?
No, but that doesn’t prevent people from thinking it should be solved (or it doesn’t exist) and the rules can be discussed, and either changed, or better understood by everyone. As far as I know, this wasn’t really done for the current rule. Let’s say it was set in an emergency in face of the risk of copyright/license tainting, and so for that reason the normal process (discussion on the mailing list or forum) was skipped or expedited quickly.
The result is that there can still be disagreement about it, as no consensus was reached.
That’s why we should have this discussion properly.
My experience with code review is that it is considerably more effort to properly review code than to write it in the first place. The activity on Gerrit seems to agree with that: lots of people submitting new changes. Few people reviewing the changes, making the fixes, getting things finetuned so they can be merged.
Doing code review on generated code is even harder. Due to the way the LLMs work, the code will always look nice at a first glance, well formatted, neatly commented, and so on. But there can be logic errors in the tiny details, and there can be sometimes a lack of understanding of the surrounding code architecture (which can lead for example to not reusing an existing function, but writing a new one to do the same thing).
These issues are not easy to see. When I review code written by a human and in good faith, I can usually tell by indirect signs where to look more closely for problems. Maybe a comment disagrees with the code. Maybe the naming for something isn’t so well chosen, which indicates that maybe the concept wasn’t fully thought through or fully understood. And when I have questions, I can ask the person who wrote the code why they did things in one way rather than another.
When reviewing code that was largely written by LLMs, this review strategy doesn’t work anymore. I have to use a method I would normally use for an adversarial code review, like if someone was intentionally trying to sneak in a vulnerability, and I’d be trying to find it. This is something I’m not good at. I trust the person who wrote the code too much. I assume they are trying to do what they say. Some of my paid job colleagues are much better than me at this. I feel so not ready for it. And to add to that, we could easily be drowned under low quality pull requests, as some other projects participating in GSoC are this and last year. I’m not sure what shielded us from that. Maybe the ban on LLM generated contributions. Maybe the fact that we’re not on Github and that was enough to stop these low-effort GSoC attempts. Anyway, I would rather keep it that way, and the proposals we got so far look better than in previous years.
Now, all of this is, of course, mainly about entirely “vibe-coded” things. There is a wide range of ways to write code from that to “I write the code directly in hexadecimal because I don’t trust the assembler/compiler” (I know some people who can do this, too).
So the question can be, where do we put the limit? I would say we can eliminate the extreme ends of the spectrum: vibe-coded slop is not allowed, and we are allowed to use a text editor and a C++ compiler. So the right answer is somewhere in between. I would say auto-completion by a traditional language server or clangd based system is allowed (that’s not LLMs, of course). But then there’s a whole range of things that may be debatable. At which point is the limit between auto-completion and vibe coding? But the answer to that very much depends on the reasons to encourage or not the use of LLMs.
If the main worry is copyright and licenses, auto-completing a few lines of code is probably OK, but an entire function is already a bit risky.
If the main worry is code quality, you probably want to keep it to a pretty low level as well. Something you can carefully review and understand. Maybe one function at a time is OK, more than that, I think the developer “writing” the code is not fully in control anymore, and you start putting more work on the person who will do the code review.
If the main worry is natural resources usage or DDoSing of servers, the right tim to stop would be before training the model at all. But, as x512 pointed out, this is currently outside of the scope of Haiku, due to the “no politics” rule. I would prefer for this to change, but there’s enough heated debate already at the moment here, so let’s keep it at the purely technical and legal arguments.
If it’s about not relying on closed source tools that may someday be unavailable, we’d have to research self hostable options (there are some). But I think this aspect wasn’t discussed as much in this thread, so I assume it is less important. In Haiku we don’t currently have a rule against use of closed source tools like Github, for example. So introducing one specifically for LLMs would be strange (although personally I would also say that would be a good idea, I’m not going to push for it at this time, same as above).
