Glenn Seto
宮城巴惠Krita’s Maintainer is awesome!
zivi@KitsuneofInari “as long as the contributor understands the code” understands is subjective, and they’re guaranteed to understand it less, so really its just apologism for converting codebases into unmaintainable dumpster fires
@KitsuneofInari based maintainer we need more takes like that in the space ^^
🌸 lumi-nhac 
🌱
Tock@KitsuneofInari I love you Krita! And all the developers, too!
Been backing Affinity for the longest time, since nobody has to wonder on Adobe's position on AI ("we're already vibing our source code, use Firefly!") Affinity 3.0 was the biggest disappointment.
Thank you for this.
catraxx 
@KitsuneofInari Fuck yeah!
Anna Miriamsdochter
fénnie 🌒🌔🍮🌖🌘
Li ~ Crystal System@KitsuneofInari im not sure id say that understanding of code is a boolean function of just either fully understanding everything including any and all bugs that could ever arise, and not understanding it at all, but rather a variable level of some understanding,
Halla Rempt@Li @KitsuneofInari So... Nobody can really understand code fully, not even the code they have written. Code has become too complex, and there is just too much of it!
But if you have thought about what your were writing, if you have recorded your thinking in commit messages and comments.
There is a chance you might have a recollection of having created that code when debugging it, ten years later.
If you asked Claude to regexp-slop it for you, not chance.
@Li @KitsuneofInari Also, the fun part of writing software is thinking, coding, testing, seeing people use it...
Doing code-review, not so much. That's what you do to help other people to level up.
But with LLM-generated code, where are the people you want to mentor?
But you still have to code-review the swill.
o lаvrоvsky@halla @Li @KitsuneofInari This is tantamount to saying don’t use tools, or libraries, or graphics cards, if you don’t review their code. I can’t agree with this sentiment. The real issue is the broken trust and overreach of agentic tools.
Should I ditch the Wacom and pick up a papyrus roll and feather quill instead because I don’t review the Krita source code or my tablet’s circuits? No, I trust the authors and their community.
So the question in my mind is: in what skilled hands, with which LLM, and to what extent, does it fall within the boundaries of acceptable use?
@halla @Li @KitsuneofInari I completely respect the decision of the maintainers, and think it’s better to have a clear policy than let the debate stew and simmer. Given the user base and sentiment against the GenAI topic, it’s possibly even the right move.
I just want to see an even better justification, like: we don’t add code, we tactically remove it. Our code base is a sculpture, and LLMs aren’t much help here.
Or even: as you kind of wrote above, Krita is an expression of the joy of art through the joy of coding. We aren’t feeling the vibes when you PR us your lobster soup, but feel free to make your own painting program from scratch. Call it Pincha, if you want!
@loleg @Li @KitsuneofInari You need retraining in logic, rethoric and thinking. Focus for now on recognizing false equivalences. I will mute you for a month, in the hope that you will have found sources of proper learning.
ImaLamer@kindalame yes. Instead of doing something good, like painting. Mea culpa
NullDott 🎩 (Comms Open 3/4)@loleg I don't think is a tantamount at all.
LLMs aren't a vital part of the development process, and they have never been... You can pretty much get all the supposed "benefits" from "agentic tools" without actually using "agentic tools", and you avoid all the hassle that they can bring in the long run.
Also the examples you gave aren't really equivalent?
I can trust Krita maintainers and community, I cannot trust slop machines owned by private companies.
Moses Izumi
Urwumpe@Li @KitsuneofInari I think the main issue is not the code you design and write, but that its interacting with code and users you don't fully know. Often requirements or the assumptions behind the code you change or the code you interact with are not documented. Often users don't tell you what they really want. Or don't know that a small action they don't tell you in the error report is the key to understand the bug.
Good software developers are not nerds or code monkeys, but good mentalists.
shared public horse .not bitey@KitsuneofInari I checked and here's the post ♥️ @halla
Policy on LLM code?
