[1505] in java-interest
Re: Systems Programming Oriented Java
daemon@ATHENA.MIT.EDU (John D. Ramsdell)
Wed Sep 6 12:35:31 1995
To: "John D. Mitchell" <johnm@emf2-003.emf.net>
Cc: java-interest@java.sun.com, ramsdell@linus.mitre.org
In-reply-to: Your message of "Wed, 06 Sep 1995 01:09:57 PDT."
<199509060809.BAA02511@emf2-003.emf.net>
Date: Wed, 06 Sep 1995 09:12:23 -0400
From: "John D. Ramsdell" <ramsdell@linus.mitre.org>
> 'J' is already taken.
Okay, call it SPOJ.
> The question I have is... why?
>
> C already exists and is supported *everywhere*.
This is good because a SPOJ compiler would probably generate C. The
compiler for the systems programming oriented dialect of Scheme by
Jonathan Rees and Richard Kelsey generates C. It is a good assembly
language.
> Why not just write ISO/ANSI/POSIX/etc. conformant C code which are
> (trivially?) wrapped in Java classes.
The idea is to write a SPOJ program fragment which is translated into
ISO/ANSI/POSIX/etc. conformant C code. If you would like to integrate
a fragment into a Java VM, it could be wrapped in Java classes. One
could also execute it as a stand alone program, or link it in as a
loadable device driver.
Do you want to write and debug C code or write Java-like code which is
debugged in a Java environment? Many Lisp implementors use Lisp for
their garbage collector. This is not a new idea. Many Lisp
implementors choose C for their garbage collector. It may be that, in
practice, people would rather use C than SPOJ, however, it would greatly
bother me if the Java language turned out to be too weak to have a
systems programming relative.
John
-
Note to Sun employees: this is an EXTERNAL mailing list!
Info: send 'help' to java-interest-request@java.sun.com
