Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1 6/24/83; site decwrl.UUCP
Path: utzoo!watmath!clyde!burl!mgnetp!ihnp4!zehntel!dual!decwrl!powell
From: powell@decwrl.UUCP (Mike Powell)
Newsgroups: net.lang.mod2
Subject: Compiling Modula-2 in one pass?
Message-ID: <3432@decwrl.UUCP>
Date: Fri, 24-Aug-84 19:25:48 EDT
Article-I.D.: decwrl.3432
Posted: Fri Aug 24 19:25:48 1984
Date-Received: Wed, 29-Aug-84 01:25:04 EDT
Sender: powell@decwrl.UUCP
Organization: DEC Western Research Lab, Los Altos, CA
Lines: 30

In the Report on The Programming Language Modula-2 (p.143 of Wirth's
Programming in Modula-2, Springer-Verlag, 1982), Section 4, paragraph 2,
it says:
    In general, the scope [of an identifier] extends over the entire block
    (procedure or module declaration) to which the declaration belongs and
    to which the object is local.  The scope rule is augmented by the following
    cases:

    1. If an identifier x defined by a declaration D1 is used in another
    declaration (not statement[!]) D2, then D1 must textually precede D2.
    ...

From the general rule, and the "not statement" comment, it seems that a
statement may make a forward reference to a declaration.  Case 1 does prevent
declarations from making forward references to other declarations.  The absence
of a forward procedure declaration, and the usefulness of allowing two local
modules to import each other's exports both argue that this was an intentional
decision to require a multi-pass compiler.

Note that I, too, had a one-pass compiler until I compiled some of the ETH
software and found it had forward references from statements to declarations.
I have heard that the Volition compiler permits forward declarations of at
least procedures and perhaps modules and other things, but I don't have direct
knowledge of it.
					Michael L. Powell
					Digital Equipment Corporation
					Western Research Laboratory
					Los Altos, CA  94022
					{decvax,ucbvax}!decwrl!powell
					powell@decwrl