Cryptography load order assumptions

To whom it may concern,

The cryptography package is unwittingly assuming that its classes are
going to be loaded in a specific order and that the call to class
#initialize is going to be preformed in that order.

I have recently changed MC1.5 to run its calls to class #initialize in a
sightly different way and it breaks convincingly.

I am on irc #squeak if anyone wants to discuss it...

cheers

Keith