in reply to RE: Re: DBI, DBD::Oracle and LD_LIBRARY_PATH
in thread DBI, DBD::Oracle and LD_LIBRARY_PATH

Thanks, this sparked a recollection and I got a solution working that didn't require exec() or messy wrappers. In fact, it is desirable in context of this code to put the "use DBD::Oracle;" in a module, so the exec was questionable for me anyway...

Use a .conf file and ldconfig. RedHat and others support this nowadays. (Google knows more about this than I do, so I leave the research to the reader.) Now on my "standard platforms" I have a .conf file listing the various LD_LIBRARY_PATHs I use, including the standardized path to Oracle client, and did ldconfig to reload that cache, and all is right with the world. No fussing with wrappers or messy execs.

Now if only we could prod Perl to force a reload of that cache (and subsequently its own), we wouldn't have an OS dependency...

  • Comment on Re^3: DBI, DBD::Oracle and LD_LIBRARY_PATH

Replies are listed 'Best First'.
Re^4: DBI, DBD::Oracle and LD_LIBRARY_PATH
by Anonymous Monk on Nov 01, 2012 at 11:11 UTC

    Now if only we could prod Perl to force a reload of that cache (and subsequently its own), we wouldn't have an OS dependency...

    Seeing how the entire concept of LD_LIBRARY_PATH is already OS dependent and flawed, perl should stay far away from it, and let the OS manage its business