Speaking as the Krita maintainer, I don’t want us to accept any LLM-generated code into Krita. Some may say, as long as the contributor understands the code it’s fine, but that’s a fallacy. No programmer really fully understands the code they are working with (otherwise bugzilla would be empty); but with LLM-generated code they haven’t even put any thought into it. That code will be unmaintainable. And that’s even completely apart from the fact that using LLM’s to generate code lowers your co...
lps
Neville Park
ratfactor
Metin Seven 🎨@KitsuneofInari Nicely worded, @halla ! 👍
mohs@KitsuneofInari thx Krita!
I think many fall for a misconception here.
The code good developers write is a result of their understanding of the code that already exists and the way they intend to improve it.
If that code is generated, it is no longer a testament of somebody understanding what is going on. It becomes a "this feels like it does what I want".
JesseBot@KitsuneofInari this is lovely to hear 
Xoa Gray@KitsuneofInari @gramarye I so wish that more people felt like this.
Hacker 
@KitsuneofInari Gedit developers too!
Azarilhⓥ@KitsuneofInari Thanks for making me love Krita more.
GezThePez@KitsuneofInari is it not just another layer of abstraction?
Noa@GezThePez It is not, no.
Pēteris Krišjānis@KitsuneofInari preach it, well said.
David Fleetwood - RG Admin
Ulysses Almeida Neto@KitsuneofInari how do they detect LLM generated code?
kie/qui (planar) 🦽🛹 HLZ@ulyssesalmeida by its character
@kindalame @ulyssesalmeida do you have any suggestions for defeating racial capitalism?
@jackie @ulyssesalmeida socialism
@jackie @ulyssesalmeida (in the form of AI...)
@kindalame @ulyssesalmeida that isn't really what's being argued here, though, is it. this is about the project governance of a drawing tool.
@jackie @ulyssesalmeida i don't know, seems like performance mostly but i don't really care beyond thinking it's cute
@jackie @ulyssesalmeida there is no argument is there?
@kindalame @ulyssesalmeida in the current form AI technologies represent a total forefeit of the means of production and expansion in surveillance and policing with no regard to digital sovereignty or even the most basic public consent. AI socialism makes the mistaken assumption that the everyday person even has access to electricity and connectivity on top of their basic needs of food, water, shelter and human connection.
> the mistaken assumption that the everyday person even has access to
yes, fix those base needs
@jackie @ulyssesalmeida you don't get it - I'm not even for AI software or vibe coding ... except it *is* taking control of production in that Adobe is down 60% since the rise of vibe coding. Salesforce is down, an ICE contributor.
@jackie @ulyssesalmeida ...It should not creep into these projects, the software doesn't exist yet - AI can't write it, but they can replace proprietary software
@kindalame @ulyssesalmeida please tell me you have some other way of engaging the community
@jackie @ulyssesalmeida I know what I've been force fed, and I see LinkedIn and Twitter and Reddit, all different takes. This is one and performative and you're buying it. But sorry you're in a bubble I guess?
@kindalame @ulyssesalmeida so we agree all tech infrastructure must be re-nationalised
bluestarultor@ulyssesalmeida @KitsuneofInari Aside from them all slapping their branding on it? Poor code quality. LLMs don't know how to call existing functions; they just write new code with some stab at the same functionality. Heck, they're finding Claude Code is vibe-coded by Claude itself and it's so incapable of calling functions that it's literally re-prompting itself instead.
If that's any indication, detecting generated code is as simple as a gut check. If your guts are wringing your lunch back out, it's generated. XD
slotos@KitsuneofInari That’s a really eloquent way to describe my issue with “as long as you understand the code” argument.
That never mattered.
I won’t merge code written entirely by human just because they understand it. Because what matters is whether after an isekai accident someone new will be able to work with it.
Flux@KitsuneofInari KDE developers try not be based challenge, literally impossible.
happyborg@KitsuneofInari ...and removes copyright from the rest of the code unless it is explicitly marked as LLM generated.
snowyfox@KitsuneofInari I mean, those are pretty strong reasons
noodle@KitsuneofInari
"No programmer really fully understands the code they are working with".
put that next to
"You can and must understand computers NOW"
"No programmer really fully understands the code they are working with".
put that next to
"You can and must understand computers NOW"