Am Dienstag, dem 17.03.2026 um 18:37 +0800 schrieb Luca Toniolo: > Yes halscope I had help from Claude, also TP2 work, mostly with > Claude. TP1 was not. I don't think is feasible to keep AI out of the > code base completely, but for the final TP2 work will make sure it is > tested, it's more of inspiration for future work, this was also > discussed a couple of meetings ago, I didn't really get any pushback > at that time, but Andy was not present... >
I don't think it is much of a problem given a human being reviewed the code. The halscope stuff was rather "trivial" and more or less obvious in the sense that there is not much leeway how to implement these changes / features, implementations of the same feature wouldn't look too different IMO. It could be argued that such a change (essentially changing a few numbers) is not copyrightable at all because it lacks the necessary level of originality. the TP is a completely different thing though,I expect the new TP will be subjected to very close inspection and thorough testing / vetting while in the spent fuel pool to cool down before it is included as default in a new linuxcnc version. the copyright situation isn't really settled. I don't think definite statements are possible one way or another. in case of linuxcnc, it is additionally complicated because the base of the code is public domain as I understand, and I wouldn't be too sure if it is even possible to "relicense" that under GPL. to be on the safe side without impacting new developments, IMO we should clearly identify stuff that has relevant code generated by AI agents and keep it somewhat separate or at least separable. IMO refactorings, merely renaming / reorganizing stuff or transmutating code e.g. from snprintf to std::format with the help of AI agents is no problem. here is some stuff from the FSF https://www.fsf.org/licensing/copilot/ that also doesn't look like they have a coherent opinion. And it is probably completely outdated, papers are from 2022. discussion from last october: https://lwn.net/Articles/1040888/ Executive summary: "Anybody hoping to exit the session with clear answers about the status of LLM-created code was bound to be disappointed; the FSF, too, is trying to figure out what this landscape looks like." IANAL so caveat emptor -- Robert Schöftner <[email protected]> _______________________________________________ Emc-developers mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/emc-developers
