Use custom GetHashCode() implementation in XmlDocumentationProvider as the hash codes may get serialized.

The normal .NET string.GetHashCode() isn't guaranteed to be stable across multiple runs of the program (e.g. with .NET 4.5 hash randomization).

ICSharpCode.NRefactory/Documentation/XmlDocumentationProvider.cs

@@ -259,13 +259,30 @@ namespace ICSharpCode.NRefactory.Documentation
 							int pos = linePosMapper.GetPositionForLine(reader.LineNumber) + Math.Max(reader.LinePosition - 2, 0);
 							string memberAttr = reader.GetAttribute("name");
 							if (memberAttr != null)
-								indexList.Add(new IndexEntry(memberAttr.GetHashCode(), pos));
+								indexList.Add(new IndexEntry(GetHashCode(memberAttr), pos));
 							reader.Skip();
 						}
 						break;
 				}
 			}
 		}
+		
+		/// <summary>
+		/// Hash algorithm used for the index.
+		/// This is a custom implementation so that old index files work correctly
+		/// even when the .NET string.GetHashCode implementation changes
+		/// (e.g. due to .NET 4.5 hash randomization)
+		/// </summary>
+		static int GetHashCode(string key)
+		{
+			unchecked {
+				int h = 0;
+				foreach (char c in key) {
+					h = (h << 5) - h + c;
+				}
+				return h;
+			}
+		}
 		#endregion
 		
 		#region GetDocumentation
@@ -277,7 +294,7 @@ namespace ICSharpCode.NRefactory.Documentation
 			if (key == null)
 				throw new ArgumentNullException("key");
 			
-			int hashcode = key.GetHashCode();
+			int hashcode = GetHashCode(key);
 			// index is sorted, so we can use binary search
 			int m = Array.BinarySearch(index, new IndexEntry(hashcode, 0));
 			if (m < 0)