Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 3 Dec 1997 20:28:54 -0500 (EST)
From:      "Alex.Boisvert" <boia01@gel.usherb.ca>
To:        chat@FreeBSD.ORG
Cc:        kelly@fsl.noaa.gov
Subject:   Re: What can we do about Java?
Message-ID:  <Pine.SOL.3.91.971203202405.21951A-100000@castor>

next in thread | raw e-mail | index | archive | help
(This is a repost since -chat seems to have filtered my previous posting 
from a non-resolvable address, I'll change my sendmail.cf soon)

On Wed, 3 Dec 1997, Sean Kelly wrote:

> I think it's high time FreeBSD become one of the leading Java runtime
> and development platforms.  Our OS is already positioned quite well in
> Internet environments and on the desktops of many of us old Unix hands. 
> I'm aware that some FreeBSD core team members also do paid Java
> development with FreeBSD.  I'd like to be able to do that as well with a
> high-performance, low-bug Java system.

I have Sun's JDK 1.1.2 running on my system now.  It's "fairly stable" 
with respect to Keith White's JDK 1.1 port.  To give you an idea, I can
run ICQ, the BISS-AWT development environment (LibBrowser), and many other
java applications.  I did not test many applets (via AppletViewer for
instance) since applets are not my priority.   I use my port exclusively
now, so that's a good indicator.

Threads are still giving me some headaches but the stability is improving.
Sadly, I can attribute most of the current bugs to Lesstif.  I've used
version 0.78 to 0.82 and each has its specific weaknesses.

I will be getting a Motif 2.0 license in the next few days (it's in the
mail).  Then, I'll be able to test threads using HotJava Browser, which I
can't run now because of bugs in Lesstif (presumably).


As soon as I can get HotJava to run, I will release my port.  So you can
expect a release soon...  (Before Chrismas?)

> 
> I've got the enthusiasm (if not the talent), and I want to help.  What
> can I do? What can *we* do?
> 

Well, let me paint the picture as I see it.  (Other people's mileage may
vary ;-)

Right now, it's "kinda hard" to work collectively on Sun's JDK port.  I
have had trouble with Sun getting JDK 1.1.3 or JDK 1.1.4... So working
with Keith White on the same JDK version has not been possible yet.  Also,
JDK versions are moving targets and just keeping up with Sun's releases is
a challenge by itself.  Hopefully things will start to settle after JDK
1.2.X.

In the long run, I think Kaffe (www.kaffe.org) is the right way to go. 
Keeping in sync with Sun will still be very hard but at least, we'll have
a clean-room implementation where many people can work simultaneously,
hands in hands.  Kaffe is also our best hope for JIT compilation on
FreeBSD since Sun/JavaSoft will unlikely release a JVM source code
license with JIT engine.

I believe that making a stable JVM (look at FreeBSD for instance :-)  is
the most important thing now.  Performance is a second-degree goal.
Keeping up with all of Java's feature (like exotic APIs) is a third-degree
goal for me.  What I want is a JVM that has a portable GUI, reliable
networking and stable execution.   The rest is luxury for now.

I not saying that we should stop porting Sun's JVM to FreeBSD.  We still
need Sun's JVM as a reference implementation (as it's meant to be) and we
will still depend on their class libraries in the short/medium term (ie. 
Swing components) but to get a *good* JVM on FreeBSD, we'll have to go
*our way*, we'll have to get our hands dirty for a while and come up with
a JVM (or maybe a compiler based on gcc) that rivals Microsoft's or
Symantec's (and many others) JVM.  

For this to happen, Kaffe and/or gcc need a good deal of work and that's
what we have work on to get it "our way" with FreeBSD without sacrifices
(with respect to performance, stability and licensing).

I understand that all of this takes time and efforts.  So much that it'll
take some time (another year, maybe more) before we can compare the "free
software" approach to commercial ones.  Again, it's not as easy as porting
Sun's JVM but, in the end, the reward is greater.

There are many active projects that need volunteers:  Kaffe, gcc/java,
kore (a java.* package clone), java-posix, BISS-AWT, and others.  Take a
look at them and see what you'd like to work on.  What are *your*
priorities regarding java?   (Don't forget to contribute to FreeBSD or
else Jordan will hate me for this one ;-).

Regards,
Alex Boisvert

---
FreeBSD:  Decouvrez la puissance de votre PC!  www.freebsd.org




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.SOL.3.91.971203202405.21951A-100000>