WebStart Install Fail
swpalmer at gmail.com
Wed Dec 12 16:13:16 PST 2012
I had 7u10-b18 32-bit already installed when this failed. The fix was to
manually install the 64-bit-bit JRE.
Filed RT-26964 <http://javafx-jira.kenai.com/browse/RT-26964> without using
the word pathetic - that was directed towards the situation not any people,
sorry for that. I think we can all agree that it was not a good user
I can think of only one "weird" thing about my environment that might have
triggered this. I had 32-bit 7u10 installed AND I made sure that the
32-bit java.exe was copied into the Windows System32 directory instead of a
64-bit java.exe. Normally on 64-bit windows the 32-bit java.exe isn't
copied there, but the 64-bit version is. As we use a lot of 32-bit JNI
DLLs the 64-bit java.exe being on the path ahead o 32-bit java causes all
sorts of frustration during development. I'm going to uninstall 64-bit
7u10 now and see if I can reproduce the problem again. It happened
consistently before installing the 64-bit JRE.
On Wed, Dec 12, 2012 at 4:40 PM, John Smith <John_Smith at symantec.com> wrote:
> I don't think this particular issue had anything to do with the browser
> being used or a browser plugin.
> The jfxtras deployment is just an html link to a jnlp file (no use of the
> cause a native browser plugin to be used).
> Due to windows file associations set from a previous jre install, when the
> link is clicked, it will run the web start launcher, from thereon it goes
> awry in Scott's environment.
> Clicking on the same link works fine for me (Win7+jre7u9) and I can use
> the excellent jfxtras demo app, but that's not surprising as I have the 32
> bit jre installed already.
> -----Original Message-----
> From: openjfx-dev-bounces at openjdk.java.net [mailto:
> openjfx-dev-bounces at openjdk.java.net] On Behalf Of Daniel Zwolenski
> Sent: Wednesday, December 12, 2012 12:40 PM
> To: Richard Bair
> Cc: openjfx-dev at openjdk.java.net
> Subject: Re: WebStart Install Fail
> Definitely this is a JFX issue and I don't believe there is anything Xtras
> could do even if they wanted (apart from write their own webstart browser
> I have seen this many times. It is one of 3 reasons why I keep saying jnlp
> doesn't work (the other being the user-off-putting steps of installing Java
> even when it works; and the forced jre auto-updating that can break your
> app at any time after release without the developer changing a thing).
> Igor was aware of this problem (and I think there should be a jira in
> there somewhere) but since he's gone the rough gist of what he said was
> happening is that many browsers are still 32bit even if you are running on
> a 64bit OS. You end up with the browser somehow getting a mix of 64bit and
> 32bit JREs installed/JFX's installed. Hence the error (eg running 64bit jfx
> on a 32bit jre or the inverse). The "fix" generally was "once browsers all
> go to 64bit it won't be a problem" - I'd be inclined to allow the
> "pathetic" adjective at this point.
> As an aside, I did notice recently when hacking at the installer package
> code that regardless of what OS you are on the jnlp gets a 32bit win
> fallback URL for jfx. In theory if it finds the jre but not jfx then it is
> suppose to use this URL. This may or may not be related - I only see the
> packager code that generates the jnlp not the actual plugin code that acts
> on it.
> Have I ever mentioned that deployment in jfx sucks ;)
More information about the openjfx-dev