Fix the computation of the length of multops.

* src/ltlvisit/length.cc (visit(multop*)): New function. "a & b &
c" has length 5, not 4, even though it is stored as And(a,b,c).
This caused reduc.test to fail on some formulae.

ChangeLog

+2010-01-22  Alexandre Duret-Lutz  <adl@lrde.epita.fr>
+
+	Fix the computation of the length of multops.
+
+	* src/ltlvisit/length.cc (visit(multop*)): New function. "a & b &
+	c" has length 5, not 4, even though it is stored as And(a,b,c).
+	This caused reduc.test to fail on some formulae.
+
 2010-01-21  Alexandre Duret-Lutz  <adl@lrde.epita.fr>
 
 	Please the style checks...

src/ltlvisit/length.cc

+// Copyright (C) 2010 Laboratoire de Recherche et Dveloppement de
+// l'Epita (LRDE).
 // Copyright (C) 2004, 2005  Laboratoire d'Informatique de Paris 6 (LIP6),
 // dpartement Systmes Rpartis Coopratifs (SRC), Universit Pierre
 // et Marie Curie.
@@ -21,6 +23,7 @@
 
 #include "length.hh"
 #include "ltlvisit/postfix.hh"
+#include "ltlast/multop.hh"
 
 namespace spot
 {
@@ -42,6 +45,17 @@ namespace spot
 	  return result_;
 	}
 
+	virtual void
+	visit(multop* mo)
+	{
+	  unsigned s = mo->size();
+	  for (unsigned i = 0; i < s; ++i)
+	    mo->nth(i)->accept(*this);
+	  // "a & b & c" should count for 5, even though it is
+	  // stored as And(a,b,c).
+	  mo += s - 1;
+	}
+
 	virtual void
 	doit_default(formula*)
 	{