Indexed memory is used as a generic protocol for handling vectors of data in genetic programming. Using this simple method, a single program can generate many outputs. It eliminates the complexity of maintaining different trees for each desired parameter and avoids problem-specific function calls for handling the vectors. This allows a single set of programming language primitives applicable to wider range of problems. For a test case, the technique is applied to evolution of behavioural control programs for a simulated 2d vehicle in a corridor following problem.