Aucbvax.5698
net.applic
utzoo!decvax!ucbvax!baden
Wed Jan  6 20:25:30 1982
identifiers in FP
I have completed a new version of my FP interpreter (implements Backus'
functional language) basically it includes the associative (tree) insert
functional form. This is due to Dorab Patel (ucla-vax!patel).

I am distributing the software via tape if you are interested.  A user's
guide is also available.

Is there anyone out there who has used a single
assigment language?  I find the use of mnemonic rather than numerical names
to be very akward.

e.g.  define fn == g @ [1,2,h@3]

where '@' stands for functional composition.

Why not:

define fn<a,b,c> == g @ [a,b,h@c]

another reason:  sometime one continually reforms  sequence after breaking
it apart (for the purposes of extracting an argument):

define fn == h @ [1,2,a@3,4] @ [1,b@2,3,4]


why not: define fn<l,m,n,p> ==
v[1] <- a : l
v[2] <- b : m
h : <l,v[2],v[3],p> 

(the final application is the result).


This is alot easier to read, and DOES NOT violate the side-effect
free property of the language (cannot use variables as object parameters
for say the constant functional form).

Any comments?