On Mon, Sep 28, 2020 at 11:43:11AM -0700, Junio C Hamano wrote: > Robert Karszniewicz <avoidr@xxxxxxxxx> writes: > > > Signed-off-by: Robert Karszniewicz <avoidr@xxxxxxxxx> > > --- > > contrib/completion/git-completion.bash | 5 ++++- > > 1 file changed, 4 insertions(+), 1 deletion(-) > > > > diff --git a/contrib/completion/git-completion.bash b/contrib/completion/git-completion.bash > > index 8be4a0316e..d98c731667 100644 > > --- a/contrib/completion/git-completion.bash > > +++ b/contrib/completion/git-completion.bash > > @@ -3016,7 +3016,10 @@ _git_stash () > > list,--*) > > __gitcomp "--name-status --oneline --patch-with-stat" > > ;; > > - show,--*|branch,--*) > > + show,--*) > > + __gitcomp "--patch-with-stat" > > + ;; > > Why is --patch-with-stat so special? > > Without completion support for "--patch" and "--stat", typing > "--<TAB>" after "git stash show" and seeing only "--patch-with-stat" > (which has been made obsolete-but-still-kept synonym immediately > after the other two were invented in 2005) would make a rather Oh, I didn't know --patch-with-stat was obsoleted. It was recently added to `stash-list`, too. I can as well use "--patch --stat". > surprising experience to the end users. For "show" alone, it may > make a lot of sense to complete "git stash show -<TAB>" and offer > "-p". Does git complete short options at all? I only see long options completed. (I'm also very new to bash-completion) > > In any case, it might make more sense to do this instead, and then > rethink what options make sense to these subcommands of "git stash". > I do not think patch-with-stat should be among them. So shall I do a v2 as per your suggestion and replace "--patch-with-stat" with "--patch --stat"? > > - list,--*) > + list,--* | show,--*) > __gitcomp "--name-status --oneline --patch-with-stat" > > Thanks. Thank you.
