v0.0.14 now uses libcurl and works nicely :)

         (file-port (open-input-file the-file))
         (handle (curl-easy-init))
         (_ (begin
              (curl-easy-setopt handle
                                'url
                                (format #f "smtp://~a:~a" access-server
                                        access-port))
              (curl-easy-setopt handle
                                'verbose #t)
              (curl-easy-setopt handle
                                'use-ssl 1)
              (curl-easy-setopt handle
                                'username access-key)
              (curl-easy-setopt handle
                                'password access-secret)
              (curl-easy-setopt handle
                                'mail-from from-address)
              (curl-easy-setopt handle
                                'mail-rcpt
                                (list to-address))
              (curl-easy-setopt handle
                                'readdata file-port)
              (curl-easy-setopt handle
                                'upload #t)))
         (r (curl-easy-perform handle #t))
         (rr (catch #t
                    (lambda ()
                      (bytevector->string r "utf-8"))
                    (lambda (key . args)
                      r))))

i understand the mixed feelings about AI, but I think we should stop and think that we sometimes have nice and fun use cases, like image generation

Sorry you feel this way ! my code is completely artisanal you might say, I only use AI for code comments and docstrings that is all. in that regard it is useful as a tool. i understand the world has now very mixed feelings about it, but it’s just another tool for us digital carpenters.

it’s quite a good idea actually to have inter-instance communication! i will have to put more thought into it, but definitely added to the TODOs

the like system is a sort of trust score, kinda like stars in ultimate guitar, so the community can downvote entries and that will help filter out bad content

I think it is certainly possible to do , but I think it is not really something interesting for most people. I therefore think that it is easier to handle it all on wikimusic side, also considering permissions and content per instance, etc