Re: [Unicon-group] Aspect Oriented Programming and Unicon ?

[Majorinc, Kazimir]

There are some limitations with that proc:=new_proc, because proc cannot 
carry any information beside its own name, unlike arg1,...,argn, they can 
be anything. It can be avoided if one write

go(proc,arg1,...,argn)

consistently, and in that case, there is almost no limit of program 
transformation which can be done in go, including 
go:=aspect1,...,go:=aspectn, i.e. dynamical selection of aspects relatively 
to procedures during runtime. It is uncomfortable to use so many go's, but 
it is at least explicit, otherwise AOP (reminds on AOR, 
adult-oriented-rock) breakpoints are even less structured than "go to", 
they are like "come from" seen in some enigmatic languages.

It could be more comfortable, without redundant "go", if  "proc" can be 
record or object. Is it possible to somehow write O(f) where O is object 
(not class!) in Unicon? Theoretically OO is association of data and 
behavior, but if f(O) is possible, but not O(f), it is still "only" data 
with associated behavior. Convenient implementation of such far-reaching 
OO/functional crossbreed can be relatively simple, through special role 
given to method/attribute named "main" if such exists in class.

----
Kazimir Majorinc, Zagreb, Croatia