/tmp>cat shebang.pl
#!/no/such/perl -w
use strict;
use warnings;
print 'Hello\n';
/tmp>chmod +x shebang.pl
/tmp>./shebang.pl
-bash: ./shebang.pl: /no/such/perl: bad interpreter: No such file or d
+irectory
/tmp>
Interpreter not executable:
/tmp>cat shebang.pl
#!/etc/passwd
use strict;
use warnings;
print 'Hello\n';pstree -al
/tmp>chmod +x shebang.pl
/tmp>./shebang.pl
-bash: ./shebang.pl: /etc/passwd: bad interpreter: Permission denied
/tmp>
Shebang not identified as such:
/tmp>cat shebang.pl
#!/usr/bin/perl -w
# ^- note: Shebang on second line, first line is empty!
use strict;
use warnings;
print 'Hello\n';
/tmp>chmod +x shebang.pl
/tmp>./shebang.pl
./shebang.pl: line 5: use: command not found
./shebang.pl: line 6: use: command not found
./shebang.pl: line 8: print: command not found
/tmp>
This also happens when there is an (invisible) Byte Order Mark (BOM) at the start of the file. What happens here?
The shell attempts to start ./shebang.pl via fork() and exec(), as usual. This will fail, because the kernel can't identify the file as binary executable or script with some interpreter (first two bytes aren't #!). At this point, a legacy mechanism in the shell kicks in. In the very early days of Unix, shell scripts did not have to have the shebang. If you chmod +x any text file and try to run it, but exec() fails, the shell will attempt to read it as a shell script. A little experiment shows this:
/tmp>echo 'pstree -Aal' >> shebang.pl
/tmp>./shebang.pl
./shebang.pl: line 5: use: command not found
./shebang.pl: line 6: use: command not found
./shebang.pl: line 8: print: command not found
init
.
.
.
|-sshd
| `-sshd
| `-sshd
| `-bash
| `-bash
| `-pstree -Aal
.
.
.
Update:
A little trick to ensure you don't run into the BOM trap:
/tmp>perl -e '$_=<STDIN>;print ord $_' < shebang.pl
10
/tmp>
This test must return 35 or the script does not start with the shebang. 10 is a LF, the end of the first line. 13 is CR, the end of the first line with MS-DOS or classic Mac line-endings, 32 is a space, 9 is a tab. Byte Order Marks may show up as 239 (UTF-8), 254 (UTF-16 BE), 255 (UTF-16 LE or UTF-32 LE), 0 (UTF-32 BE), 43 (UTF-7), 247 (UTF-1).
Update:
See also Re^2: Shebang behavior with perl
Alexander
--
Today I will gladly share my knowledge and experience, for there are no sweeter words than "I told you so". ;-)
|