[fedora-java] Re: java issues on ppc

[Andrew Haley]

Orion Poplawski writes:
 > Robin Green wrote:
 > > It's tricky - but possible - to debug interpreted bytecode. JDWP protocol
 > > support is being worked on, but it's not ready yet.
 > > 
 > > Is plplot multithreaded? I've just discovered a thread-safety bug in the
 > > gij interpreter which has lain unfixed for several years:
 > > 
 > >  http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23367
 > > 
 > > I'm told that this bug only affects the interpreter.
 > > 
 > > There are also a couple of other threading bugs that I know of (one of
 > > which is not so easy to reproduce on x86 but easy to reproduce on x86-64,
 > > FWIW). But both of those are not specific to interpreted code.
 > > 
 > 
 > Well, plplot is compiled with threading, but I'm not sure it make much 
 > use of them.  I'll try compiling without threading to see if that makes 
 > a difference.
 > 
 > > If you want to join the #fedora-java IRC channel on
 > > irc.freenode.net, I can try and help you debug it over IRC.
 > >
 > 
 > I'll try soon...
 > 
 > > What kind of problems are you experiencing on ppc, when running plplot as
 > > interpreted?
 > > 
 > 
 > Basically, I've got the following call path:
 > 
 > main:
 > pls.w3d( 1.0, 1.0, 1.0, -1.5, 1.5, -0.5, 1.5, zmin, zmax, alt[k], az[k] );
 > 
 > ->
 > 
 > package plplot.core;
 > 
 > class plplotjavacJNI {
 >    public final static native void plw3d(double jarg1, double jarg2, 
 > double jarg3, double jarg4, double jarg5, double jarg6, double jarg7, 
 > double jarg8, double jarg9, double jarg10, double jarg11);
 > }
 > 
 > 
 > If I print out the values of the arguments in plplotjavac.plw3d, some of 
 > them are bogus (i.e. basex = 0.0 when it is passed 1.0).  Others are okay.

Ah, right.  This m,ight well be a libffi problems that is specific to
PPC.

Andrew.