On 1 Dec 2008, at 13:30, 山东 wrote:
...
I think it just uses the standard method as described in books like
Aho, Sethi & Ullman, "Compilers..." (the "Dragon book"), an iteration
until it stops. There is a more efficient method using Tarjan's SCC
algorithm - search the archives for the Use
c2ompute_FOLLOWS()
{
register int i;
digraph(includes);
for (i = 0; i < ngotos; i++)
{
if (includes[i]) FREE(includes[i]);
}
FREE(includes);
}
digraph(relation)
short **relation;
{
register int i;
infinity = ngotos + 2;
INDEX = NEW2(ngotos + 1, short);
VERTICES =
