Vim vs nano: which editor to use on a server

Open a file with nano and it behaves like most editors: the cursor is in the text and keys type. The bottom two lines list the most useful commands, where ^ means Ctrl and M- means Alt (Meta), so ^X Exit means Ctrl-x to leave. If you have changed the file, nano asks whether to save it on the way out. The nano manual calls it "a small and friendly text editor", and it started life as a free clone of Pico, the editor from the Pine mail client.

Open the same file with vi or vim and you are in Normal mode, where letters are commands rather than text. That is the source of the "how do I exit Vim" joke: typing a sentence runs a string of commands instead. Press Esc to be sure you are in Normal mode, then :q! and ↵ to quit without saving, or :wq to save and quit. The full escape guide is how to exit Vim.

vi is specified by POSIX (in its User Portability Utilities option) and has been part of Unix systems for decades. In practice you will find a vi on almost any Linux server, BSD or macOS machine; on Debian and Ubuntu it is often a small build called vim.tiny, and on BusyBox-based systems it is BusyBox's own vi, which covers the basics.

nano is installed on many Linux distributions and is easy to add, but it is not guaranteed. Very small container images may have neither editor, which is one reason it is worth knowing the vi basics: they work in whichever vi you end up in.

nano is a good choice when the edit is small and rare: changing one line in /etc/hosts, uncommenting an option in a config file, fixing a typo in a cron entry. Nobody on your team needs to have learned anything to do it safely, and the shortcuts on screen make mistakes unlikely. It is also a sensible default for shared servers where people of mixed experience log in.

Two options worth knowing:

• nano -v file opens a file read-only, for looking without risk.
• nano --modernbindings swaps in the keys most desktop programs use: Ctrl-s to save, Ctrl-q to quit, Ctrl-z to undo, Ctrl-c and Ctrl-v to copy and paste.

Vim earns its learning time when you edit often, edit large files, or make the same change many times. A few examples of what takes one command in Vim:

• :%s/old/new/g replaces every match in the file (find and replace).
• :g/DEBUG/d deletes every line that mentions DEBUG.
• ci" replaces the text inside the quotes the cursor is in, and . repeats that change wherever you go next.
• gg, G and :120 jump to the top, the bottom or line 120 of a long log.

You don't need all of Vim to be comfortable on a server. A survival set of i, Esc, :wq, :q!, dd, u, / and n covers most one-off edits, and the vi cheat sheet lists the keys that work even in a minimal vi.

Programs that open an editor for you (git commit, crontab -e and many others) look at environment variables. Set both in your shell's startup file, for example ~/.bashrc or ~/.zshrc:

• export EDITOR=nano (or vim)
• export VISUAL=nano

For Git alone, git config --global core.editor nano sets it. Git checks GIT_EDITOR, then core.editor, then VISUAL, then EDITOR, and falls back to the default it was built with, which is usually vi. That fallback is why a first git commit so often lands people in Vim.

On Debian and Ubuntu there are two more layers:

• select-editor asks which installed editor you want and stores your choice in ~/.selected_editor, used by sensible-editor. VISUAL and EDITOR still win over it.
• sudo update-alternatives --config editor picks the system-wide editor command, the fallback for every user who hasn't chosen.