From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <zsh-workers-return-11678-mason-zsh=primenet.com.au@sunsite.auc.dk>
Received: (qmail 8897 invoked from network); 31 May 2000 06:10:05 -0000
Received: from sunsite.auc.dk (130.225.51.30)
  by ns1.primenet.com.au with SMTP; 31 May 2000 06:10:05 -0000
Received: (qmail 7854 invoked by alias); 31 May 2000 06:09:53 -0000
Mailing-List: contact zsh-workers-help@sunsite.auc.dk; run by ezmlm
Precedence: bulk
X-No-Archive: yes
X-Seq: 11678
Received: (qmail 7847 invoked from network); 31 May 2000 06:09:52 -0000
Date: Wed, 31 May 2000 08:09:50 +0200 (MET DST)
Message-Id: <200005310609.IAA21412@beta.informatik.hu-berlin.de>
From: Sven Wischnowsky <wischnow@informatik.hu-berlin.de>
To: zsh-workers@sunsite.auc.dk
In-reply-to: "Bart Schaefer"'s message of Tue, 30 May 2000 17:12:32 +0000
Subject: PATCH: Re: tag-order problem.


Bart Schaefer wrote:

> On May 30, 11:22pm, Tanaka Akira wrote:
>
> ...
> 
> } Z(3):akr@serein% Src/zsh -f
> } serein% bindkey -e;autoload -U compinit; compinit -D
> } serein% zstyle ':completion:*' tag-order '*:label' 
> } serein% which <TAB>
> } 
> } This completes nothing.
> 
> I used the example of completing after `lynx' because _lynx has not been
> edited since before this broke, whereas _which was changed recently.
> 
> The problem is with the change to _arguments to use a counter instead of
> a shift when going around this nested loop:
> 
> ...
> 
> In the previous implementation, `shift descrs' at the point where I've put
> `# lots deleted' would mean that the inner while loop consumed all the
> descriptions on the first time around the _tags loop, so they were not
> tried again for the second and succeeding tags.  Now, all the descriptions
> are tried for every tag.  I'm not sure if that's what Sven intended, but
> it breaks the `all tags not explicitly selected are tried last' behavior
> in some way.

It definitely is what I intended. But, yes, the problem shows up
because now the inner loop is run more than once, leading to the
state-name being more than once in $state, which makes the test in the 
caller fail.

The patch below just ensures that we don't add states more than once
to $state. Another question is if the test in line 345 should be
changed (or if $tried should be set when a `->state' was
executed). We'll have to play some more with it, I guess.

Bye
 Sven

Index: Completion/Base/_arguments
===================================================================
RCS file: /cvsroot/zsh/zsh/Completion/Base/_arguments,v
retrieving revision 1.26
diff -u -r1.26 _arguments
--- Completion/Base/_arguments	2000/05/30 07:49:00	1.26
+++ Completion/Base/_arguments	2000/05/31 06:03:10
@@ -223,15 +223,18 @@
           fi
 
           if [[ "$action" = -\>* ]]; then
-            comparguments -W line opt_args
-            state=( "$state[@]" "${${action[3,-1]##[ 	]#}%%[ 	]#}" )
-	    if [[ -n "$usecc" ]]; then
-	      curcontext="${oldcontext%:*}:$subc"
-	    else
-	      context=( "$context[@]" "$subc" )
-	    fi
-            compstate[restore]=''
-            aret=yes
+	    action="${${action[3,-1]##[ 	]#}%%[ 	]#}"
+	    if (( ! $state[(I)$action] )); then
+              comparguments -W line opt_args
+              state=( "$state[@]" "$action" )
+	      if [[ -n "$usecc" ]]; then
+	        curcontext="${oldcontext%:*}:$subc"
+	      else
+	        context=( "$context[@]" "$subc" )
+	      fi
+              compstate[restore]=''
+              aret=yes
+            fi
           else
             if [[ -z "$local" ]]; then
               local line

--
Sven Wischnowsky                         wischnow@informatik.hu-berlin.de