Posted by mauromol
on October 8, 2009 at 8:37 AM PDT
Java Web Start has suffered from multiple problems in almost all of the latest Java 6 updates. We have compiled a list of regressions from Java 6u6 on, up to 6u16, so that our application must enforce the use of 6u5 at most, but sometimes it is not yet enough, because the jws client fails before our JNLP can even intervene to specify the JRE to use...
As of 6u16 there are still two regressions that are hurting our application, one of which is serious and blocking.
The first bug, the most critical, is the following:
- java.applet.AppletContext.showDocument(URL, String) does not work with 1.6.0_13
This bug seems to have been introduced in 6u11 or 6u12 (although in the bug report I've written 6u13 because in that version its severity is even higher) and causes the method java.applet.AppletContext.showDocument(URL, String) not to work at all when running an applet using Java Web Start. This bug is critical for us, because the funtionality is totally broken.
The bug report has an attachment with a simple application that reproduces the problem.
The second bug, which is less critical, is a regression in the sense that it appeared in the past, then it was fixed, but since 6u14 it has been there again:
[REGRESSION] Bug #6533653 is back in Java 6u14
The problem is that if an empty JAR is present in the section of your JNLP, the jws client throws a JNLP exception saying that there's an unsigned application that is requesting unlimited access to the system... but that JAR is actually signed and is not giving any problems at all with the jws client of previous JRE releases. The workaround is to remove the empty (and useless) JAR from the section of the JNLP, but of course this can still be a problem for existing deployments of our application, because the jws client fails before our JNLP can even force the use of a previous JRE.
So, please, go and fix these problems in 6u18 at last!!! We're stuck at 6u5 because of such regression problems that almost every new update introduces, so that Java Web Start technology appears not so reliable from a user perspective...
P.S.: if you're curious to know which regressions I'm talking about, this is the complete list of bug numbers regarding Java Web Start that are preventing us to upgrade from 6u5:
6u6, 6u7, 6u10, 6u11: 6741349
6u11, 6u12, 6u13, 6u14, 6u15, 6u16: 6835450
6u14, 6u15, 6u16: 6850598