Sam writes:
> Although I am still trying to come to grips with Java, I have hacked
> Matthias and Marcus' vt320 code about a bit and (using Preview Release 1
> of Symantec's Visual Cafe) now have an hp2621.class which will almost
> handle character mode screens with dynamic function keys from Quick.

Sounds cool.

> I would also be interested to hear from anyone who can comment on:
> a) the differences between Java 1.0 and 1.1

1.1 adds significant new functionality and replaces much of the old API
for doing GUI stuff with a better design.  All the old 1.0 code still
runs fine of course.  Only in the last week has a mainstream browser
(Netscape 4.03/4.04) acquired full support for the Java 1.1 features.
This means that if you are trying to write Java "applets" for public
web pages then you probably want to be using only Java 1.0.x features
for widest compatibility.  For standalone Java applications, Server
programs, and IntraAnet stuff, the advantages of the 1.1 features are
significant enough that you probably don't want to use less.  Virtually
all the current Java development environments are 1.1 "only" these days.

> b) the capabilities and features of Java compilers including:
>    . the standard SUN compiler
>    . the HP port available on JAZZ

These are the same thing.  The HP3000 port is the straight Sun compiler
ported to the 3000.  The topic of compilers is rather dull compared to
the topic of the Virtual Machines that execute the resulting code.  The
HP3000 VM includes a JIT or Just-In-Time compiler, so your Java code will
actually run as native PARISC code rather than being purely interpreted.
This doesn't mean that all the performance issues for Java on the HP3000
have been solved yet though.  Note that you only need the HP3000 Java
port if you want to run Java code *on* the 3000.  You can write Java
applets using your favorite Java IDE (Integrated Development Environment)
and serve them to web clients from a web server on the 3000 without
having the Java VM software on the 3000.  Of course being able to write
the server side of your application in Java too can a big advantage, and
it's perfectly possible to write complete applications in Java on the
3000 that don't have any idea of what a web browser or even a network is.

>    . Symantec's Visual Cafe (the current version)

I've been using IBM's VisualAge for Java lately, so I don't have any
personal experience with the 2.0+ versions of Visual Cafe.  I believe
that the Java code in Netscape is from Symantec, and my experience
with the Symantec tools has in general been excellent.

> c) do any of the above, or others you have seen, have the ability to
>    produce both Java pseudo-code and native (WINTEL).exe code?

Yes.  Both Visual Cafe 2.0 from Symantec and Borland's JBuilder (Pro?)
are supposed to have this ability (I'm not positive that both have
shipped this functionality yet).  Supercede from Asymetrix has had this
capability for some time.  While you do end up with a "standalone" .EXE
file, the definition of standalone in these cases may mean multiple
megabytes of required companion .DLL files and .JAR files.

> d) and why, oh why, do Visual Cafe, IE 3.02, Netscape 4.0 and sun's
>    jdk1.1.2 appletviewer all display my code a little differently?

Unfortunately, each of the Java implementations typically will be based
on a different patch level release of the JDK from Sun.  Additionally,
each vendor either tweaks things or the "standards" don't rigidly
require very specific behavior.  Developing a tool that can really
provide cross-platform compatibility is very hard.  Making one that can
make any application look good on every possible platform is practically
impossible.  While newer versions of Java are addressing many of these
issues, writing applications or applets that work on multiple/all
platforms currently requires a lot of testing and experimentation.

G.