Hi Oleg,
On 10/12/26 21:01, Sharlatan Hellseher wrote:
Thank you for your calm replay, I should practice more smooth writing
style ;-)
I'm not sure whether that is sarcasm. In case it is, my apologies. I
feel like we are starting to chase people away, and I'm getting a bit
frustrated with that. As a result, I sometimes fail to convey my
enthusiasm.
Also there is a higher order Codeberg policy which
may affect any LLM generated code to be dropped immediately [1]
That is not at all what the codeberg policy says. Before your quote it
says:
If your work fits into these cases, it is unlikely that you are affected at all:
- Projects who have an active community that cares about and maintains the
software
- Projects with a significant pre-LLM history
- Maintainers who unknowingly or willingly accepted LLM-generated contributions
from other contributors, if your project otherwise does not involve the heavy
use of LLMs
We fit all three. It is therefore totally fine for codeberg if there is
the occasional LLM-generated code. Don't hide behind codeberg; please
stand for what you believe yourself.
That is a very strange conclusion. We got to meet people where they
are somewhat and try to get them to become a contributor.
As [2] is already part of contributing guidance - how we need to
implement the following statemen during day-to-day review process (?):
We do not accept contributions that were written by, or based on code
written by, a @acronym{LLM, Large Language Model} coding assistant. This
is due to the uncertain copyright status of the outputs generated by an
LLM, which could be incompatible with and undermine the
@url{https://www.gnu.org/licenses/gpl-3.0.html, license} used by the GNU
Guix project.
Yeah, something like that, give them a chance to rewrite the code. Or
help them with that, as the general sentiment seems to be that we want
prospective contributors to ask us humans for help instead of relying on
an LLM. We need to be proactive in showing how we want people to behave.
And please ask why people feel the need to use an LLM, so we know what
we need to improve. Like the other week when someone used an LLM to
help with understanding the Guile peg-parser. That means we could
document it better.
People who use LLM's try to make the project better, so we need to have
constructive responses that hopefully transform such drive-by P.R.
authors into regular contributors.
Best wishes,
Hugo
