Is anyone else having problems with Java 1.6 update 12 and Webstart? I just tried it out with Gravitational Fourks and it appears to be completely broken - I’m not even getting a console window, just a splash screen which vanishes after a very short period of time and a process which sits using up half my CPU.
Not only that, but it turns on a semi-broken version of OpenGL acceleration which runs horrendously slow in XOR mode.
For those that don’t know, XOR mode is when you draw the inverted color of what is onscreen, a feature used to select regions in a couple of programs I use.
Really regret downloading this release.
I haven’t had any problems myself. Everything I have tried has worked for me. Though I don’t think I’ve used anything that has used XOR.
What you have un u12, I’ve had for ages in u11.
Webstart is just plain broken. Eventually it works, but I sometimes have to launch a JNLP files 10 (or more!) times before it works.
I thought I was the only one with this problem, as nobody else ran into it.
as always the best what we can do is to file bugs or vote for already filed bugs (e.g XOR perf regression). Thats what early access builds are for.
Since JavaFX is Sun’s new marketing flagship for the desktop, webstart bugs should be taken pretty seriously internally since ws is currently the only deployment mechanism of JavaFX apps and applets.
regarding OpenGL pipeline, i pretty much gave up the hope that we will get a reliable pipeline anytime soon.
I gave up on filing bug reports…
I submitted about 5 or 6. All were closed withing a few weeks… “could not reproduce”.
I mean… it’s like JFileChooser. You can’t even traverse directories, a doubleclick renames the directory, [enter] selects it. It’s been like that for 5 years? People complaining everywhere, Sun is not going to fix it, because they cannot reproduce the damn thing. The Java Plugin is only (more or less) stable since u10. I’ve given up all hope on Sun in the clientside. The JVM is a masterpiece, but everything they release clientside is just buggy, and bugfixes take 5+ years. Webstart never worked reliably, and now it’s entirely broken. Don’t the folks have a Quality Assurance team there? With 20 PCs with the most common combinations of operating systems and browser versions?
They fixed bugs which are actually pretty old (AFAIK the heavyweight/leightweight thingy was around 10 years old).
Almost the whole desktop team works on JavaFX… webstart is currently the only supported way to get javafx running. Its suicide if deployment doesn’t work (They know that). You have to see it from their perspective too - what should they do if they can’t reproduce the bug? Next time you see something very rarely reproducible (with latest update release) you could e.g take threaddumps or zip your entire webstart cache and sent it to them. Race conditions are really a pain in the ass and very difficult to track…
Regarding the filechooser. I heard at last J1 that they plan a rewrite for JDK7, not sure if it is still up2date since Sun runs currently with reduced resources.
I talked to some devs about the browser embedded notifications for applets as alternative to the notification dialogs and they think about including it in 7 and will provide a workaround for javafx applets (for some scenarios) since they can’t implement it in a jre update release.
Sun changed priorities.
For starters, they shouldn’t close the bug immediately.
I mean, if there a dozens of developers complaining, they should be taken serious, even if you cannot reproduce.
I’m fairly disgusted about all resources put at JavaFX: a technology that nobody really is waiting for, and given Suns trackrecord, will finally be stable in 5 to 10 years. We need SERIOUS bugfixes in Swing, where components fire double events, or render totally wrong graphics, even in the DirectX pipeline (I just wasted 1 entire day at work tracking it down) under very specific conditions, in very specific versions.
i agree, in this case this wasn’t very nice.
How do we know whether something’s rarely reproducible or not? The correct action in case you can’t reproduce is to ask for more details, not close immediately.
Actually Sun’s testing is tip-top. I saw something on the web about they’re batteries of computers which they test the jre’s on in Russia.
Sounds like webstart is being left out of the tests though…
Riven did you try this?
http://www.shankh.com/2008/12/14/java-web-start-jnlp-splash-recv-failed/
If that’s the problem, this is too strange a fix for the casual user. Shouldn’t JWS just work with or without firewalls?
:o Uninstalling Comodo Firewall now… I was not too happy with it anyway.
Update: problem solved.
is the problem the update or the firewall then?(I dont have 1.6 12, or this problem for that matter)
just curious
hope oyu guys can work it out.
So, after all that Sun’s client team bashing the problem turned out to be caused by third party software, huh?
Anyway, FYI, the XOR perf bug has been fixed - or, rather, worked around - in 6u14 (the build with the fix isn’t out yet).
Dmitri
While the “Splash recv failed” error is a little better than “General error 38”,
I think it’s fair to say that almost no users are going to interprete it as meaning “Unable to connect to host, please check your internet and firewall connection settings”.
Case in point;
It’s taken 1 tech. guru searching forums to find the solution, another to link his blog onto a Java developers forum to help a 3rd resolve the problem.
That’s 3 tech gurus more than the average casual web gamer is going to have access to.
What’s on my mind is:
Is the out-of-process Java Plugin sometimes stopped by firewalls, too?
Or is this a bug in the splash screen code of JWS?
While it may seem unfair to you that I was ‘bashing’ you for something not your fault, it seems unfair to me, that now that one bug isn’t caused by the Java client team, suddenly all my other instantly closed bugreports are discarded in the same category? I mean, there are very sound reasons for my so called ‘bashing’.
My problems aren’t caused by a firewall, because I don’t have one. (I run Linux, before anyone tells me that my computer is probably zombied).