Re: git git<tab> completes non-existent command `git gitk`

From: Matheus Tavares Bernardino <matheus.bernardino@usp.br>
Date: May 25, 2020, 01:53 UTC
Message-ID: <CAHd-oW7b2191haSeoaBu4jvS+PBG04RKGxOhp7+rRQF+b9eErg@mail.gmail.com>
In-Reply-To: <xmqqlflgq2nr.fsf@gitster.c.googlers.com>

On Sun, May 24, 2020 at 9:27 PM Junio C Hamano <gitster@pobox.com> wrote:

Show 10 quoted lines

>
> Matheus Tavares <matheus.bernardino@usp.br> writes:
>
> > In the meantime, you could use:
> >
> >       git config completion.commands -gitk
> >
> > To locally remove the completion for "gitk".
>
> I think an equivalent of this in-core would be the way to go.

I'm not sure I understand the idea, but would that be something along these lines? Adding a "nocomplete" category in commands-list.txt, to do the opposite of the "complete" we already have[1]. Then adding a new group name to "--list-cmds" (such as "--list-cmds=completeonly"), which would include the commands that are in "complete" and exclude those in "nocomplete"? We then might be able to complete subcommands with the output of:

git --list-cmds=list-mainporcelain,others,nohelpers,alias,completeonly,config

[1]: "nocomplete" would only be relevant in conjunction with "list-mainporcelain", as the commands in the latter (as gitk) are currently completed by default (even without "complete").

> I wonder if there are any commands other than gitk that cannot be
> invoked as a subcommand of "git" potty.

In the list suggested by git-completion, I think gitk is the only one (if I haven't missed any).