In the case of global variable V, the paths of definition-reference pairs is context insensitive between functions. Again, the reason for this is that binding variables to the procedures they were defined in makes the analysis far more complex. This means that for some C, there may be a reference on some UN-realizable path which is considered a use. This makes for a conservative analysis. This is why, in the "raw" data file, you may see up to twenty uses for some C with global variable g on the right side of the copy. This is a limitation of the algorithm.
p1()
{
int x = g;//definition
p2();
}
p2()
{
print ("hello");
}
p3()
{
p2();
int z = g + 1;//use
}
