Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Path: utzoo!mnetor!seismo!ut-sally!pyramid!decwrl!sun!rmarti From: rmarti@sun.uucp (Bob Marti) Newsgroups: net.micro.mac Subject: Re: true code-generating M2 for the Mac ? Message-ID: <5495@sun.uucp> Date: Fri, 25-Jul-86 17:45:36 EDT Article-I.D.: sun.5495 Posted: Fri Jul 25 17:45:36 1986 Date-Received: Sat, 26-Jul-86 06:10:40 EDT References: <145@tnocsda.UUCP> <1392@runx.OZ> Organization: Sun Microsystems, Inc. Lines: 44 > > > Is there a true 68000 code-generating (as opposed to p-code) > > Modula-2 compiler forthe Mac ? > > As far as Modula-2 products go, there are no native code generating Modula-2 > compilers available. They all produce M-code. > > > /* Jason Haines ACSnet: baron@runx > * ElecEng Undergraduate CSNET: baron@runx.oz > * Australia ARPA: baron%runx.oz@seismo.css.gov > */ I don't know where you got that information from, but I happen to have not one but TWO different Modula-2 (well almost, see below) compilers generating 68000 code. Both were produced at ETH in Zurich, Switzerland: MacLogimo consists of a 5-pass compiler, a linker, and a simple TTY-based interactive post-mortem debugger. It has been posted to the net in 13 parts about a year ago. MacMETH is a new Modula-2 development system comprising a 1-pass compiler, an editor, and an interactive source-level debugger, all of which were derived from the latest Lilith/Ceres Modula-2 software. Unfortunately, the compiler is non-standard in that it does not allow type-transfer functions of the form y := T(x) where y is a variable of type T. Instead, you have to use the function VAL which has to be imported from module SYSTEM (!), e.g., y := VAL(T,x). Finally, NEW and DISPOSE are not supported, so you have to call Allocate and Deallocate -- imported from module System (!!) -- instead. This system was announced on the net about a month or so ago. I have used both systems a little. The 5-pass compiler is fairly slow. Moreover, you have to go through an additional linking step. However, the quality of the generated code is fairly good. I am truly amazed at the speed of the 1-pass compiler! Moreover, since the editor, compiler and debugger remain in main memory once they have been started up, so the edit-compile-debug cycle is very fast indeed! -- Robert Marti, Sun Microsystems, Inc. UUCP: ...{cbosgd,decvax,decwrl,hplabs,ihnp4,pyramid,seismo,ucbvax}!sun!rmarti ARPA: rmarti@sun.com