Re: [Unicon-group] Proposal for changing the parameter type and default value syntax...

[Nevin :-] Liber]

[Note:  I don't know the Unicon macro facility very well; I'm just 
adding a general discussion of what would have to be done with macros]

          procedure p(l,m,n,x,y,z)

equivalent of the p(x:=a; z:=b) seems to be expression

         (every (l|m|n|x|y|z):=&null) & x:=a & z:=b & p(l,m,n,x,y,z).

The first and the last of the conjuncts are fixed, used always in 
the same form, so even limited Unicon macros allow further 
simplification. One of the possibilities is

         P(x:=a & z:=b)Pend  .

Hence, those who like (or tolerate) macros can manage CBPN for 
themselves, perhaps not in the prettiest possible form, however, it 
is possible.
How would this work for recursion, where the variable names are the 
same, but you only want to be modifying the local instance?  Take 
something like:

        procedure F(..., x, ...)
                ...
                F(..., x := x+1, ...)
                ...
        end
If it were macro based, there would be severe restrictions on naming 
conventions, as what you name a parameter in one function may affect 
your ability to call it or be called from another function.  This 
violates locality, making it very difficult to manage.

Now, there are ways around this as well; basically, have the macro 
change each parameter name to a name with the function name encoded 
in it.  In the easy way to get it right, you would need to do that at 
the point of function declaration, not at the point of function 
invocation, which would add overhead to each and every function that 
takes parameters.  A macro expansion of the top might be:

        procedure F(..., _F_x, ...)
                x := _F_x
                ...
                F(..., x+1, ...)
                ...
        end
Alternatively, if you want to do it at the point of function 
invocation to make only the people who use call by name pay for it, 
you have to take into account calls like:

	F(..., x := F(x := x + 1), ...)

and have the macro come up with multiple unique local variable names 
(say, by appending it with an increasing number).

However, this does point out a problem with the syntax.  Is

	F(..., x := x + 1, ...)

invoking call by name on the parameter x, or is it incrementing x by 
1 and passing that in positionally to F?
--
 Nevin ":-)" Liber  <mailto:[EMAIL PROTECTED]>  (773) 961-1620

-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn
_______________________________________________
Unicon-group mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/unicon-group