Re: [NonGNU] new: hermes
Philip Kaludercic <[email protected]> Sat, 08 Aug 2026 17:59:43 +0000
| Newsgroups | gmane.emacs.devel |
|---|---|
| Message-ID | <[email protected]> |
Jean Louis <[email protected]> writes: > On 2026-08-08 14:00, Philip Kaludercic wrote: > >> The proposal was always to add the package to NonGNU ELPA, see the >> subject line. The question here, and with the condex package is just >> that this is not SaaSS situation, which from Eli's responses I >> understand is not the case. > > My opinion is related to GNU free software principles. I do not know > what is "condex". "Condex" is a typo of "Codex", see https://lists.gnu.org/archive/html/emacs-devel/2026-08/msg00178.html. > It is true that Hermes agent can be used with local models. It is also > true and undisputed fact that users of Hermes agent are from first > steps driven to non-free SaaSS LLMs. > > Package like that takes away users' freedom, doesn't teach them > nothing about GNU principles, it in fact acknowledges proprietary LLM > APIs (SaaSS LLMs). I don't think it has ever been a rule that we require all packages to only use software that acknowledges GNU principles. I agree that it would be better if it did, but our requirement is that the software is fully functional on a system that only uses free software. > Why not be frank and truthful?! > > We are not even discussing here how Emacs can be used with local models. > > Yet, we are introducing packages that support proprietary SaaSS > LLMs. Fine, you say there is 1 chance among hundreds, to use the local > model. But realistically and objectively, so far on this mailing list > I did not hear others really doing it so. > > It is very easy to deflect my opinion and say "hey, but there is way > to use it locally". But practically, do you? This is not deflecting, it is explaining why your objection is not a valid reason to reject the package. > Does Thanos?