Clarify My doubt on #!/usr/local/bin/perl

kprasanna_79 has asked for the wisdom of the Perl Monks concerning the following question:

Replies are listed 'Best First'.
Re: Clarify My doubt on #!/usr/local/bin/perl by holli (Abbot) on Jun 01, 2005 at 05:16 UTC
Is it necessary to put this line while doing code in windows platform. It is not necessary. The perl interpreter to be used in Windows is determined by the file associations of the system. But it is not completely useless. You can pass switches to perl via the shebang line like `#!perl -w` [download] which switches warnings on. holli, /regexed monk/	[reply] [d/l]
Re: Clarify My doubt on #!/usr/local/bin/perl by tlm (Prior) on Jun 01, 2005 at 05:24 UTC
The shebang (or bang) line is not specific to perl. In, say, Bourne shell scripts you see this on the first line: `#!/bin/sh` [download] See this article for a quick description. I imagine that a lengthier description would require delving into the history of shells and interpreters, but there is no profound significance to the "!". The term "bang" comes from equating the exclamation symbol "!" with a loud noise. "shebang" is American slang; it's used in the expression "the whole shebang", meaning "the whole thing". As far as I know its meaning has nothing to do with interpreters; it just sounds similar to "bang", so it got adopted by extension, I imagine. the lowliest monk	[reply] [d/l]
Re: Clarify My doubt on #!/usr/local/bin/perl by monarch (Priest) on Jun 01, 2005 at 05:22 UTC
The `#!` combination of characters is known as the "hash-bang". I don't know why it is called bang.. I thought it was an exclamation mark myself, but I suspect an American thought it would be cute to give that character a shorter name. The `#!` characters are interpreted by linux/unix shells to mean "run whatever program follows, then pass the text on all subsequent lines to that program". It's a simple way of constructing scripts in any language, merely by defining which interpreter should be used on the first line of the script. As the previous poster attests, you don't need it in windows. That's because windows is different to linux in that it determines which program to run your file by the file extension (in most cases this will be `.pl` on your system). Alternatively you can run your perl scripts by typing "`perl <filename>`". There is one situation where you must provide the "hash-bang" on scripts in windows - this is in Apache for Win32, when executing perl CGI scripts. Apparently Apache looks for which interpreter to use based on the first line of any CGI scripts.	[reply] [d/l] [select]
Re^2: Clarify My doubt on #!/usr/local/bin/perl by ysth (Canon) on Jun 01, 2005 at 05:43 UTC
I thought it was an exclamation mark myself, but I suspect an American thought it would be cute to give that character a shorter name. Most multisyllabic character names get shortened in various ways. According to the jargon file, "bang" originated at CMU.	[reply]
[OT] Re^2: Clarify My doubt on #!/usr/local/bin/perl by fokat (Deacon) on Sep 20, 2005 at 14:25 UTC
Dear monarch (...)I don't know why it is called bang.. I thought it was an exclamation mark myself(...) A few years back, email and news were sent from server to server using a protocol (and program) called UUCP (Unix-to-Unix-CoPy). This used phone lines to push packets of text files from machines that for the most part, were "connected" to the network intermitently, a few times a day. In those days, email addresses had to include the routing information, telling each system how to forward your message. An email address could have been something like `ucbarpa!dino!usb!lem` (If memory serves well). Each word in the address would be a host/system/site identifier, as known by the corresponding node. The `!` would then instruct said node to "bang" or bounce the file to the next system in line. As you may easily guess, this system was very awkward (sp?) to use. Later it was entirely replaced with SMTP and NNTP for direct transmission of email and Usenet news from server to server, using the Internet as transport. During that transition, things were complex because provisions had to be made for rewriting the addresses to RFC-822 format (ie `user@site`) and for combining RFC-822 and UUCP in the same address, to deal with hybrid or incomplete connections (ie `usb!lem@dino.conicit.ve`). To this respect, Internet is so much simpler now. Google for UUCP for more information. Best regards -lem, but some call me fokat	[reply]
Re: Clarify My doubt on #!/usr/local/bin/perl by ikegami (Patriarch) on Jun 01, 2005 at 05:21 UTC
And for your second question, the "!" differentiates shebang lines from comments.	[reply]
Re: Clarify My doubt on #!/usr/local/bin/perl by gellyfish (Monsignor) on Jun 01, 2005 at 08:44 UTC
The shebang line, contrary to what has already been stated, is not used by the Unix shell but by the `execve` system call - as per the execve(2) manpage: execve() executes the program pointed to by filename. filename must be either a binary executable, or a script starting with a line of the form "#! interpreter arg". In the latter case, the interpreter must be a valid pathname for an executable which is not itself a script, which will be invoked as interpreter arg filename. Windows uses another mechanism based on the extension of the file to determine how to execute the file, however Perl does take note of the shebang line however the program was invoked: if there is a shebang line and it contains "perl" then it will be parsed to determine any additional command line switches that may need to be applied - if `perl` finds a shebang lime that doesn't contain "perl" then it will attempt to re-execute the script using the "interpreter" found there - thus on windows you can create a file with a shebang line `#!notepad` and run it with perl then "notepad" will be started with your "script" open for editing. /J\	[reply] [d/l] [select]
Re^2: Clarify My doubt on #!/usr/local/bin/perl by Tanktalus (Canon) on Jun 01, 2005 at 13:59 UTC
Still a bit off. The shebang line is actually interpreted by the kernel. When the kernel is told to run something, it checks the first few bytes of a file looking for a magic number to figure out how to execute something. The magic number for passing it to an interperter, such as the shell, is whatever the characters '#!' mean on that machine (depending on whether it's big-endian or little-endian). For example, the magic number for ELF-formatted files is whatever ^?ELF is (where ^? is ctrl-?, not literally ^ and ?). The kernel uses this to use the ELF loader to load and execute that file. Other formats may be supported as well, depending on the kernel and architecture, and, on Linux, whether you've compiled in support for other formats.	[reply]
Re^3: Clarify My doubt on #!/usr/local/bin/perl by gellyfish (Monsignor) on Jun 09, 2005 at 10:07 UTC
Er, yes. But it is through the agency of the `execve(3)` system call that this happens - you can see the code for this (for example) in `/arch/i386/kernel/process.c` in the Linux source. Okay in Linux it is further complicated by the `binfmt_misc` stuff, but in general for all Unix based OS `execve` is a kernel function that is exposed via the `syscall` mechanism in the C standard library. /J\	[reply] [d/l] [select]
Re: Clarify My doubt on #!/usr/local/bin/perl by gawatkins (Monsignor) on Jun 01, 2005 at 14:09 UTC
The only case in win32 where a `#!c:\perl\bin\perl.exe` is required would be CGI scripts that run through Apache. Otherwise it is only optional for switches and such. Greg W	[reply] [d/l]


Welcome to the Monastery
	PerlMonks