Why use eval here? To simply run a command and check its return code, I would use something like:

my $out = `$command`;    # $out contains the command stdout
my $rc  = $? >> 8;       # $rc contains command return code
[download]

I think eval might not be the right tool to use here. In Perl, eval takes a string (or block) of Perl code and executes it. If you put a backtick expression where it's looking for a string, it's going to take the output of the shell command in the backticks and then try to execute it as Perl. Here's a demonstration:

sub whoops {
    print "whoops() called\n";
}
eval `echo whoops`;
print $@ ? "EVAL ERROR: $@\n" : "no error!\n";

eval `echo no_such_sub`;
print $@ ? "EVAL ERROR: $@\n" : "no error!\n";
__END__
whoops() called
no error!
EVAL ERROR: Bareword "no_such_sub" not allowed while "strict subs" in 
+use at (eval 2) line 1.
[download]

I'm guessing that what you want to do is:

1. Run an external command.
2. Get its output.
3. Find out if it died (gave an abnormal exit status).
4. Find out also if it spit out a warning (output anything on STDERR).

I think the responses in Parsing STDERR and STDOUT at the same time might help here. I think that IPC::Run can do all of this, but I haven't used it myself.

For just perl's internal warnings and registered warnings (which don't seem to have caught on), use warnings FATAL => "all";