Help using the pipe symbol

[zedan85]

So i read that pipe symbol helps redirect the output of one command as an input to another command.. like

ls grep 'Something'

Anyways i know the command which dispays a directory of the utility
like
which gcc would display
usr/bin/gcc

and

dirname 'usr/bin/gcc' would display
usr/bin

So how can i mix those commands up so that the output of which command serves as an input to the dirname command . I tried the following combinations but couldnt get it to work

[zedan@localhost bin]$ which gcc | dirname 
dirname: missing operand
Try `dirname --help' for more information.
[zedan@localhost bin]$ dirname | which gcc
/usr/bin/gcc
dirname: missing operand
Try `dirname --help' for more information.

Any suggestions ???

[atreyu]

you've almost got it, but in this case, you don't actually need a pipe.

The dirname command wants an argument passed to it (i.e., a filename with a path), not STDIN. So you'd go:

dirname `which gcc`

or, alternatively:

dirname $(which gcc)

the 2nd form is better as it can nest.

[jayd512]

A pipe acts kind of like a connector for two programs.
Linux pipe tutorial | Linux basic configurations
Tips For Linux - Pipes - Get the most out of your shell

[zedan85]

I did read some tutorials on the Linux pipes. However the $ thing seems to make sense. This might sound a bit weird since i am asking this question without doing any kind of research but how do u we know know when to use a pipe or a $. The $ makes sense since it just treats it like a command

dirname $(which gcc)

however the pipe seems to confuse me like
ls | grep 'SearchString'

does placing a pipe after a command make it an input or what ??

[atreyu]

When you add "|" after a command, you are piping the program's output to the STDIN (standard input) of another command.

So when you do "ls | grep blah", you are running "ls" and sending it's output to the STDIN of grep. You can do this b/c grep accepts STDIN as a valid source to act upon (grep can also look in a file, of course). Lots of commands take STDIN in, but not all. When in doubt, read the manpage of the command, e.g. "man grep" (and look for STDIN or standard input).

[zedan85]

@atreyu Thanks that clears a lot..
Since grep states

grep  searches the named input FILEs (or standard input if no files are
       named, or if a single hyphen-minus (-) is given as file name)

I believe standard input here is STDIN

[atreyu]

you believe the truth.

STDIN = standard input (can be thought of as the stuff you type on your keyboard that gets sent to the computer)
STDOUT = standard output (i.e. what gets displayed when you run commands, send to the terminal by default)
STDERR = standard error (also output generated by programs, but sent to a special file handle, also defaults to terminal)

In bash, you can send STDOUT to a file like this:

ls > ls.log

This will redirect the output of contents of the current directory to the file "ls.log". However, say you run "ls" on a file that does not exist. The "ls.log" file created by the above redirection would be empty and you'd see errors on the terminal, e.g:

[root@localhost ~]# ls /foo > /tmp/foo.log
ls: /foo: No such file or directory

you can send STDOUT to one file and STDERR to another like this:

ls /bin /nobin 1> stdout.log 2> stderr.log

you can send both STDOUT and STDERR to the same file with:

ls > output.log 2>&1

The use of the '>' is called a redirection, as in, you are redirecting the output (either STDOUT or STDERR) to a file descriptor other than the default.