If it’s about the pleasure to work with other people as a team, help each other and learn together, rather than with machines, I don’t see why one would want to use LLMs. Maybe I’m missing something. I’d rather ask a fellow human.
And if it’s the personal pleasure and pride of having written the code all by oneself, that probably doesn’t belong to a project-wide policy.
I’m not sure if I’m missing other arguments. The thread is quite long and I’m not going to re-read it all.
10 Likes
I would say another valid reason to oppose LLM-generated code is that code should be submitted and maintained by people who understand, and are responsible for, how it works.
Being able to ask a human person “why did you implement this feature in X way” or “is there a gotcha when using Y” is incredibly valuable.
Of course, I also freely admit I despise LLMs on principle, and as a software professional, I feel it represents engineering malpractice to write code with them.
5 Likes
I hope we can find a solution that both pro- and anti-LLM camps will be satisfied with. But if not, a separate repo / depot server might be a viable option. We can even have an AltPorts Depot published on the HaikuDepot, provided the app will be coded without LLM 
I’d frankly would not want to trust an operating system that’d use slop code extensively (Stability and the ability to understand what the code is doing is absolutely crucial there imho and slop code undermines that usually) and those companies that like to push LLMs (and similar generative models) also tend to be those that also support policies that are directly aimed against the right to exist for the likes of me and my friends. I thus am very much opposed to any use of such technology for now (apart from that, a lot of stuff that “AI” could “solve” could be better solved by already existing solutions).
Probably. But survivability will get worse. Climate change greatly increases the risk of zoonotic diseases, droughts and heat waves. I’d personally would prefer a world where all of that is kept at a minimum. FOSS software is dependant on (international) collaboration which is (to put it mildly) hindered by (resource shortage) conflicts and helped by contributors having a safe surrounding and lifelyhood, both things that are directly endangered by climate change.
- If it’d be questionable, would that mean that we’d not want to at least try and do to the best of our abilities?
- For germany at least, there is a study comissioned by FridaysForFuture Germany that acually showed the 1.5°C target to be reachable. Granted, this study is now 5 years old, but should at least hold up for the 2°C target
The name of the study is:
Wuppertal Institut (2020). CO2-neutral bis 2035: Eckpunkte eines deutschen Bei-
trags zur Einhaltung der 1,5-°C-Grenze. Bericht. Wuppertal.
Whilst Haiku might not be the UN, it still is a quite big software project (It’s a complete operating system after all). As software engineering is a type of engineering, it bears the burden of enabling or shaping the life of people and society, thus having political power. And an operating system can make quite grave politcal decisions: Windows 11 is quite infamous for not wanting to run on computers that are even a bit older than 2018. With the widespread adoption of Windows, MS can thus force both private persons and buisnesses to do (especially for some people) quite heavty investments to renew their equipment and generated tons of just avoidable e-waste. This can be seen as a political decision by MS. And whilst we probably don’t even have a fraction of the power that Microsoft does, we do still have political power and I think that it’s important to recognize that and confront problems that the project itself can influence.
kind regards,
zeldakatze
5 Likes
Alright, so let’s meet for lunch today via Telegram video call—strictly using Haiku! Get your webcam ready because I’m really curious to see how well it works! 
Sometimes the truth hurts and it’s hard to face it. 
I’m away for work and unfortunately don’t have much free time, but I’d love to hop on a 10-minute video call; you’d be only the second person I know with a working USB webcam 
I’m glad you were wrong about me.
What time can you confirm for our slot?
1 Like
Oh fine,continue your trolling,I’m sure that will make you more friends here 
This forum was really a nice and friendly place before you opened the can of worms with your “tutorial” here that insults the biggest part of the Haiku community.
Probably the 1-day ban just wasn’t long enough.
2 Likes
I’m not trying to troll anyone. If your webcam is working, hop on too and I swear I’ll apologize to you all in person. 
Regarding the ban, I’m waiting for a proper apology since it wasn’t justified it was just a deliberate choice by one single person.
