Because all coded
datasets are directly processed (no copies are generated at all), a program
interrupt in the coding phase can irretrievably destroy (!) one or more
datasets, i.e. you absolutely have to make backup copies e.g. on
other storage media (!). In success these copies can be replaced by the
encoding results to get protection against troubles during decoding.
To rule out an unintentional mistake regarding this, the
program only starts if you have explicitly confirmed that you have made backup
copies !
