Rémi Vanicat <[email protected]> writes: > Le 23 avril 2019, Lev Lamberov a écrit: > > [...] > >> Good to hear that it helped! >> >> So, 'require' loads a package if it is not already loaded. Typically any >> additional (not built-in) package needs to be loaded (by means of >> 'require' or other elisp functions). Since elpa packages from the Debian >> archive are installed system-wide we typically do not load them >> automatically, so each user in case of multi-seat environment could >> choose which packages to load in her/his Emacs. > > Glad it's fixed, but just for the record, elpa package should not have > to be required. They have their autoloads loaded automatically. >
Dropping the bug as this is a more general policy question. My first thought was similar to Remi's, that require should not be needed. On the other hand, I also see Lev's point about multi-seat environments. Is there one overall policy we can agree on, or does it depend on the package? On a related note, I also see the problem when different modes try to adjust auto-mode alist to take over an extension. I (sortof) maintain geiser (somewhat better) racket-mode, and they are not really sensibly co-installable because of auto-mode-alist conflicts. I'm not sure if this requires policy or just a technical solution. d
