Subroutines should be used when you have any significant portion of code that appears at least twice. The benefits are huge. First, in code maintenance, anything that needs fixed or altered only need be altered in one place. Too often I have seen broken code appear throughout a program, partially repaired - left broken because not every instance of it was found (or known to exist). It also make for much less total code. That makes the Sesame data engine run faster. Subroutines also reduce the amount of temporary storage the running program will need to use. In line repeating code (not using a subroutine) will run a tiny bit faster in that it doesn't need to setup a stack and move execution to the subroutine. But the tiny performance gain is not worth the unreliability and difficulty of maintaining repeating inline code. The performance gain in many database will be offset by the slowdown due to increased memory usage and communications speed.

If only one event uses a subroutine, put it with that event. If many use it, put it in the Global Code section for that form.