Checking mail command for success

ronix has asked for the wisdom of the Perl Monks concerning the following question:

Venerable ones, I searched and could not find any obvious information on this. I am using the mail command in a script on AIX (I can't use any of the the slick mail modules for Perl, as it would require updating thousands of production servers). Simply put: I want to continue if the mail command is successful and exit out if it fails.

# subroutine for mailing/transferring the changes
sub mail_changes {
  my $mailer = "|/bin/mail -s $hostn $recipients < $disk_changes";
  open (MAIL, $mailer) || die "Cannot execute mail command; $!\n";
  close(MAIL);
        if ($? > 0) {
                print "Do some error stuff if needed\n";
        }
  }
[download]

Any suggestions for a better way of checking for success other then the old $? ? thanks for your time.

Comment on Checking mail command for success Select or Download Code

Replies are listed 'Best First'.
Re: Checking mail command for success by ikegami (Patriarch) on Apr 21, 2009 at 20:19 UTC
What's the problem? Is `$?` not suitable for some reason? Note that you're subject to attacks by passing unescaped text to the shell.	[reply] [d/l]
Re^2: Checking mail command for success by ronix (Novice) on Apr 23, 2009 at 19:27 UTC
Note that you're subject to attacks by passing unescaped text to the shell. Sorry, but could you explain this in more detail? thanks!	[reply]
Re^3: Checking mail command for success by ikegami (Patriarch) on Apr 23, 2009 at 19:37 UTC
"`John Doe <john@example.org>`" (minus the quotes) is a valid email address, but you'll have problems if `$recipients` contains that. And of course, there's the malicious who might find a way of getting "`\| rm -rf /`" (minus the quotes) into `$host` or `$recipients`. The multiple argument form of various commands gets around that problem. `use IPC::Open3 qw( ); open(my $mailer_fh, '\|-', '/bin/mail', '-s', $hostn, $recipients) or die("Cannot execute mail command: $!\n"); open(my $msg_fh, '<', $disk_changes) or die("Cannot execute message file: $!\n"); print $mailer_fh $_ while <$msg_fh>; close($mailer_fh) or die("Error sending mail message: $!\n"); $? == 0 or die("Error sending mail message: $?\n");` [download]	[reply] [d/l] [select]
Re: Checking mail command for success by mr_mischief (Monsignor) on Apr 22, 2009 at 02:48 UTC
How are you deploying the changes to your program wrapping the mail command? Perhaps you're using ftp, sftp, scp, rsync, or something similar? Chances are any of the pure-Perl mail sending modules could be deployed the same way. Some of those mail sending modules even just wrap a Sendmail mail submission executable (or the mock sendmail executables offered for compatibility by any of Postfix, Exim, qmail, SMail, or a long list of others) usually found at /usr/sbin/sendmail, /usr/bin/sendmail, or (for hysterical porpoises) /usr/lib/sendmail. The trick is, they wrap it properly so you don't have to. They let you just call a simple sub from your Perl program so you don't have to sweat the details. If you really want to wrap an external program, consider a module that does so. If you can't stand the thought of a module, consider grabbing the code from one and crediting the author. That would let you just update the one file on those servers instead of two. I generally use my own interface module around Net::SMTP::TLS and have hacked support for that module into NMS formmail, too. However, you might prefer Mail::Mailer, Mail::Sendmail, MIME::Lite, or Mail::DWIM (which is one of those that will wrap around a Sendmail mail submission executable). Update: fixed a typo in PM link tag for Mail::DWIM	[reply]
Re: Checking mail command for success by kwaping (Priest) on Apr 21, 2009 at 23:06 UTC
Have you tried it using system? That's too bad that you can't use any of the nice CPAN mail modules. --- It's all fine and dandy until someone has to look at the code.	[reply]


Perl: the Markov chain saw
	PerlMonks