Message5537

From IRC:

17:43  wmeissner> sabi: yes - the problem is that the jaffl.jar and
                  jnr-posix.jar in jython are more recent than the ones in jruby
17:43  wmeissner> well, in jruby-1.4 anyway
17:43  sabi> sure, but the idea would be that jython could use jarjar to
             maintain its own copy of jaffl/jnr-posix so it doesn't conflict
17:43  sabi> is there any reason that couldn't happen?
17:44  sabi> or would the native bits conflict?
17:44 * sabi really doesn't know too much about how these things work
17:44  wmeissner> the native bits should be ok ... jaffl might be tricky, it
                  loads its provider via reflection, so that might break
17:45  wmeissner> and unfortunately, you can't jarjar rename jffi at all, since
                  it has jni methods
17:46  sabi> yargh. ok.
17:46  sabi> is jffi at least relatively stable? :)
17:46  wmeissner> I'm really not sure why renaming all the stuff in /jni makes 
                  any difference to that bug
17:46  sabi> i think it's causing jython to not be able to find jni any more
17:47  wmeissner> jffi should have a version 1.0 out soon
17:47  sabi> er jffi or whatever
17:47  sabi> so it probably never even tries to load it
17:47  wmeissner> and given that 0.6 has been stable for 6 months or so, I'd 
                  say it should be ok for another 6 ... or 12 .. or more 
                  hopefully
17:47  sabi> ok. so hopefully the next time jruby and jython release, 
             everything should be good
17:48  wmeissner> yes.  I'm trying to round up people to compile the stubs on 
                  $favourite_platform

So hopefully this will be resolved before we release 2.5.2.  Sorry about the inconvenience.