Very good results with 0.8.3

[Pierre Guerrier]

We just updated to 0.8.3 which was really important given the huge performance gains of Java 1.4.2 update 1 (up to 10x over plain 1.4.2 on some tasks in our applications).

Everything seems to work fine after 3 minutes of testing (with Mozilla 1.6 and Firefox 0.8).

Even better, making JPI absolutely super-fantastic-psychedelic: it fixes some applets that were broken by update 1 in Safari !!

Example: go to www.boursorama.com, the leading stock-quote website in France. They have a new Java applets for dynamic plotting, e.g. Intel quotes at http://www.boursorama.com/graphiques/graphique_histo.phtml?symbole=INTC

This applet is 1.4.x only, required Safari to load. Since Update 1, it became broken, console displayed:
java.lang.ClassFormatError: StartApplet (Bad magic number)
    at java.lang.ClassLoader.defineClass0(Native Method)
    at java.lang.ClassLoader.defineClass(ClassLoader.java:539)
    at java.security.SecureClassLoader.defineClass(SecureClassLoader.java:123)
    at sun.applet.AppletClassLoader.findClass(AppletClassLoader.java:157)
    at java.lang.ClassLoader.loadClass(ClassLoader.java:289)
    at sun.applet.AppletClassLoader.loadClass(AppletClassLoader.java:123)
    at java.lang.ClassLoader.loadClass(ClassLoader.java:235)
    at sun.applet.AppletClassLoader.loadCode(AppletClassLoader.java:561)
    at sun.applet.AppletPanel.createApplet(AppletPanel.java:617)
    at sun.plugin.AppletViewer.createApplet(AppletViewer.java:1946)
    at sun.applet.AppletPanel.runLoader(AppletPanel.java:546)
    at sun.applet.AppletPanel.run(AppletPanel.java:298)
    at java.lang.Thread.run(Thread.java:552)
... when loading the page !

But it works fine in Firebird and Mozilla with JPI ?

Steven, have you got any explanation for this miracle ?

[Steven Michaud]

I'm very glad to hear of your good results ... and
that the Java Embedding Plugin works with a site
that breaks Safari.  (I've confirmed what you
report with Mozilla 1.7.2 on Mac OS X 10.2.8 with
Java 1.4.1, with Mozilla 1.7.2 on OS X 10.3.5 with
Java 1.4.2 Update 1, and with Safari 1.2.3 on OS X
10.3.5 with Java 1.4.2 Update 1.)

No, I don't believe it was divine intervention :-)

I agree that Java 1.4.2 Update 1 was a substantial
improvement over Java 1.4.2 "plain".  But under
the hood (where ordinary users and developers
can't see it), Apple's implementation of Sun's JVM
has recently been undergoing major changes -- the
change to Update 1 was a particularly big one.  So
it's not too surprising that the latest change
broke some things.  It did (of course) also break
things in previous versions of the JEP ... though
it seems I've been lucky enough to fix all of
these things in version 0.8.3 :-)

Your bug report on Safari is interesting to me for
a completely unrelated reason:  I've had several
reports of the "bad magic number" error from
people trying to use specialized "console" applets
with the Java Embedding Plugin (the Citrix and
WebLogic "consoles").  These reports are very
puzzling, because "bad magic number" usually means
a corrupt class file or an attempt to load
something that isn't a class file -- but there's
no reason to think that either of these things is
happening.  So your Safari bug report may help me
figure out those other problems.