From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <zsh-users-return-19127-mason-zsh=primenet.com.au@zsh.org>
Received: (qmail 10918 invoked by alias); 21 Sep 2014 20:27:57 -0000
Mailing-List: contact zsh-users-help@zsh.org; run by ezmlm
Precedence: bulk
X-No-Archive: yes
List-Id: Zsh Users List <zsh-users.zsh.org>
List-Post: <mailto:zsh-users@zsh.org>
List-Help: <mailto:zsh-users-help@zsh.org>
X-Seq: 19127
Received: (qmail 26250 invoked from network); 21 Sep 2014 20:27:44 -0000
X-Spam-Checker-Version: SpamAssassin 3.3.2 (2011-06-06) on f.primenet.com.au
X-Spam-Level: 
X-Spam-Status: No, score=-1.4 required=5.0 tests=BAYES_00,RCVD_IN_DNSWL_LOW,
	RCVD_NUMERIC_HELO,SPF_HELO_PASS,T_FSL_HELO_BARE_IP_2 autolearn=no
	version=3.3.2
X-Injected-Via-Gmane: http://gmane.org/
To: zsh-users@zsh.org
From: Yuri D'Elia <wavexx@thregr.org>
Subject: Re: rsync --progress stops completion
Date: Sun, 21 Sep 2014 22:27:26 +0200
Message-ID: <lvncbe$h5b$1@ger.gmane.org>
References: <lv9jav$8l6$1@ger.gmane.org>  <2002755.9ryFYYVtTN@note>	<5418786F.8030001@thregr.org>	<140916175124.ZM5742@torch.brasslantern.com>	<54194198.2010607@thregr.org>	<140917085133.ZM6725@torch.brasslantern.com>	<541AA918.8060503@thregr.org>	<140918093602.ZM7963@torch.brasslantern.com>	<lvjuvl$mru$1@ger.gmane.org>	<140918190130.ZM8366@torch.brasslantern.com>	<lvjv8o$r67$1@ger.gmane.org>	<140920112024.ZM29459@torch.brasslantern.com>	<lvmv6s$o5c$1@ger.gmane.org>	<140921102006.ZM31805@torch.brasslantern.com>	<lvn24g$qhj$1@ger.gmane.org> <140921123854.ZM1299@torch.brasslantern.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit
X-Complaints-To: usenet@ger.gmane.org
X-Gmane-NNTP-Posting-Host: 193.106.183.18
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:31.0) Gecko/20100101 Icedove/31.0
In-Reply-To: <140921123854.ZM1299@torch.brasslantern.com>

On 09/21/2014 09:38 PM, Bart Schaefer wrote:
> You might be better off using "compadd -x" to display a message, but
> that only shows up if there's a completion listing.  Hmm.

After some testing, I got this:

#autoload
local -a _complete_fallback_precmd

_complete_fallback_cleanup()
{
  precmd_functions=_complete_fallback_precmd
  _complete_fallback_precmd=()
}

_complete_fallback()
{
  if [[ -n $_complete_fallback_precmd || $LASTWIDGET = *complete* ]]
  then
    if [[ -z $_complete_fallback_precmd ]]
    then
      _complete_fallback_precmd=precmd_functions
      precmd_functions+=( _complete_fallback_cleanup )
      compadd -x "fallback: enabled"
      return 0
    fi

    [[ $compstate[context] = command && $CURRENT -gt 2 ]] || return 1
    words=("$words[1]" "${(@)words[$CURRENT,-1]}")
    CURRENT=2
    _compskip=default
    _complete
  fi
}

_complete_fallback "$@"

===

When the hook is added, I display a message and then stop to make it
readable, as otherwise it's obliterated by the completion list. Felt a
bit odd in the beginning, since you need to press tab *again*, but it
gives you a chance to just accept/cancel the command if you're in doubt.
I like the behavior, as it makes obvious that the completion broke at
that point.

I also removed the local curcontext, since after
_complete_fallback_precmd is set, the context is already
:completion::complete-fallback:* due to the fact that _complete_fallback
is immediately entered.

Any obvious problems you can spot?