From: Eric Sunshine <sunshine@sunshineco.com>
Date: Mon, 14 May 2018 03:33:07 GMT
Subject: Re: [PATCH/RFC] completion: complete all possible -no-<options>
Message-ID: <CAPig+cRUD7FnJqSZ=hkz0GAgOROiY+gvHKRjKVfkKtqDZyt5mA@mail.gmail.com>
In-Reply-To: <CACsJy8BGs7EOYFKayL-bgvEbKOJiROF52o3SneLyG9Nm6nUngA@mail.gmail.com>

On Tue, May 8, 2018 at 11:24 AM, Duy Nguyen <pclouds@gmail.com> wrote:
> On Mon, Apr 23, 2018 at 7:36 AM, Eric Sunshine <sunshine@sunshineco.com> wrote:
>> I haven't looked at the implementation, so this may be an entirely
>> stupid suggestion, but would it be possible to instead render the
>> completions as?
>>
>>     % git checkout --<tab>
>>     --[no-]conflict=                   --[no-]patch
>>     --[no-]detach                      --[no-]progress
>>
>> This would address the problem of the --no-* options taking double the
>> screen space.
>
> It took me so long to reply partly because I remember seeing some guy
> doing clever trick with tab completion that also shows a short help
> text in addition to the complete words. I could not find that again
> and from my reading (also internet searching) it's probably not
> possible to do this without trickery.

Okay.

>> It's also more intuitive than that lone and somewhat weird-looking
>> "--no-" suggestion.
>
> It's not that weird if you think about file path completion, where you
> complete one path component at a time not full path, bash just does
> not show you full paths to everything.

The "path completion" analogy and the dotted configuration variable
analogy (below) don't really help me find "--no-" less weird. We're
used to "/" as a separator in paths, and "." a separator in
configuration variables, so they are easier to digest than "-" somehow
being a separator for --no-<option>.

It _might_ feel as bit less weird if it was presented as --no-<option>
or --no-{...} or --no-<...> or --no-... or something, but those seem
pretty weird too, so perhaps not. Anyhow, it's not a major issue; the
--[no-]foo idea seems pretty intuitive, but if it can't be easily
implemented, then falling back to your --no- idea makes sense.

> I'm arguing about this because I want to see your reaction, because
> I'm thinking of doing the very same thing for config completion. Right
> now "git config <tab>" gives you two pages of all available config
> variables. I'm thinking that we "git config <tab>" just shows the
> groups, e.g.
>
>> ~/w/git $ git config
> add.              interactive.
> advice.           log.
> alias.            mailmap.
> am.               man.
>
> Only when you do "git config log.<tab>" that it shows you log.*

Just wondering out loud (again): add.<var> | add.{...} | add.<...> |
add...; those aren't very attractive either, so plain "add." may
indeed be best.