Dave VanHorn wrote: > Filling with randoms means to me, that it might get out the door, and > work, then fail in the field... :( No, no! That's only a tool for use during development to help catch uninitialized variables. And I don't even rely on that; first I do a simulation of the reset code, and make *sure* that the variables are all initialized *before* there is any chance for external inputs to influence things. The randomization is definitely *not* something to ship. But once I'm sure that my variables are correctly initialized before use, there's no benefit to having all RAM initialized to zero first. IMHO, doing so encourages laziness that will bite you later. Eric