For a while now—in the terminal(kitty)—your author has been putting up with: characters in the command line displayed with an offset; and more annoyingly, characters being randomly jumbled up during auto-completions. After some fiddling, the root cause was un-earthed: your author’s prompt contains non-ASCII characters, and zsh and his terminal have a different idea of how wide they are. This is because there is a mismatch between the encoding in the terminal and the one declared by the shell; and these 2 encodings result in different widths for certain byte sequences. This happened for your author since he has 2 different encodings: one declared by my base-system: Arch Linux; and the other defined by GNU Guix.
To remedy this conundrum, your author resorted to generating his own locales and setting them in “$HOME/.zshenv” as follows:
mkdir -p $HOME/.locale I18NPATH=$HOME/.guix-profiles/share/i18n/locales localedef -f UTF-8 -i en_US $HOME/.locale/en_US.UTF-8 LOCPATH=$HOME/.locale LC_ALL=en_US.UTF-8 date echo "export LOCPATH=\$HOME/.locale" >> $HOME/.zshenv echo "export LANG=en_US.UTF-8" >> $HOME/.zshenv
This is a common issue that faces many people running GNU Guix in a foreign distro; and as such your author hopes that this article is relevant to people with a similar issue. To read more on alternative reasons/solutions, check out this StackOverflow answer.