Hi,

On 2026-11-01 at 18:36+01:00, Ludovic Courtès wrote:
> Without a clear message that LLM-produced pull requests
> are not accepted at all,

I've merged https://codeberg.org/guix/guix/pulls/12143
to broadcast this message.

On 2026-11-01 at 18:36+01:00, Ludovic Courtès wrote:
> enabling the weakening and rejection of GCD 008
> is what brought us here.

This can't be right, as GCD 8 started with allowing
any contribution under 15 LoC, then modified (and IMHO strengthened)
to any non-creative contribution incorporating LLM output,
but the current policy disallows all LLM output.

I would get if if by _weakening_, you referred to the removal
of the following clause

On 2026-05-15 at 17:19+02:00, Ludovic Courtès wrote:
> Software where the majority of commits were authored
> or co-authored by genAI **will not be packaged in Guix**.

but this does not seem to be what is discussed in this thread.

On 2026-11-01 at 18:36+01:00, Ludovic Courtès wrote:
> On 2026-11-01 at 12:51+09:00, Maxim Cournoyer wrote:
> > An honest, direct question to the author and a gentle approach would
> > seem best to me.  Of course if the same person keeps disregarding the
> > rule, there can be an escalation, ultimately a ban.
>
> The problem is that it creates a toll on human reviewers.
> [...] volunteer time will be wasted kindly asking the author
> whether they’re a robot—kinda like Cloudflare’s challenge,
> but made by humans.
>
> Volunteers ought to be entitled to just close pull requests that look
> LLM-generated.  The onus should be on the submitter.
>
> This is a bad vibe, but the volunteers dealing with this are not the
> ones to blame;

I strongly request the retraction of these statements,
especially when they come from the project's founder and one
of very few developers of the critical parts of the Guix.
Per my understanding, they are normalizing throwing insults
at contributors if they did not submit a LLM-generated patch.

Code of conduct aside, I think it's basic decency to communicate
with each other like human beings.  How do you feel I go around
accusing your patches of being generated by LLM without even
talk to you?

FWIW, both before and after the ban, I've raised my concern on patches
and in every case the contribution was just kindly withdraw.
I do not see any reason to assume bad faith, especially when reviewers
are in a position of power.

On the other hand, it disheartened me to see our infrastructures
are slowly depending on LLM-generated software.

Kind regards,
Phong

Attachment: signature.asc
Description: PGP signature

Reply via email to