Battling LiveVariables

Still battling with the class LiveVariables. I don't like the way it stores live range information. Also I don't like the traditional way of storing it. In/out sets simply do not reflect the state of a SSA program. I think liveness information should be stored in the edges of the control flow graph. The out set is the same for each instruction, but sometimes the same instruction has different branches, and the set of temporaries alive at each branch is not the same. Of course, this problem never happens with PHI functions. I will implement a liveness algorithm by myself.

But, what did I do today? Well, I worked with the LiveVariables class, and end up printing a graph like this one here... The source code of the class to print live ranges is in Fernando_LiveSet.cpp and Fernando_LiveSet.h. I shall have problems with Bernie...