Suppose A and B are fixed length arrays. C = A + B - A is easily optimized in fortran, to a no op if C is not modified afterwards, and to a copy if it is. I

Suppose A and B are fixed length arrays. C = A + B - A is easily optimized in fortran, to a no op if C is not modified afterwards, and to a copy if it is. In C this is pretty much impossible. To be fair, there are C like languages (ispc, glsl) which makes this work with heroic compiler efforts.