XML parser now initializes child-list elements as empty arrays.

Previously, a child element that was described as a list but that
did not appear in the XML response would not be set in the response
data.  This makes them more consistent, always getting initalized
to an empty array, reguardless if the list child is in the XML.

lib/xml/parser.js

@@ -55,16 +55,24 @@ AWS.XML.Parser = inherit({
     } else if (error) {
       throw AWS.util.error(error, {code: 'XMLParserError'});
     } else { // empty xml document
-      return {};
+      return this.parseStructure({}, this.rules);
     }
 
   },
 
   parseStructure: function parseStructure(structure, rules) {
     var data = {};
-    AWS.util.each.call(this, structure, function (name, value) {
-      var rule = rules[name] || {};
-      data[rule.name || name] = this.parseMember(value, rule);
+
+    // force array members to always be present
+    AWS.util.each.call(this, rules, function(memberName, memberRules) {
+      if (memberRules.type == 'list') {
+        data[memberRules.name || memberName] = [];
+      }
+    });
+
+    AWS.util.each.call(this, structure, function (xmlName, value) {
+      var rule = rules[xmlName] || {};
+      data[rule.name || xmlName] = this.parseMember(value, rule);
     });
 
     return data;

test/unit/xml/parser.spec.coffee

@@ -86,6 +86,18 @@ describe 'AWS.XML.Parser', ->
       parse xml, rules, (data) ->
         expect(data).toEqual({items:[]})
 
+    it 'returns missing lists as []', ->
+      xml = '<xml></xml>'
+      rules =
+        type: 'structure'
+        members:
+          items:
+            type: 'list'
+            members:
+              type: 'string'
+      parse xml, rules, (data) ->
+        expect(data).toEqual({items:[]})
+
     it 'Converts xml lists of strings into arrays of strings', ->
       xml = """
       <xml>