We do not know how many URLs there will be when we start parsing
so let's use a Vector
Single threaded for simplicity but a real program would use multiple threads
One to load and parse the data
Another thread (probably the main thread) to serve the data
Early data could be provided before the entire document had been read
The character data of each url element needs to be stored.
Everything else can be ignored.
A startElement() with the name
url indicates that we need to start
storing this data.
An endElement() with the name url indicates that we need to stop
storing this data, convert it to a URL and put it in the
Vector
Should we hide the XML parsing inside a non-public class to avoid accidentally calling the methods from unexpected places or threads?