> From: Mattias Engdegård <mattias.engdeg...@gmail.com> > Date: Sun, 25 Aug 2024 22:08:30 +0200 > Cc: Eli Zaretskii <e...@gnu.org>, > Stefan Kangas <stefankan...@gmail.com>, > Andrea Corallo <acora...@gnu.org>, > caso...@gmail.com, > 70...@debbugs.gnu.org > > 25 aug. 2024 kl. 19.55 skrev Stefan Monnier <monn...@iro.umontreal.ca>: > > > The variable solution has one advantage: users can (setq ... t) in their > > `.emacs` to keep old code working, tho potentially at the cost of > > breaking new code. > > Actually that's a terrible solution because it affects all code, not just a > single package. In addition, it will linger in the user's system indefinitely > (.emacs being the prime bioaccumulation tissue) and potentially cause trouble > when he or she installs a perfectly working package years later. > > So my preferences are: > > First place: always unibyte, with no controlling variable. > Very close second place: always multibyte, idem. > Distant third place: contrived new function name for unibyte, like > json-serialize-encode (no, I don't like it either). > Very distant fourth place: variable controlling the string type returned.
Stefan Kangas and Andrea, can I have your opinions on this, please?
