The biggest difference is what each one returns. system returns the return value of the whatever you executed, whereas backticks capture and return whatever said program printed on STDOUT. This is the property which usually determines which you want to use. There are a few other differences, however; the multiple-argument form of system skips the shell, thus eliminating the risk of having to deal with shell metacharacters, for example.

perl -pe '"I lo*`+$^X$\"$]!$/"=~m%(.*)%s;$_=$1;y^`+*^e v^#$&V"+@( NO CARRIER'

# All similar in functionality
`tar czf /tmp/stuff.tar.gz /stuff`
open ("tar czf /tmp/stuff.tar.gz /stuff");
system ("tar czf /tmp/stuff.tar.gz /stuff");
system ("tar", "czf", "/tmp/stuff.tar.gz", "/stuff");

# Don't search path, specify explicitly
system ("/bin/tar", "czf", "/tmp/stuff.tar.gz", "/stuff");
[download]

Perhaps I'm missing something obvious here. Could you explain to me how this:

system ("/bin/tar", "czf", "/tmp/stuff.tar.gz", "/stuff");
[download]


is different from this:

`/bin/tar czf /tmp/stuff.tar.gz /stuff`
[download]


Is there some inherent advantage to the system call other than those listed by Chmrr above?

Thanks for the clarification.

Vavoom

The Perl Cookbook claims (in the Discussion 16.1) that there is overhead associated with the backticks that does not exist with the system call. I imagine perl does the openpipe/fork combination to get the STDOUT from the backticks; if you don't need the output (as in the tar example above), it is thus more efficient to use system.

-Ton
-----
Be bloody, bold, and resolute; laugh to scorn
The power of man...

The 'system' call treats its first argument as the command to run and the remaining arguments as arguments to that command. Bypassing the shell means that even if the other arguments contained pipe/redirect characters or other commands, those redirects, commands, etc. will not processed. That helps to eliminate a lot of security problems that can be caused by tainted data.

(Of course, that's not to say that you shouldn't run in taint mode and untaint your data anyway.)

Impossible Robot

I think it's almost always better to use the perl builtin or a module than a system call. First, it makes your code more portable. What if the script needs to be run on an OS without a "mv" command? Second, it's probably a bit easier to do error handling when calling a perlfunc than having to trap (and sometimes fiddle with) the return value of system. Given the plethora of great modules such as Archive::Tar, File::Copy, Compress::Zlib, Archive::Zip, etc. I find that I very rarely need to do a system call. There are times when you need to call an OS specific or some other 3rd party program but I find those times to be very rare.

Whenever I find myself using system I always ask myself "Self, is there an all-perl way to do this?" It's saved me time more than once when I had to port the script to a different OS.

unless (open (CMD, "$anycommand |")){
  die "Error in executing $anycommand: $!\n";
}
else {
  my @result = <CMD>;
  close (CMD);
}
[download]

Best regards,
perl -e "$_=*F=>y~\*martinF~stronat~=>s~<A HREF="/index.pl?node=%5E%5Cw&lastnode_id=1072">^\w</A>~~g=>chop,print"