Hi, On Mon, 2026-11-09 at 08:40 +0100, Jonas Smedegaard wrote: > Quoting [email protected]Â (2026-11-09 07:46:26) > > On Nov 9, 2026 05:09, Maytham Alsudany <[email protected]> wrote: > > > The new system is clearly a "large-scale or automated" project action. > > > > It is not clear to me. The fact an LLM review is made on each upload is an > > automation. But it isn't making any decision, and it is optional to read > > the output. > > So you are saying that it *is* clear to you that it is automation - > what is unclear to you is a) whether or not that automation involves a > decision, and b) whether or not that automation requires certain action > from the Debian contributor who triggered it. > > Do I understand you correctly on that? > > Also: what, in your interpretation, is the threshold of "large-scale", > if beyond a project action applied to every package entering NEW queue?
The same as for mass-bug filings. Do you think NEW processing needs the same action as mass-bug filings with a prior discussion on -devel@ that NEW processing will take place? If yes, then that would be required no matter whether a LLM is involved or not. If not, then the use of the LLM doesn't matter. I think an explicit discussion as for mass-bug filings is not needed for NEW processing and the team responsible for some area gets to decide how they handle their area of responsibility as usual. Of course some bitter rearguard battle enthusiasts want to kick them out of the project for that in their first mail, but that is just business as usual. This will probably continue for at least a decade (see: systemd). Ansgar
