FYI; Both: Fix nullable again

[Akim Demaille]

I think this bug was in the original Bison.

Index: ChangeLog
from  Akim Demaille  <address@hidden>

        * src/nullable.c (set_nullable): Useless rules must be skipped,
        otherwise, since we range over their symbols, we might look at a
        nonterminal which no longer ``exists'', i.e., it is not counted in
        `nvars', hence we overflow our arrays.

Index: src/nullable.c
===================================================================
RCS file: /cvsroot/bison/bison/src/nullable.c,v
retrieving revision 1.20
diff -u -u -w -u -r1.20 nullable.c
--- src/nullable.c 13 Dec 2001 11:02:20 -0000 1.20
+++ src/nullable.c 15 Dec 2001 16:21:16 -0000
@@ -69,6 +69,8 @@
   p = relts;
 
   for (ruleno = 1; ruleno < nrules + 1; ++ruleno)
+    if (rule_table[ruleno].useful)
+      {
     if (ritem[rule_table[ruleno].rhs] > 0)
       {
        /* This rule has a non empty RHS. */
@@ -98,6 +100,7 @@
          {
            nullable[rule_table[ruleno].lhs] = 1;
            *s2++ = rule_table[ruleno].lhs;
+             }
          }
       }