Using readline-8.3 (with bash-5.3_p3 or bash-5.2_p37) there seems to be a history-search-backward regression inserting stray partial lines into the history. It does not occur with readline-8.2_p13 (with bash-5.2_p37). Other functions (like history-search-forward) might also trigger this.
Using history-search-backward (PgUp) on a partial line will commit that partial line to the .bash_history file. Example: On a system with readline-8.3, e.g. Debian Sid (tested on debian-sid-genericcloud-amd64-daily-20251120-2302) Start with a .bash_history containing only echo hello echo world And /etc/inputrc containing (at least) "\e[5~": history-search-backward Open a shell Up Up Enter -> echo hello Up Esc-Left Ctrl-K "w" PgUp Enter (or) Up Esc-BkSp "w" PgUp Enter -> echo world Exit (Ctrl-D) But now there is a truncated "echo w" entry in .bash_history (possibly replacing "echo hello“, I guess) $ cat .bash_history echo hello echo world echo w echo world I found the latest 8.2 (readline-8.2_p13) to be unaffected and all of 8.3 affected, tested were - https://ftp.gnu.org/pub/gnu/readline/readline-8.3-alpha.tar.gz - https://ftp.gnu.org/pub/gnu/readline/readline-8.3-rc2.tar.gz - https://ftp.gnu.org/pub/gnu/readline/readline-8.3.tar.gz - http://git.savannah.gnu.org/cgit/readline.git/snapshot/readline-master.tar.gz (8.3_p1) - http://git.savannah.gnu.org/cgit/readline.git/snapshot/readline-devel.tar.gz I also did a partial install by placing just libreadline.so.8.3 in a 8.2 install, which reproduces the bug. How can I investigate this further? Are there incremental change sets between 8.2 and 8.3 to bisect? Thanks, Christian
