Shell Programming and Scripting

Below is a chuck of a shell script I have written. The problem I am having is with the test section. I want to test the status of the f60gen command but since I'm running the output of the f60gen command into tee, the test command is actually testing "tee" and not "f60gen". Is there a way to test the second to last command? If not, how can I make this work?

$ORACLE_HOME/bin/f60gen module=$PLL_FILE \
userid=apps/$APPS_PWD \
output_file=$PLX_FILE \
module_type=library \
batch=yes \
compile_all=special | tee -a $LOG_FILE 2>&1

if [ $? = 0 ]; then
log_msg "Successfully generated $PLX_FILE"

Why can't you test for your return code before you do the compile_all?

$ORACLE_HOME/bin/f60gen

if [[ $? != 0 ]]; then
print "F60gen failed" | tee -a $LOG_FILE
fi

I can't do that because "compile=all" is an option for the f60gen command.
I think I found a way though. I just redirected the output from the command to a tempfile then I set a variable = $?.
I can then recall and test the value later and I also can tee tempfile into my proper logfile.

It works.

Rather than creating a temp file that needs to be cleaned up afterward, you can do the same thing as follows:

Code:

{
    f60gen ...
    print variable=$?
} | while read LINE
do
    case $LINE in
        variable*)
            if [[ ${LINE#*=} -gt 0 ]]
            then
                ... report error ...
            fi
        ;;
    esac
done

I have used a ksh coprocess for stuff like this. See exit status of command in a pipe line for an example. But r2007 presented an arguably better solution in Pipelining.

you can use switch command if you want it also.