Still no electric, running on battery power. Luckily for me I can mount my PDA and sync against it so that the most up top date code is on my PDA. I have about an hour of battery. So my plan yesterday was no refactor the ZIP code so that it is a bit faster and does not rely on the structures and the reading in general is a bit faster and more multi-threaded friendly.
Hopefully the more direct approach will end up being faster and not require all the structure information to be in memory, since it is only needed when the structures need to be read.
Actually my battery is getting a bit low. In the future when there are native SquirrelJME ports, I could quite literally work on this on my PDA despite it being small. The PDA although slower uses far less battery. However, that would be months away. I suppose what I can do is think of the PVM JVM. Once all the interfaces are setup, I can cleanly determine the best route to take when it comes to native code generation. I will have to fixup and change the java-byte-code behavior so that it is exactly as is represented in the byte code. Rather than adjusting or performing slight optimization, being exact will allow me to more easily map code when the PVM JVM needs to change all the class references to process prefixed ones. One problem of the PVM JVM is that it will cost much memory.
And the electric is back again, so I do not have to worry about battery life as much anymore.