> . . .
> I added the import visad.*; into all sources in the visad.util package.
> But I think I just had set my CLASSPATH wrong, so this might really not be
> necessary. (I had it at "..", not the actual directory name.)
> If the Makefile would be in the root directory, you just would need a "." in
> your CLASSPATH and could compile everything from there?
> Or set it in the Makefile? Well, I see that you have
> to go through the directory tree for native code.
> The test program in ucar.tests needs to change the package to ucar.tests, not
> ucar.test .
> I think I can't use native code programs with the linux jdk on FreeBSD,
> can I? There would be a lot to change. And the FORTRAN programs don't compile
> nicely with f2c (complains about variable declarations after DATA blocks)
> And the hdf package I got from the FreeBSD ports seems not to be compatible.
> For me it works quite well now. But if I have to do all the changes every
> time a new release comes out, it is quite a lot of work ;-)
> It would be nice to include the date in the filename, so that you know
> whether it's newer than the one you already have.
If your CLASSPATH includes the parent directory of visad and '.'
then it should not be necessary to add 'import visad.*;' to any
classes in the VisAD source distribution.
Someone from Unidata will need to address the issue of the test
program in ucar.tests.
If come up with mods to our Fortran programs that use #ifdefs
to compile on multiple platforms (Solarix, Linux, FreeBSD) we
will look at making them part of our distribution.
The visad.data.hdfeos package depends not only on HDF native
code, but also on HDF-EOS native code that I believe is not
supported on FreeBSD (a definitive answer to thsi will have
to come from Tom Rink, who just left for two weeks vacation).
Bill Hibbard, SSEC, 1225 W. Dayton St., Madison, WI 53706
hibbard@xxxxxxxxxxxxxxxxx 608-263-4427 fax: 608-263-6738