Newsgroups: comp.lang.apl
Path: watmath!watserv2.uwaterloo.ca!watserv1!70530.1226@compuserve.com
From: Mike Kent <70530.1226@CompuServe.COM>
Subject: Storage class integration
Message-ID: <921001034337_70530.1226_CHC201-1@CompuServe.COM>
Sender: root@watserv1.uwaterloo.ca
Organization: University of Waterloo
Date: Thu, 1 Oct 1992 03:43:38 GMT

In article <ROCKWELL.92Sep301992223@socrates.umd.edu>,
Rockwell@socrates.umd.edu (Raul Deluth Miller-Rockwell) suggests that
more transparent storage class integration would be a boon to programmers.

This is of course true.

What's more, IBM's mainframe APL2 provides such a facility (for almost-
transparent access to disk file storage) via an associated processor.
Name association to a file establishes the file as a vector in the
workspace (without actually moving data into main memory until it is
referenced), each record in the file being one item of the vector.
There may be some restrictions on the types of files which work.
Indexing and index specification are supported, as are catenation 
and drop (among the structural functions), and if the file consists of
e.g. 1000 25-by-12 numeric matrices, the sum of the matrices can be
constructed by +/ followed by disclose (just asif it were a "real"
array).  I don't know just what operations will cause very large objects
to materialize in the workspace, but there are some.  Perhaps someone from
IBM could post a more complete description, with appropriate caveats?

