On 06/01/15 09:13, Ade Lee wrote:
In the past, I suspected that we needed to do these kinds of paths so
that the JVM could find the C code for the JSS JNI modules.  Is that no
longer true?
Once upon a time, there were some proposed changes for JNI:

All of this was undone as of Fedora 19.

Consequently, JSS attempts to load its 'C' code portion using the following java code:

The changes that I made only remove the LD_LIBRARY_PATH, not the CLASSPATH which must still be specified.

IIRC, even earlier in the history of this product, there was code that required environment variables to locate the locations of various libraries, and as the product existed on platforms other than Linux, not all of them used 'LD_LIBRARY_PATH' for this purpose (e. g. - AIX used LIBPATH, HP-UX used SHLIB_PATH, and Windows used PATH).

I do not see any tests below that necessarily test out JSS JNI
functionality.  For example, one test might be to use pki to create a
client NSSDB.
Sorry,

I had also done the following test (but forgot to attach it to the original email) using the changed 'pki' script:
If we still do need this, it would be nice to put in paths for Debian as
well, so that they do not need to patch this.

Ade

On Fri, 2015-05-29 at 18:02 -0600, Matthew Harmsen wrote:
Please review and test the attached patch out on platforms other than
'x86_64' which addresses this issue:
      * PKI TRAC Ticket #1392 - Remove i686/x86_64 architecture
        limitations (e. g. - ppc64/ppc64le)

I did apply the attached patch and build it on an x86_64 machine, and
successfully tested out the following:


      * built, installed, and tested out a CA
      * built, installed, and tested out a CA console
      * built, installed, and tested out a TKS and TPS
      * built, installed, and tested out tpsclient
      * AtoB and BtoA
_______________________________________________
Pki-devel mailing list
Pki-devel@redhat.com
https://www.redhat.com/mailman/listinfo/pki-devel