what I mean by a connection pool is that, we have say 15 "open" connections in the pool and any client requesting a db connection will use one of the already open. if the client finishes with it, that connection still open and back to the pool for other clients request.
But in case of pg_pconnect, any client request for a db connection, the server will "open" a persistent connection which lead to delays if we have many clients requesting a connection in a same time.
so, i think pg_pconnect differs from a connection pool which already have a pool of open connections.
Fat.