Thanks to everybody.
Allready now I feel much wiser.
Stephen came up with a good description of the three ways variable
declarations and when to use what. I may use (even quote) the description
in an eventual patch to some documentaiton one day.
Steve came up with a nice ASCII-drawing that I will save for eventual
inclusion in some doc.
I will probably also put a cross-ref from somewhere in the lispref manual
to the node describing how to create "real" packages in the xemacs manual.
Regarding the real-world example you came up with steve, I would consider
it's me who did some bad coding (I must have been sleepy that day ;-). In
that example I would say that the random table should be an optional
parameter to the eicq-v5-encrypt funciton and hence the eicq-v5-encrypt
math-definition, Discussion regarding that should continue on
eicq-devel(a)lists.sf.net
Jarl