git/list[1] front-page[2] threads[3] people[4] search[5] about
 

Re: [PATCH] completion: complete refs after 'git restore -s'

From
Ákos Uzonyi <uzonyi.akos@gmail.com>
Date
Sep 25, 2020, 19:59 UTC
Message-ID
<CAJrAtOWH8Upwov5dL4OfHPAb9v3NynmooC8mr3hAVA+Yo9Yi5w@mail.gmail.com>
In-Reply-To
<xmqqlfgxlq6o.fsf@gitster.c.googlers.com>
Hi,
Thanks for your detailed review.
On Fri, 25 Sep 2020 at 19:30, Junio C Hamano <gitster@pobox.com> wrote:
Show 12 quoted lines
> Ákos Uzonyi <uzonyi.akos@gmail.com> writes:
>
> > From: Uzonyi Ákos <uzonyi.akos@gmail.com>
> >
> > Currently only the long version (--source=) supports completion.
> >
> > Add completion support to the short (-s) option too.
>
> I am not too familiar with the completion library, but what makes
> the "-s" option of restore so special?  I've scanned the entire file
> and did not find that many special cases for short options that have
> their longer counterpart supported already.

There are multiple commands already having this kind of short-long option completion. The "-c" options of commit, switch and checkout each have longer counterparts, and both the short and long versions have completion support for their arguments.

Show 7 quoted lines
> I do not know if the "feature" this wants to bring in is a good
> idea---we may want to try to be more systematic (e.g. perhaps it
> involves teaching the parse-options subsystem about equivalence of
> short and long options, so that we can reuse existing support for
> the the long option "--source=<TAB>" to complete "-s <TAB>"), if we
> were to do something like this.  Singling out "-s" of "restore"
> smells not quite right, as the approach would not scale well.

I think these cases are not too frequent, so it doesn't seem to be a big scaling problem.

Show 21 quoted lines
> > Signed-off-by: Ákos Uzonyi <uzonyi.akos@gmail.com>
> > ---
> >  contrib/completion/git-completion.bash | 12 ++++++++++++
> >  1 file changed, 12 insertions(+)
> >
> > diff --git a/contrib/completion/git-completion.bash b/contrib/completion/git-completion.bash
> > index 8be4a0316e..50e6e82157 100644
> > --- a/contrib/completion/git-completion.bash
> > +++ b/contrib/completion/git-completion.bash
> > @@ -2853,6 +2853,18 @@ _git_restore ()
> >       --*)
> >               __gitcomp_builtin restore
> >               ;;
> > +     *)
> > +             local prevword prevword="${words[cword-1]}"
>
> Why duplicated prevword here?  Did you mean
>
>         local prevword=${words[cword-1]}
>
> instead?
Thanks, I'll fix it.
Show 14 quoted lines
> > +
> > +             case "$prevword" in
> > +                     -s)
> > +                             __git_complete_refs
> > +                             return
> > +                             ;;
> > +                     *)
> > +                             ;;
> > +             esac
> > +             ;;
>
> Wrong indentation.  In this file, as can be seen on the line "*)"
> you added at the top of this hunk, the case arms like "-s)" and "*)"
> must align with "case" and "esac" in this file.
Thanks, I'll fix it.

By the way, I copied this piece of code from _git_switch (it's also there in _git_checkout), so these problems have to be fixed there as well.

Also, reading _git_commit it looks that we already have a "$prev" variable, so I'll use that instead of "$prevword".

Previous: Junio C Hamano
Message 3 of 3 in “completion: complete refs after 'git restore -s'”
  1. completion: complete refs after 'git restore -s'Ákos Uzonyi, Sep 25, 2020
  2. Junio C HamanoSep 25, 2020
  3. Ákos UzonyiSep 25, 2020

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.