[1236] in java-interest
Re: java-interest-digest V1 #133
daemon@ATHENA.MIT.EDU (Robert Stephen Rodgers)
Sun Aug 27 13:59:09 1995
Date: Sun, 27 Aug 1995 10:08:35 -0400 (EDT)
From: Robert Stephen Rodgers <rsrodger@wam.umd.edu>
To: java-interest@java.sun.com
cc: java-interest-digest@java.sun.com
In-Reply-To: <199508270223.TAA17741@webrunner.neato.org>
> I instinctively liked Java, the first I saw it. Java is a personal
Me, too.
[...]
> Yes, writing a Java Extensible Operating Sytem, in Java, ala Oberon,
> is what naturally springs to my mind as a first priority. Therefore I
> have been trying to find out about Java Secureness, and whether it's
> syntax will lend itself to the yoke of Formal Semantic Analysis.
But frankly, I can't think of anything more pointless or silly. Every
time a new fad language comes along some numbskulls get it in their mind
that what they really need is to write an operating system in the language.
It's like exploratory programmers are all trapped in 1970, scurrying to get
a 50k LOC operating system written in their latest toy.
What good would an operating system do you and Java? Do you think people
*really* want to take the back-asswards approach to computers that Oberon,
"run the OS on another OS", brings? Look how successful Oberon is (putting
aside its horrendous interface) -- before you say, "yeah, yeah, it's a
success" sit back and think about the numbers. Oberon isn't a success, it's
not even as popular as widely-regarded-as-dead-but-aren't languages like
Fortran77, Pascal and Modula 2.
Do you fellows really think that you can write an operating system on the
level of Microsoft _Windows 3.1_(+DOS), let alone a "real" operating system
like Linux, NT, Os/2 or Unix?
Forget about it. Operating systems today are few and far between. There
hasn't been a major OS turnover for twelve years, and before that, ten or
more years. You can't possibly put the kind of effort, time, or expertise
into an operating system that commercial companies already _have_ and continue
to improve. All the Microsoft whining aside, Windows NT is not a toy, and
I doubt very much that all the Java enthusiasts in the world could equal it.
The same goes for many other popular operating systems.
Leave operating systems to the people who know how to make them successful
and who have the money to do it. Another toy operating system isn't going
to accomplish anything.
Writing an operating system gets you nothing but an intellectual curiousity
and a good resume bullet. It will do nothing for Java -- the people running
Java programs aren't going to be running them on HappyJAvaNonMicrosoftOS,
they're going to be running them on Solaris, Windows NT and Windows 95.
You can guess on your own which of the two versions in question, Win32 vs.
Solaris, are going to be the most popular.
If you want to through a bunch of development talent behind Java and produce
something worthwhile, make realistic plans and aim for a laudable, not
silly 70's-retro, goal.
What does Java need?
The No List:
------------
An Operating System Written In Java
Exclusionist (e.g., "We hate MS so we're not going to support Java/Win32")
baby-whiner developers who go through life with gun aimed squarely
at foot
The YES List:
-------------
Source code beautifiers
Integrated development environment
Java-centric source code editors
A debugger worth using
A debugger that works
A debugger that works well
Something akin to BoundsChecker or Purify
JavaLint
Install-time pre-translation*
The ability to tag sections of code as performance-critical and translate
on install, retranslate every time the runtime/installtime translator
has been updated)
Persistent storage of translated code, all tagged with CPU id and translator
version
A good Java installer, for both the runtime environment and for installation
of applications
Most developers do without an adequate debugger. There are people who
consider gdb "state of the art." So perhaps the debuggers can wait. But
the last 4 are going to be *critical* if Java is meant to ever be anything
more than a neat way to add applets to people's web pages. *I* want Java
to replace C++ as a mainstay of commercial and private development. That's
not going to happen if performance continues to be as bad as it is and there
exists no way to ensure that performance critical code will be treated as
performance critical.
I'd like to note that for you tots who haven't gotten over your
I-hate-Microsoft-Whine-Whine-Whine penis envy, the last few
would help vanquish one of their key strategic OS strongpoints. Part of
Microsoft's strength is that they get the important stuff (UI, installers,
documentation) done first and the less-important stuff (actual program
code) almost-done second. This may offend some purists, but yes, Windows
and DOS won because they were better in end-user critical ways than the
competition. I hate it, you hate it, but lets get over it and beat them
at their own game instead of sitting in ivory towers crying about how
evil MS is and how unfair it is that the world doesn't recognize the
pure beauty of marvelous tools like EMACS and bash.
If you want Java as a language, and not as some web hacker's toy, to
succeed on a huge scale, the place that it's got to happen is on Windows.
Like it or not, that's where the money is, both for development and from
consumers.
RSR
-
Note to Sun employees: this is an EXTERNAL mailing list!
Info: send 'help' to java-interest-request@java.sun.com
