Stefan Kangas writes: > Thierry Volpiatto writes: > >>> Do you have a recipe to reproduce this bug, starting from "emacs -Q"? >> >> I could reproduce from emacs -Q as described previously. > > I couldn't reproduce it using this recipe: > > 0. emacs -Q > 1. C-h i g > > Are there any other steps required to reproduce it? I can't reproduce anymore now. I think vanilla Emacs completing-read fails silently when an history var is not initialized (correct me if I am wrong) i.e. it behave just like if it were not defined. But, do you agree that code using completing-read with an history variable should declare this var? If not could you please explain me how such an history variable is defined, loaded etc... Here we have: (completing-read prompt #'Info-read-node-name-1 nil t nil 'Info-minibuf-history default) and Info-minibuf-history is defvared nowhere. Could you please fix it? -- Thierry