The lifetime of current AO systems can span decades, from the first ideas to decommissioning. During this time, computer systems, from operating systems to programming languages to data analysis methods can change fundamentally. This can be a challenge for AO simulations, where very old results sometimes need to be "resurrected", re-analyzed, reproduced or changed.
In this contribution, I will present a range of methods that have been used to shield simulation tools from unwanted changes, and how we manage to still exactly the same results as 15 years ago, if needed. We can still run exactly the same simulations as when they were first done, if needed, and can update only the desired parameters in a controlled fashion.
I will also present a range of things that can go wrong if a simulation environment is left to drift uncontrollably.
|