1 Like
that’s so called irony or sarcasm, if he’s not right, can you point me a working webcam in haiku cuz i found no one? it’s been hard even to find an usb sound card to have audio (bought several but not working until I get a cheap 1.1)
I was really trying to stay out of this toxic discussion as much as possible,but your rude behavior makes it almost impossible for me to ignore.
To also add something relevant to this post,here’s my stance when it (hopefully soon) comes to a vote about banning slop contributions:
- Completely ban it from contributions to Haiku itself.
- Completely ban it from contributions to the Haikuports repository
- It’s probably impossible to remove all ports where the upstream uses AI from Haikuports,but at least clearly label software where the use is known and substantial
- Completely ban it from use in forum posts as well (stating that it’s not helpful is a good first step,but some people seem to not care that much about being helpful)
- Also ban announcements of projects substantially based on AI from the forum,this place doesn’t need to be and shouldn’t become a stage for showing your latest slop,and you can do that somewhere else.
4 Likes
there are a lot of rude people here and I never witnessed a similar ostracism, is this ad personam? you dont like him so you want to kick him out?
1 Like
My extremely heavy, nasty, and rude post which deserves to be hidden is simply asking to connect via webcam.
1 Like
That’s really not a personal thing at all.
I’ve read his DesktopOnFire blog since the beginning and like most of it.
I also found Sestriere quite interesting before I realized that it’s 100% slop.
The initial post neglecting all the work that has been done on Haiku so far and calling things that have been available for years unavailable despite knowing better feels like a personal attack on me as a contributor who has put many hours into improving Haiku and fixing numerous bugs in my valuable spare time,however.
Yes,I call that behavior rude,and also the permanent jokes on unfinished functionality.
It’s true,most camera drivers don’t work yet,this is a widely known thing that some people (also myself included,I think I put in a weekend or so) tried to fix without a solution yet.
Feel free to use Windows if that makes Haiku so unbearable for you.
Personally,I don’t even need a webcam but anyway I put in the work hours knowing that many others would be happy if it worked.
5 Likes
Abahhahahahahaha 
Thanks a lot, man! Since we’re talking about being constructive, let’s hear it: what sucks about it?
1 Like
Sure, no problem. You just need the right WebCam.
And you also need to consider the tone you are using in your messages is not appropriate and disrespectful.
4 Likes
At this point, would an Apple II work for you anyway? 
1 Like
Taking a deep breath here and with the utmost respect for every virtual friend and soul in this group, considering we’ve all been part of this journey for over 25 years. I’d like to share a few points:
- Point 1 The work done on porting is beyond dispute. The quality and the immense value it has brought to the community is unquestionable.
- Point 2 I understand that some views can be emotional rather than rational. However, claiming that USB webcam support is limited to a single model via IP streaming is, frankly, disappointing to hear from you.
- Point 3 In a democratic and rational environment, “branding” or gatekeeping third-party software, regardless of how it’s used, is something I simply cannot accept.
- Point 4 When someone in 2026 dismisses a technology in such a harsh, absolute way without even trying it, relying only on a few blog posts, that shouldn’t be taken as the absolute truth, especially compared to those who are actually testing and analyzing these new tools.
- Point 5 It is disheartening to realize that to use the OS we love, I might have to build a personal fork that will be nearly impossible to maintain. All this because a standard $9.99 USB webcam isn’t supported, and when a driver is modified thanks to “Pippo” [or the specific tool name], the official response is to set up a Wi-Fi network for an IP camera. That is just not rational.
- Point 7 I’ve likely missed something in this whirlwind of extremist messages, none of which are mine!
Now, back to work, someone has to save the planet!
2 Likes
I am considering that, yes. I have a lot of fun with 8 bit computers and no people telling me that I should use LLMs to write code for me.
If my contributions here are not welcome and getting in the way of the robots, no problem, I can go do something else for a while, interesting projects are not lacking.
