DragonFlyBSD

I wanted to submit these before I attempt any serious surgery on /bin/sh.

These two patches *mostly* bring /bin/sh up to WARNS?=3. They don't correct
the "comparison is always true due to limited range of data type" errors that
also appear on WARNS?=0.

The eval.c patch fixes possible logjmp clobbering. The histedit.c patch fixes
a warning of possibly returning an uninitialized value from str_to_event. (I
initialize 'i' as well for good measure.)

Index: eval.c
===================================================================
RCS file: /home/dcvs/src/bin/sh/eval.c,v
retrieving revision 1.5
diff -u -p -r1.5 eval.c
--- eval.c 12 May 2006 01:56:11 -0000 1.5
+++ eval.c 4 Jun 2006 08:15:20 -0000
@@ -977,6 +977,12 @@ struct jmploc loc, *old;
struct strlist *sp;
const char *path;
int ch;
+#ifdef __GNUC__
+ /* Avoid longjmp clobbering */
+ (void) &path;
+ (void) &argc;
+ (void) &argv;
+#endif

for (sp = cmdenviron; sp ; sp = sp->next)
setvareq(sp->text, VEXPORT|VSTACK);
Index: histedit.c
===================================================================
RCS file: /home/dcvs/src/bin/sh/histedit.c,v
retrieving revision 1.7
diff -u -p -r1.7 histedit.c
--- histedit.c 17 Jan 2006 15:06:26 -0000 1.7
+++ histedit.c 25 May 2006 19:05:33 -0000
@@ -441,6 +441,7 @@ const char *s = str;
int relative = 0;
int i, retval;

+ i = retval = 0;
history(hist, &he, H_FIRST);
switch (*s) {
case '-':