From: "Theo de Raadt" Subject: Re: wscons 256 colour support To: Ingo Schwarze Cc: Mark Kettenis , tech@openbsd.org Date: Sat, 28 Jun 2025 11:45:04 -0600 Ingo Schwarze wrote: > Hello, > > Theo de Raadt wrote on Sat, Jun 28, 2025 at 11:04:29AM -0600: > > > As I see it the problem is different: > > People now run with TERM=xterm by default, > > Do they? > > I just checked what happens for me. > > When i start an xterm(1), i get: > > $ echo $TERM > xterm That is not the use case where you end up on the console with TERM=xterm. Why bring it up? > Normally, i'm using st(1) nowadays, and there, i get > > $ echo $TERM > st-256color > > At the console, i get > > $ echo $TERM > vt220 That is not the use case where you end up on the console with TERM=xterm. Again, why bring it up? > I do not recall configuring any of that by hand, so the defaults > seem sane to me. I'm not sure how "people" might end up > with TERM=xterm at the console "by default". Obviously they force it manually, because eventually they will ssh to other machines. On other machines, the xterm termcap entries are simple and complete and similar, whereas other terminal types (including vt220) are highly varied and incomplete. > > so all the worries you describe are already occuring. > > There will be misbehaviours, already. > > Only for people explicitly lying "TERM=xterm" to their console, > i guess? Yes, and people will do this. > > I said "minimal xterm", for a reason. Surely an intentional minimal > > xterm model is doable, considering people are using xterm against the > > codebase already. > > Good point, that should indeed be possible. > > I'm unsure though whether it will improve user experience. > For people already setting TERM=xterm at the console now, likely > yes, for them, it can hardly make matters worse. > > But for users relying on the default setting of TERM=vt220 now, > switching the default to TERM=xterm may possibly cause some > fallout, because that may encourage part of the software they are > running at the console to use xterm(1) features that we intentionally > will not support, possibly causing breakage, which may not occur > with TERM=vt220. Not sure how widespread or serious that will be, > maybe not very. They currently use TERM=vt220, except our code is not doing a pure vt220. We don't want a pure vt220, either.