• Results table is persistent and temporary table is transient now.
• WebCrawler does not go to links if they are already in file passed to crawl()
• Crawler serializes and deserializes results as XML.

To-Do list (revised after asking Keith about specification):

• Rewrite WebCrawler so that HTMLread methods are used in the class.
• Type check the XML when it is deserialized to object which is case to List. InstanceOf is frowned upon; use some kind of dtd/schema. Indeed if the file contains a load of rubbish or is empty we need a way of dealing with that
• Hide Database class in some way or other; inner class?!

• Test,test,test!

• Add some basic awareness of robots.txt.
• Add some concurrency; one crawler per core, with a shared temporary table.