CONTRIBUTING.md: clarify AI usage in commit trailers - #166
Conversation
|
|
||
| Use your real name (sorry, no pseudonyms or anonymous contributions.) | ||
|
|
||
| AI tools, services, or assistants may be used to assist with contributions, but |
There was a problem hiding this comment.
Co-authored-by: AI seems fine as long as Signed-off-by is added by a human?
There was a problem hiding this comment.
It's hard to tell that during review 😂. Contributor can add them in description
There was a problem hiding this comment.
I'd like us to encourage an Assisted-by trailer. I think that's better than Co-authored-by and better than just adding into the PR description.
There was a problem hiding this comment.
FYI k8s community forbids AI co-authored-by/assisted-by etc.
https://github.com/kubernetes/community/blob/main/contributors/guide/pull-requests.md#ai-guidance
"Listing AI tooling as a co-author, co-signing commits using an AI tool, or using the assisted-by, co-developed or similar commit trailer is not allowed.":
There was a problem hiding this comment.
It does. I disagree with that conclusion and it is not a CNCF-wide policy so we are not bound by it.
There was a problem hiding this comment.
I also disagree with that conclusion.
CNCF is our parent foundation, yes. However kubernetes arguably being it's biggest project, I figured it needed referencing, for completion, here in this PR.
Made the argument for assisted by at kubecon/cncf con when this was presented on maintainer day. Response was no AI signatures of any kind in the commit message, but you can state assisted by, etc, in the PR.
Also recognizing that this allowance of the assisted-by AI commit trailer, is not a MUST. So I've been using PR text for that.
Also avoiding extra end of file AI assisted comments in code.
fuweid
changed the title
|
ping @containerd/committers |
| Use your real name (sorry, no pseudonyms or anonymous contributions.) | ||
|
|
||
| AI tools, services, or assistants may be used to assist with contributions, but | ||
| do not add them to `Signed-off-by` or `Co-authored-by` lines. These commit |
There was a problem hiding this comment.
Can we add Co-Authored-By: Claude * <noreply@anthropic.com> to the allowlist?
Too useful to reject by default.
There was a problem hiding this comment.
The Change is to avoid them. So, I would like to say no
There was a problem hiding this comment.
The purpose of the human authored by records is for humans to agree to the apache2 license for the commit diff, and we're not supposed to allow anonymous.
There was a problem hiding this comment.
The purpose of the human authored by records is for humans to agree to the apache2 license for the commit diff, and we're not supposed to allow anonymous.
Isn't it covered by Signed-off-by: <HUMAN>?
There was a problem hiding this comment.
Signed-off-by: <HUMAN> covers the certification. To me, Co-authored-by usually represents another human author, and I don't think an AI tool should be treated as a co-author there. If it does, we probably can add vim/awk/sed tool as co-author too. It doesn't make senses to me
There was a problem hiding this comment.
this ^.. while I understand the reporting honesty of co-authored by attribution to a tool... The co-authored by is bound, legally, to the license requirement, warrant that the code is theirs etc.. not copied from a GPL license.
There was a problem hiding this comment.
The co-authored by is bound, legally, to the license requirement
I don't believe any of us are lawyers. If you are concerned about this from a legal perspective, I think we can ask CNCF for their advice. But I don't expect the co-authored-by trailer has any specific weight in a dispute (at least in the context of the US), especially if it's a known LLM tool rather than a human.
660851f to
b7f0d11
Compare
Signed-off-by: Wei Fu <fuweid89@gmail.com>
b7f0d11 to
63315a5
Compare
|
kindly ping @dmcgowan @AkihiroSuda @dims @kzys @kiashok any objection? |
9 participants
containerd/containerd#13483 (comment)