Oleg Pykhalov writes: > Marius Bakke writes: > >> Oleg Pykhalov writes: >> >>> * etc/completion/bash/guix (_guix_is_dash_L): Replace with >>> ‘_guix_is_dash_L_or_m’. >>> (_guix_complete): Add this. >> >> [...] >> >>> diff --git a/etc/completion/bash/guix b/etc/completion/bash/guix >>> index 6b84193db..286c4f1c5 100644 >>> --- a/etc/completion/bash/guix >>> +++ b/etc/completion/bash/guix >>> @@ -97,9 +97,10 @@ _guix_is_removing () >>> $result >>> } >>> >>> -_guix_is_dash_L () >>> +_guix_is_dash_L_or_m () >>> { >>> - [ "${COMP_WORDS[$COMP_CWORD - 1]}" = "-L" ] \ >>> + comp_words_minus_1="${COMP_WORDS[$COMP_CWORD - 1]}" >>> + [ "$comp_words_minus_1" = "-L" ] || [ "$comp_words_minus_1" = "-m" ] \ >>> || { case "${COMP_WORDS[$COMP_CWORD]}" in >>> --load-path=*) true;; >>> *) false;; >> >> I think this fails to account for the long form "--manifest". Perhaps >> it's better to add a new "_guix_is_dash_m" and use >> >> "if _guix_is_dash_L || _guix_is_dash_m" >> >> below. WDYT? > > Nice catch! Thanks. > > But I don't know how to implement a file completion for ‘--manifest=’. > ‘--load-path=’ doesn't have a file completion too. Only completion for > equal sign before. Oh, OK. That's another issue I guess. The patch below LGTM, but the commit message needs adjusting ;-) Thank you!