> You have to transfer terminfo to all ssh servers because the kitty term does not exist there.
terminfo made sense decades ago when most people were connecting to 1 big centralised mainframe from a mix of random terminals. Now we have people using one (software) terminal to connect to many servers big and small. Requiring terminfo to be updated on the remote side is just nuts.
Kitty should just identify as xterm like every other terminal written in the last 20 years.
> Kitty should just identify as xterm like every other terminal written in the last 20 years.
This is not my experience at all. I'd say most of terminals, including modern ones, have unique terminfos and work miserably when it's not installed. Examples include ST, kitty, alacritty. On top of that we have terminal multiplexers like tmux and screen which provide their unique terminfos as well.
I find myself in between. I've seen many terminfo's set to other things, which don't work, or have to be set every time I'm connecting to a new server which is too much extra work for every new server in my opinion. So I always just set it to xterm-256color and forget it
I use rxvt-unicode which sets $TERM to "rxvt-unicode-256color," which is not recognized by a number of server to which I have to connect to from time to time.
There's a whole story about that, part of which is that at the time terminfo had to cater for old systems where there's a limit on the length of a terminal type name. But the same applies as for a missing kitty terminfo entry. The terminfo database is designed so that one can copy records around, and one can put them in a directory in one's home directory if one does not have administrative access to update the system database.
Debian systems have had an rxvt-unicode-256color entry for approaching 10 years, now. FreeBSD for about the same.
> Kitty should just identify as xterm like every other terminal written in the last 20 years.
You can do that by adding "term xterm-256color" to kitty.conf. The manual page warns against this but it seems to be mostly compatible. Kitty-specific extensions probably won't function remotely but it still serves as a nice portable terminal which can be configured with dotfiles.
The fact that the terminal is identified based on a TERM environment variable made sense if you want to support many hardware terminals which can't otherwise identify themselves and predate terminfo.
Maybe in the modern world applications could use an escape sequence to check for extensions on top of standard xterm?
terminfo made sense decades ago when most people were connecting to 1 big centralised mainframe from a mix of random terminals. Now we have people using one (software) terminal to connect to many servers big and small. Requiring terminfo to be updated on the remote side is just nuts.
Kitty should just identify as xterm like every other terminal written in the last 20 years.