Re: External functions and SHLIB_PATH

Thomas,

Thanks for your reply. Incidentally, can you tell me where in the documentation you found this. I searched in the Net8 manual, and PL/SQL manual (under external procedures), but couldn't find any reference to the 'ENVS' parameter.

This does work, however I would now like to extend it to be a little more dynamic. I have the situation where on the one machine I have several 'versions' of my application (eg development, system test, acceptance test). Each version uses external functions and shared libraries from its own bin directory. This strategy you suggested doesn't really lend itself to my situation. What I don't understand is why the SHLIB_PATH environment variable specified in the listener.ora seems to be used, whereas when I set it dynamically it doesn't. Do you know if the load of external functions is performed by the extproc process, or is it another process - which could explain why my dynamically set environment variables don't seem to have any affect.

Thanks again Thomas.

Steve Baldwin

Thomas Kyte <tkyte_at_us.oracle.com> wrote in message news:Wn3nN8KxKMrB4EroRAXIRMeIr8vb_at_4ax.com...
> A copy of this was sent to "Steve Baldwin" <steven.baldwin_at_hancorp.com.au>
> (if that email address didn't require changing)
> On Tue, 21 Sep 1999 10:45:57 +1000, you wrote:
>
> >Hi,
> >
> >I have a situation where I have an external function which makes a call
to
> >another 'C' shared library. I'm working on HP-UX (10.20), which gives a
> >link option to search an environment variable SHLIB_PATH at run time when
> >loading shared libraries. The problem I have is when I build my external
> >functions and then copy them to another machine with a different
directory
> >structure, they fail to load when I reference them from SQL.
> >
> >I then thought it may be because the SHLIB_PATH environment variable of
the
> >extproc process (ie the process that actually executes the external
> >function) was not set up correctly. So, I wrote a small external
function
> >that dumped the Unix environment variables to a file. Sure enough, there
> >was a minimal environment which didn't even contain a SHLIB_PATH
environment
> >variable. Since that process is started by Oracle, rather than my own
Unix
> >session, that was what I was expecting. My next step was to write my own
> >sessions environment variables to a file, and then read that file in an
> >external function, and create the environment variables (using putenv) to
> >mirror my user session. I then figured any subsequent invocations of
> >external functions would use the new environment. Unfortunately I
figured
> >wrong. The environment variables seemed to be set up fine, but it made
no
> >difference. The external function still wouldn't load on the other
machine.
> >
> >Is there any way I can get my external functions to execute on the
'other'
> >machine without copying object files over, and re-linking on that machine
??
> >
> >Your help is much appreciated.
> >
> >Steve Baldwin
> >
> >P.S.
> >Please 'cc' me on steven.baldwin_at_hancorp.com.au as well as posting back
to
> >the group.
> >
>
> Try this:
>
> LISTENER =
> (ADDRESS_LIST =
> (ADDRESS=(PROTOCOL=IPC)(KEY=oracle81.world))
> (ADDRESS=(PROTOCOL=TCP)(Host=sheepdog)(Port=1521))
> )
> SID_LIST_LISTENER =
> (SID_LIST =
> (SID_DESC =
> (GLOBAL_DBNAME=sheepdog.us.oracle.com)
> (ORACLE_HOME= /d01/home/oracle81)
> (SID_NAME = oracle81)
> )
> (SID_DESC =
> (SID_NAME = extproc)
> (ORACLE_HOME = /d01/home/oracle81)
> (PROGRAM = extproc)
> (ENVS=SHLIB_PATH=/x/y/z,DEBUG_FILE=/tmp/test/ext_proc.log)
> )
> )
> STARTUP_WAIT_TIME_LISTENER = 0
> CONNECT_TIMEOUT_LISTENER = 10
> ----------------------------------------------------
>
>
> o (ENVS=SHLIB_PATH=/x/y/z,DEBUG_FILE=/tmp/test/ext_proc.log)
>
> if you want a specific set of environment variables setup for your extproc
> routines, put them here, separated by commas. The above for example, sets
up 2
> environment variables for the extproc servcies, SHLIB_PATH and DEBUG_FILE.
> Since extprocs run detached from a console and are a little touchy to
debug,
> having copious amounts of 'tracing' that you can enable/disable via the
> environment is suggested (by me)...
>
> remember to bounce the listener after doing this change.
>
> --
> See http://govt.us.oracle.com/~tkyte/ for my columns 'Digging-in to
Oracle8i'...
> Current article is "Part I of V, Autonomous Transactions" updated June
21'st
>
> Thomas Kyte tkyte_at_us.oracle.com
> Oracle Service Industries Reston, VA USA
>
> Opinions are mine and do not necessarily reflect those of Oracle
Corporation Received on Tue Sep 21 1999 - 17:47:05 CDT