Re: PATCH: Small memory leak and doc fix

[Sven Wischnowsky <wischnow@xxxxxxxxxxxxxxxxxxxxxxx>]

Felix Rosencrantz wrote:

> >The patch below seems to make sense, but I'm not sure if that was the
> >problem. Felix, could you try, please?
> 
> Sven, I built a cvs pull with your changes, and noticed that compmatch test
> failure that Vin is also seeing.

Yes, a thinko in the patch. Sorry.

> I used the shell during the day, and didn't see the memory problem I reported.
> However, some of completions I tried that I think will cause the problem also
> are failing.  So, the patch took care of the memory problem, but probably
> caused this new matching problem.

Ok. Could you try again with this patch? It should make things even
safer.

And I forgot to say: this might also have to do with the build.out
test failure some people saw (11653) and which couldn't
reproduce. Could someone who can reproduce that reliably (Andrej?
Felix?) try it again?


Bye
 Sven

Index: Src/Zle/compmatch.c
===================================================================
RCS file: /cvsroot/zsh/zsh/Src/Zle/compmatch.c,v
retrieving revision 1.17
diff -u -r1.17 compmatch.c
--- Src/Zle/compmatch.c	2000/06/07 13:25:32	1.17
+++ Src/Zle/compmatch.c	2000/06/08 09:18:18
@@ -535,8 +535,7 @@
 		    }
 		    /* Give up if we don't have enough characters for the
 		     * line-string and the anchor. */
-		    if (ll < llen + alen ||
-			(sfx ? (lw < alen + aol) : (lw < alen || iw < aol)))
+		    if (ll < llen + alen || lw < alen)
 			continue;
 
 		    if (mp->flags & CMF_LEFT) {
@@ -561,8 +560,9 @@
 			if (!pattern_match(ap, l + aoff, NULL, NULL) ||
 			    (both &&
 			     (!pattern_match(ap, w + aoff, NULL, NULL) ||
-			      (aol && !pattern_match(aop, w + aoff - aol,
-						     NULL, NULL)) ||
+			      (aol && aol <= aoff + iw &&
+			       !pattern_match(aop, w + aoff - aol,
+					      NULL, NULL)) ||
 			      !match_parts(l + aoff, w + aoff, alen, part))))
 				continue;
 		    } else if (!both || il || iw)
@@ -572,19 +572,20 @@
 		     * string matched by the `*'. */
 		    if (sfx && (savl = l[-(llen + zoff)]))
 			l[-(llen + zoff)] = '\0';
-		    for (t = 0, tp = w, ct = 0,
-			     ict = lw - alen + 1 - (sfx ? aol : 0);
+		    for (t = 0, tp = w, ct = 0, ict = lw - alen + 1;
 			 ict;
 			 tp += add, ct++, ict--) {
 			if ((both &&
 			     (!ap || !test ||
 			      !pattern_match(ap, tp + aoff, NULL, NULL) ||
-			      (aol && !pattern_match(aop, tp + aoff - aol,
-						     NULL, NULL)))) ||
+			      (aol && aol <= aoff + ct + iw &&
+			       !pattern_match(aop, tp + aoff - aol,
+					      NULL, NULL)))) ||
 			    (!both &&
 			     pattern_match(ap, tp - moff, NULL, NULL) &&
-			     (!aol || pattern_match(aop, tp - moff - aol,
-						    NULL, NULL)) &&
+			     (!aol || (aol <= iw + ct - moff &&
+				       pattern_match(aop, tp - moff - aol,
+						     NULL, NULL))) &&
 			     (mp->wlen == -1 ||
 			      match_parts(l + aoff , tp - moff,
 						      alen, part)))) {

--
Sven Wischnowsky                         wischnow@xxxxxxxxxxxxxxxxxxxxxxx