From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <zsh-workers-request@math.gatech.edu>
Received: (qmail 13481 invoked from network); 4 Jan 1999 09:32:06 -0000
Received: from math.gatech.edu (list@130.207.146.50)
  by ns1.primenet.com.au with SMTP; 4 Jan 1999 09:32:06 -0000
Received: (from list@localhost)
	by math.gatech.edu (8.9.1/8.9.1) id EAA01086;
	Mon, 4 Jan 1999 04:30:51 -0500 (EST)
Resent-Date: Mon, 4 Jan 1999 04:30:51 -0500 (EST)
Date: Mon, 4 Jan 1999 10:29:22 +0100 (MET)
Message-Id: <199901040929.KAA09706@beta.informatik.hu-berlin.de>
From: Sven Wischnowsky <wischnow@informatik.hu-berlin.de>
To: zsh-workers@math.gatech.edu
In-reply-to: "Bart Schaefer"'s message of Sun, 3 Jan 1999 18:14:36 -0800
Subject: PATCH: was Re: zsh-3.1.5-pws4. accept-and-menu-complete bug?
Resent-Message-ID: <"TxH3g.0.sG.Bf8as"@math>
Resent-From: zsh-workers@math.gatech.edu
X-Mailing-List: <zsh-workers@math.gatech.edu> archive/latest/4878
X-Loop: zsh-workers@math.gatech.edu
Precedence: list
Resent-Sender: zsh-workers-request@math.gatech.edu


Bart Schaefer wrote:

> 
> On Dec 31, 10:58am, Andrej Borsenkow wrote:
> } Subject: zsh-3.1.5-pws4. accept-and-menu-complete bug?
> }
> } bor@itsrm2:/tools/src%> bindkey '^[^M'
> } "^[^M" accept-and-menu-complete
> } bor@itsrm2:/tools/src%> rm -rf mc-4.5.1/^@mc-4.5.3/
> } mc-4.5.1/  mc-4.5.3/  
> } 
> } Note this ^@ and not removed suffix instead of blank.
> 
> In acceptandmenucomplete() in zle_tricky.c, there's this fragment:
> 
>     cs = menuend + menuinsc;
>     inststrlen(" ", 1, 1);
>     menuinsc = menulen = 0;
>     menupos = cs;
> 
> menuend is supposed to be "the end position of this string in the command
> line" and menuinsc "is the length of any suffix that has been temporarily
> added."  However, at that point menuend is large enough to *include* the
> trailing slash.  In 3.0.5's acceptandmenucomplete(), menuinsc is not added 
> when computing cs.
> 
> So I don't know if this is a general problem -- that menuend is just plain
> wrong, and anything that refers to it will be messed up -- or if that ref
> to menuinsc crept in by mistake and should be deleted (in which case the
> suffix doesn't get autoremoved), or if there's a whole lot of suffix-removal
> code that ought to be copied from 3.0.5, or what.

It seems that I forgot to test accept-and-menu-complete in the final
version  of my patch, sorry.

`menuend' now points to the position where the cursor is (which may be 
in the word, e.g. for completion in paths). The patch below fixes a
bug in the calculation of `menuend' in do_single() and the stuff in
acceptandmenucomplete() (positioning the cursor and removing the
suffix).

Bye
 Sven

*** os/Zle/zle_tricky.c	Thu Dec 17 12:20:33 1998
--- Src/Zle/zle_tricky.c	Mon Jan  4 10:19:19 1999
***************
*** 408,418 ****
  void
  acceptandmenucomplete(void)
  {
      if (!menucmp) {
  	feep();
  	return;
      }
!     cs = menuend + menuinsc;
      inststrlen(" ", 1, 1);
      menuinsc = menulen = 0;
      menupos = cs;
--- 408,422 ----
  void
  acceptandmenucomplete(void)
  {
+     int sl = suffixlen[' '];
+ 
      if (!menucmp) {
  	feep();
  	return;
      }
!     cs = menupos + menulen + menuinsc;
!     if (sl)
! 	backdel(sl);
      inststrlen(" ", 1, 1);
      menuinsc = menulen = 0;
      menupos = cs;
***************
*** 4991,5000 ****
      if (m->suf) {
  	havesuff = 1;
  	menuinsc = ztrlen(m->suf);
! 	if (menuwe && (m->flags & CMF_REMOVE)) {
! 	    makesuffix(menuinsc);
! 	    if (menuinsc == 1)
! 		suffixlen[m->suf[0]] = 1;
  	}
      } else {
  	/* There is no user-specified suffix, *
--- 4995,5008 ----
      if (m->suf) {
  	havesuff = 1;
  	menuinsc = ztrlen(m->suf);
! 	menulen -= menuinsc;
! 	if (menuwe) {
! 	    menuend += menuinsc;
! 	    if (m->flags & CMF_REMOVE) {
! 		makesuffix(menuinsc);
! 		if (menuinsc == 1)
! 		    suffixlen[m->suf[0]] = 1;
! 	    }
  	}
      } else {
  	/* There is no user-specified suffix, *


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