native JDK on Dragonfly BSD

[Andrew Houghton]

Found the answer (see the end)! But in response..

Greg Lewis wrote:

>On Fri, Jul 30, 2004 at 12:26:51PM -0700, Andrew Houghton wrote:
>  
>
>>I've been playing around with DragonFly BSD for a few weeks, and thought 
>>I'd try compiling a native JDK from ports.  Surprisingly enough (more on 
>>that later) almost everything seems to work.  Admittedly, I'm not doing 
>>deep testing -- but the plugin, the JVM, and javac have handled 
>>everything I've thrown at them, including IntelliJ IDEA, various web 
>>applications, and things like HPJmeter.
>>
>>The one place I *am* running into problems is with Java Web Start.  When 
>>I try starting it I get:
>>
>> Java Web Start splash screen process exiting .....
>> Bad installation. No JRE found in configuration file: No such file
>> or directory
>>    
>>
>
>I assume that it works on FreeBSD?  I must admit I've never tried it.
>  
>
Sure does.. and at least one mission critical part of my work 
environment uses it, surprisingly enough.

>  
>
>>I traced this message down to line 243 of 
>>/deploy/src/javaws/src/share/native/launcher.c; the helpful comment on 
>>line 241 or so says "This should never happen."
>>
>>The closest bug report I found was from Evan Easton:
>>
>>http://lists.freebsd.org/mailman/htdig/freebsd-java/2003-December/001403.html 
>>
>>using his script I was able to get javaws to start, but i still can't 
>>get any of the demo apps to actually run.
>>    
>>
>
>What errors do they produce?
>  
>
It wasn't clear that they did produce errors.. they just never started, 
and the java console didn't give me any info.

>  
>
>>Now, looking at the patchset there's no mention of DragonFly (not 
>>surprisingly).  The reason I was surprised that most things work is that 
>>the patch seems to be making many choices based on system type, and it's 
>>just not clear whether DragonFly defines anything close to the expected 
>>system types.
>>    
>>
>
>I would guess that DragonFly still defines __FreeBSD__ (probably as 4) so
>tha tmost things will just work.
>
>  
>
>>So, I'm poking through the patchset and seeing what I can do, but 
>>someone actually familiar with this stuff (i.e., Greg or Alexey) could 
>>do this about a million times faster than I could, I think..
>>    
>>
>
>Assuming we had a DragonFly system, maybe so :).
>
>  
>
>>Is there any chance of getting DragonFly into the list of "officially" 
>>supported systems for the native JDK?
>>    
>>
>
>Yes.  The easiest way is to do what you're doing and submit patches
>when you're done.  All other ways involve waiting till someone else
>does it :).
>
>I'll try and have a quick look if I get a chance and let you know if I
>see anything obvious, but its going to be a little hard without a DragonFly
>system to test on.
>
>  
>

Martin Hellwig, over on the DragonFly bugs list, found the right answer 
-- simply by switching the OS name in the 
~/.java/.deployment/deployment.properties file from 'FreeBSD' to 
'DragonFly', everything seems happy.  I should have caught this myself 
-- very embarrassing.

With this known, it shouldn't be too difficult to find what needs 
changing where and provide a patch -- I'll see what I can do this week.

- a.