Patches since pre3 that are not in pre6

["Bart Schaefer" <schaefer@candle.brasslantern.com>]

About a week ago I wrote that there appeared to be some patches from pre3
that didn't make it into pre4, but that I hadn't figured out which ones.
I finally had a chance to do so; most of the issues they addressed were
patched differently in pre5 or pre6, but a few are still completely missing
(all from me, as it turns out).

All the following patches are reversed diffs because they're changes made
to an older file that are missing from the newer file.  Use "patch -p1 -R".
All these patch hunks are very short, so I've interspersed commentary.

First is a simple change to the cvs compctl-example; it's not essential,
but it looks more consistent with the rest of the function.

--- zsh-3.0-pre3-work/Misc/compctl-examples	Mon Jul 15 10:46:51 1996
+++ zsh-3.0-pre4/Misc/compctl-examples	Sat Jul 27 08:52:55 1996
@@ -439,9 +439,8 @@
 	pref=
     fi
     [[ -n "$pref" && "$pref" != */ ]] && pref=$pref/
-    if [[ -f ${pref}CVS/Entries ]]
-    then
-	reply=( "${pref}${^${${(f@)$(<${pref}CVS/Entries)}#/}%%/*}"
+    if [[ -f "${pref}CVS/Entries" ]] then
+	reply=( "${pref}${^${${(f@)$(< ${pref}CVS/Entries)}#/}%%/*}"
 		${pref}*/**/CVS(:h) )
     else
 	reply=( ${pref}*/**/CVS(:h) )

Next is my change to ignore "insignificant" white space in case patterns,
approximately as per POSIX.  I don't *think* this was fixed in any other
way between pre4 and pre6; the ChangeLog doesn't mention it.

--- zsh-3.0-pre3-work/Src/lex.c	Mon Jul 22 14:38:56 1996
+++ zsh-3.0-pre4/Src/lex.c	Sat Jul 27 08:52:56 1996
@@ -638,10 +635,6 @@
 	if (inblank(c) && !in_brace_param && !pct)
 	    act = LX2_BREAK;
 	else {
-	    if (incasepat && pct == 1 && !in_brace_param && iblank(c)) {
-		c = hgetc();
-		continue;
-	    }
 	    act = lexact2[STOUC(c)];
 	    c = lextok2[STOUC(c)];
 	}

Finally, there's this little tweak to avoid wandering off the end of a
string during subsitution if the string happens not to contain 'endchar'
anywhere.  The current code will spit out a warning message upon crossing
the string terminator when DEBUG is defined, but otherwise it silently
goes off into never-never land.  Maybe this is an "impossible" condition,
but if we're going to bother producing a debug message, shouldn't we also
avoid creating the wild pointer?

--- zsh-3.0-pre3-work/Src/subst.c	Fri Jul 19 13:05:37 1996
+++ zsh-3.0-pre4/Src/subst.c	Sat Jul 27 08:52:56 1996
@@ -137,9 +137,8 @@
 		endchar = *str;
 		*str = '\0';
 
-		while (*++str && *str != endchar)
-		    ;
-		DPUTS(!*str, "Oops. parse error in command substitution");
+		while (*++str != endchar)
+		    DPUTS(!*str, "Oops. parse error in command substitution");
 	    }
 	    *str++ = '\0';
 	    if (endchar == Outpar && str2[1] == '(' && str[-2] == ')') {

Again, these are reversed diffs, so the interesting lines are the ones with
the '-' prefix.

-- 
Bart Schaefer                             Brass Lantern Enterprises
http://www.well.com/user/barts            http://www.nbn.com/people/lantern

New male in /home/schaefer:
>N  2 Justin William Schaefer  Sat May 11 03:43  53/4040  "Happy Birthday"