current doesn't build jdk14.

[Poul-Henning Kamp]

In message <20040810222327.GB16802 at misty.eyesbeyond.com>, Greg Lewis writes:

>> + export OPERA_DIR OPERA_PLUGIN_PATH PATH
>> + LD_LIBRARY_PATH=/usr/local/share/opera/bin:/usr/local/jdk1.4.2/jre/lib/i386:/usr/local/jdk1.4.2/jre/lib/i386/native_threads:/usr/local/jdk1.4.2/jre/lib/i386/client:/usr/local/jdk1.4.2/jre/lib/i386/classic:
>> + export LD_LIBRARY_PATH
>> + eval exec '/usr/local/share/opera/bin/opera' --binarydir '/usr/local/share/opera/bin' 
>> + exec /usr/local/share/opera/bin/opera --binarydir /usr/local/share/opera/bin
>> opera: Could not initialize spell checker interface. File not found or could not be opened (-7)
>> opera: /usr/lib/libpthread.so.1: Undefined symbol "_kevent"
>> 
>> The jdk is obviously built against libpthread.so.1 and therefore need the
>> _kevent system call which the 4.x libc doesn't have...
>> 
>> I've tried to preload libc_r.so.5 but that doesn't work.
>> 
>> Any good ideas ?
>
>It looks like the best idea would be to use a jdk14 built on 4.x.  That
>_should_ run under 5.x now that the libm compatibility issues have been
>resolved.
>
>I'm not convinced this will be successful though.  For example, I am
>using a 4.x machine and I can run Opera, it claims the JDK I point
>it at is valid, but no applets will run.

yeah, I think their test is just for the precence of some particular
file, and I belive there are two different places which match that
test.

-- 
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.