Hi Guix,
On 2/5/26 08:12, Nguyễn Gia Phong wrote:
Hello Guix,
Let's continue the discussion
over https://codeberg.org/guix/guix/issues/6087 here.
For context (there are also comments not quoted in this message,
please also visit the issue),
For me it is good to have some experience before forming an opinion, and
I'd like to share this small experiment with you.
I asked Codex four times to fix "a random failing package", and it
succeeded four out of four times.
- qdl:
https://codeberg.org/hugobuddel/guix-mirror/commit/8e7277063f48a951deb2e62c638da703f1c62b1d
- perl-text-iconv:
https://codeberg.org/hugobuddel/guix-mirror/commit/d61b4914be636c283bec2f65f51ca42dcbf7106f
- go-github-com-matttproud-golang-protobuf-extensions-v2:
https://codeberg.org/hugobuddel/guix-mirror/commit/04901453566e4af57305e87d9360955e5cdbc6bb
- audmes:
https://codeberg.org/hugobuddel/guix-mirror/commit/434384988392f117608592b99e4ebbf7dda1b7d8
I think it did a pretty good job. It only took a few minutes per
package, and that was mostly due to build times and because it had to
ask permission for everything.
Only for qdl I manually changed the code, because the URL was broken. I
did not change any other code, only some commit messages.
I'm not going to make P.R.s for these, as any of you could have done
exactly what I have done, and I don't care for these particular packages
(as they were selected at random by Codex).
We do need a policy for LLM work, because apparently we can fix a large
fraction of our failing packages entirely automatically. To me, it
doesn't make sense to either keep them broken, or to spend human-time
fixing these manually.
And to stay on topic, these changes seem to be fine copyright wise, but
IANAL. The commits are very 'factual' and short.
Best wishes,
Hugo
P.S. This email was written entirely by hand.