I don’t understand why you (our your LLM) brings this point here. How is it related to the current discussion?
I did not claim that. I gave one example of how you can use a webcam on Haiku. Any IP webcam will work, since the protocol is standard. USB ones currently don’t. Eventually someone will fix the driver, it looks like this is just not a very interesting feature to any of the current devs. What are you trying to achieve, get me to do it out of spite? You are doing quite the opposite, making it not want to touch that part.
I don’t even understand what you mean. I guess it relates to the suggestion to label software that is developped with LLMs, or maybe software that is entirely “vibe coded”. As I have explained before, this is similar to printing the ingredients list on food products. It does not put any judgement of value on the software. It just lets people know, and make their own personal choices.
You asked for a way to use a webcam. I show you one. It is just one way, and it is not official.
It replies to your question. If you are not happy with the answer, maybe your question was badly formulated.
I expect your apologies because I think I answered to your challenge and shown that it is indeed possible to use a webcam for a video chat in Haiku. But you don’t hold to your promise and change the goalpost. Be more careful next time you set up a challenge, because humans can implement creative and unexpected solutions
5 Likes
Sorry, even if I was comfortable making video calls (I’m not and avoid them like the plague), I’m not wasting my time indulging a so-called “engineer” that says 3 is equal to 4524.
[REDACTED because apparently only the OP is free to insult others in here]
(but that’s okay, I went a bit too far with it afterall)
It’s very telling that you are unwilling to apologize for being such a massive a**hole to the porting community. You should be grateful people are willing to use their free time to port stuff to an obscure platform, you spat on each of our faces instead. Shame on you.
2 Likes
There is an UVC webcam driver which already partially works: External USB-Webcam experiment with UVC media add-on (I did get it working with some webcams though it still is unstable iirc). You could work from there and I doubt that you’d need a LLM or similar to do that. But having basic support for webcams is not an emotional argument. I already saw some that actually worked.
ehm:
for me this frankly sounds like a form of free speech absolutism. There is nothing stopping you from acquiring the source code, building and running programs that have been entirely vibe-coded. However, the probability for such programs to be badly written and thus in the worst case actually be dangerous (C++ is not a memory-safe language and an invalid array index can already pose a severe security risk) is pretty high. It’s totally valid to not want to endorse such programs by putting them on a repository or to include slop code in the main system, which could be a development and security nightmare.
I had the fun excercise to fix up code for a project that certainly was auto-generated for a project. Since it was a small project and memory-safety was at least enforced by the language (rust), it was manageable. However all of this could have been circumvented if the person that previously wrote the stuff would have actually sat down and had done some even basic designing. I’d prefer not having to see this in a bigger project. I already saw more than enough C programs that were done with the “aid” of LLMs and trying to debug that was a nightmare. I also recently had a fun time trying to fix a FDM 3D printer config that was slop-generated by someone after a sd-card had died. Eventually I just gave up and did everything from scratch myself which lead to much better quicker and actual results. (And I’m pretty clueless about FDM 3D printers, I should perhaps add)
Then do that. But maintaining slop code is pretty difficult too and rejecting code whose intentions are unclear as there is no person that could later be asked for the rationale why a thing was done in this or that way and with unclear licensing problems and that was generated with models that break the basic morals of FOSS software (i.e. not (being able to) attribute code to it’s original source and being reliant on having huge proprietary datasets that contain data of which the original authors did not give permission to) does sound quite rational to me.
The big models also usually don’t make a profit. OpenAI does (to my knowledge) not make any profit and is infact loosing a LOT of money. The only reason I could see for such a buisness model would be trying to make people dependant on their services and later ramping up the pricing, which would force people to pay these prices. Making yourself dependant on that does not sound like a smart idea to me and I’d rather try to solve problems myself than to make myself dependant on a model that might soon be inaccessible to me for economical reasons.
8 Likes