Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.3 alpha 4/15/85; site cbosgd.UUCP
Path: utzoo!watmath!clyde!cbosgd!db
From: db@cbosgd.UUCP (Dave Bursik)
Newsgroups: net.micro.mac
Subject: Re: Adding Assembly Language routines to Microsoft BASIC V2.0
Message-ID: <1077@cbosgd.UUCP>
Date: Wed, 24-Apr-85 02:17:15 EST
Article-I.D.: cbosgd.1077
Posted: Wed Apr 24 02:17:15 1985
Date-Received: Thu, 25-Apr-85 02:57:56 EST
References: <1740@decwrl.UUCP>
Organization: AT&T Bell Laboratories, Columbus, Oh
Lines: 36
Summary: More on machine language with MBASIC

In addition to the documentation available from Microsoft (undoubtably
the "right way" to do it), the quick-and-dirty approach to calling
user-supplied machine language routines from BASIC V2.0 as described in
the manual (putting the binary into an array and CALLing it) works fairly
well once you realize that the parenthesis in the middle of the calling-
sequence example is supposed to be a comma and that the VARPTR value for
the entry point MUST be stored as a 4-byte quantity.  However, should you
run into problems and get the bomb box in your ML routine, I don't know of
any straightforward way to get more information using BASIC alone.

I recently acquired Aztec C (commercial version) and was pleasantly
surprised to get a register dump from the shell when my BASIC program
bombed.  At first, I suspected that I might be seeing the interpreter's
registers (which wouldn't do me much good), but, having made a little
more progress, I became convinced that I was getting the Right Stuff.

So far, I have found that Aztec package and MS-BASIC V2.0 complement
one another fairly well.  I haven't done much with the compiler itself,
but the assembler, shell, and Z (vi) editor are all pretty nice to use,
and have relieved most of my occasional frustrations with the Finder
(e.g., waiting for the desktop to appear, not being able to grep or
diff files, etc.).

Getting the hex off the assembly listing is infinitely preferable to
hand-assembly, which I've found to be rather error-prone.  My only
complaint about using the Aztec assembler to get the hex to insert in
BASIC programs is that it seems to always generate the long form of the
branch instruction and doesn't fill in the displacement, even though the
target address is within the same source module and well within the short
branch range.  Perhaps there's a way to force the assembler to optimize a
bit (:-)), but I haven't tumbled to it yet.

Dave Bursik/..cbosgd!db

[The Disclaimer: I have no financial interest in any of these companies
 above and beyond the dollars I shelled out for their products.]