April 9, 2026 at 1:29 PM, "Sam James" <[email protected] mailto:[email protected]?to=%22Sam%20James%22%20%3Csam%40gentoo.org%3E > wrote:
> > "Dr. Arne Babenhauserheide" <[email protected]> writes: > > > > > [email protected] mailto:[email protected] writes: > > > > > > > > April 8, 2026 at 5:14 PM, "Maximus Minter" <[email protected] > > > mailto:[email protected]?to=%22Maximus%20Minter%22%20%3Cfiddlebert2020%40gmail.com%3E > > > > wrote: > > > > > I used deepseek-V3.2 to debug some files in the GNU Mach, and I did > > … > > > > > > > > Maybe this is a dumb question, but did you vibe code this? > > > > > > Did you use AI to code this patch? > > > > > “I used deepseek” says that. > > > > But the latter > > > > some manual intervention like testing the patches and reviewing > > them. I also tested them in qemu and performed some stress tests, > > > > (specifically: "reviewing them") > > > > is not consistent with just throwing in a new file with complex without > > any explanation. Same for moving some locking around. > > > > At the current state of this patch, I consider it as not reviewable. > > > Yes, I think Samuel was rather generous in finding anything to fish out > of it that was legitimate. I'm assuming that the Hurd's current policy is not to accept AI contributed code. Opperating under that assumption, I'll add a patch to the wiki documented that we are ok with AI to debug the Hurd codebase, but NOT to write code. Thanks, Joshua > > > > Best wishes, > > Arne > > >
