Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!think.com!zaphod.mps.ohio-state.edu!cis.ohio-state.edu!ucbvax!CHERRY.CRAY.COM!cargo
From: cargo@CHERRY.CRAY.COM (David S. Cargo)
Newsgroups: comp.lang.icon
Subject: Records
Message-ID: <9105311313.AA03855@cherry04.cray.com>
Date: 31 May 91 13:13:31 GMT
Sender: daemon@ucbvax.BERKELEY.EDU
Distribution: inet
Organization: The Internet
Lines: 31

I use records instead of lists in cases where I want a more
mnemonic way of accessing structures.  In a program I'm working
on right now, I can tell what structures are compatible because
I have named the fields the same names in the same order.  It
would not be possible to decide that so easily if the lists were
simply the same size.

Likewise, the fact that records are their own types allows me to
base actions based on the type of a particular parameter.  I have
exploited these factors together in the following fragment of code.
Because the name of the record type is also the name of the record
constructor for that type, the following routine returns a value
whose type is the same as the type passed to it without knowing
what that type is.  As long as the type is a record with at least
one field named entries, this fragment will work.

## entry_name -- return the entry portion of an absolute path
procedure entry_name(a_path)
#   Uses globals
#       verified   (record constructor)
#       unverified (record constructor)
#   record verified(entries)
#   record unverified(entries)
    if *a_path.entries > 1
    then return type(a_path)(a_path.entries[-1])
end

A lot of the code I write takes advantage of the type information
carried along by record values.

David S. Cargo (cargo@cray.com)