I'm in the process of writing a test tool in Perl for a product we're developing. I've re-directed the Perl STDERR to a logfile, which is logging Perl errors as expected. However, the application I'm testing writes some error messages to the console (STDOUT). I need to capture any application errors in a file (which I've partially done), because the test tool is designed to run overnight. I've done this by redirecting STDOUT to the log file. The trouble is, once I've re-directed STDOUT to the file, I'm unable to redirect it back to the console as it was before, to avoid sending everything to the log file.

Has anyone got any suggestions on how I could achieve this?

Regards,

Dan.

maybe try this:


Look at perl's help docs on what the select function will do for you and what caveats to look for.

Do you not need to flush the buffers before re-selecting the original filehandle?

Not that I know of, a quick check of perldoc -f select doesn't seem to suggest something. I think he's fairly safe anyways, a quick test that would never of filled the buffer worked for me.

Thanks for the suggestions. I tried them, but unfortunately they didn't quite do what I was trying to achieve. I've managed to get my hands on a copy of the Perl Cookbook where I found this:

#first, dup STDOUT

open (OLDSTDOUT, "&>STDOUT) or die "Couldn't dup STDOUT: $!";

#re-direct STDOUT to log

open (STDOUT, ">> $logfile") or die "Can't open $logfile: $!";

system "$executable";

close STDOUT;

#restore original STDOUT

open (STDOUT, "&>OLDSTDOUT) or die "Can't restore STDOUT: $!";

close OLDSTDOUT;



Regards,

Dan.

Just a small correction, it is >& and there is a missing double quote after the handle name.


open (OLDSTDOUT, ">&STDOUT") or die "Couldn't dup STDOUT: $!";



Kinetik