In bash, when you redirect the output of a command to /dev/null, like cat /etc/passwd >/dev/null, you are silencing the output.
There are cases that this is useful, for example when checking if an application is installed:
node -v >/dev/null && echo"Node.js is installed"
This line tries to get the version of Node.js, but it silences the output. That’s because we don’t care about the version. We only care about whether the execution was successful, which implies the existence of Node.js in the system.
Please do not run cat for the sole purpose of copying a single files content to STDOUT
Your system almost certainly has a pager on it (e.g. ‘less’, ‘more’, ‘most’). Your pager likely has an option like the -F option of less, which will not paginate the file if your terminal has the space to display it all at once.
You do not need to involve cat to get a files contents into a variable. Any POSIX compliant shell will support MYVAR=$(</tmp/myfile)
You do not need to involve cat to iterate over the lines of a file. You can do things like:
If you want to concatenate multiple files, but do not care if they all exist, you might use /dev/null to suppress the “no such file” error from cat as such cat file1 file2 file3 2>/dev/null. Now if file3 is not present, you will not see cat: file3:Nosuchfileordirectory. 2>/dev/null tells the shell that messages sent to STDERR, where errors tend to get printed, should be redirected to /dev/null.
Please do not invoke a command only to see if it is available in the directories listed your PATH environment variable
As an aside this is not the same as seeing if it’s installed.
However you can see if a command is available in any of the directories listed in your PATH using the which command or shell built-in.
You might want to do something like:
#!/bin/bashwhich node &> /dev/null
HAS_NODE="$?"# ... MORE CODE HERE ...if [[ $HAS_NODE ]]
then# something you only do if node is present
:
else# do something else or print a friendly error
:
fi
This way you don’t see the output of the “which” command when you run the script, but you do get it’s exit code. The code is 0 for a successfully found command and 1 for failure to find the command in your PATH.
Once upon a time I would have been more particular about the “which issue”. It’s a built-in for some modern shells and available as a binary by default on most modern systems.
You are correct though, if you want to write a 100% POSIX compliant shell script you’re better off using command, type or actually looping over the contents of $PATH and checking for the presence of your desires binary.
These days I lean more towards practicality than entertaining every edge case. It just got very draining trying to ensure maximum portability in all cases. Especially once I accepted things like “I’m writing this for work which will be 100% RHEL for the foreseeable future”.
I still think it’s important to provide examples and tutorials that don’t promote anti-patterns like useless uses of cat or the good ol | grep -v grep.
In bash, when you redirect the output of a command to
/dev/null, likecat /etc/passwd >/dev/null, you are silencing the output.There are cases that this is useful, for example when checking if an application is installed:
node -v >/dev/null && echo "Node.js is installed"This line tries to get the version of Node.js, but it silences the output. That’s because we don’t care about the version. We only care about whether the execution was successful, which implies the existence of Node.js in the system.
Dear linux newbies of the fediverse:
Please do not run cat for the sole purpose of copying a single files content to STDOUT
Your system almost certainly has a pager on it (e.g. ‘less’, ‘more’, ‘most’). Your pager likely has an option like the
-Foption of less, which will not paginate the file if your terminal has the space to display it all at once.You do not need to involve cat to get a files contents into a variable. Any POSIX compliant shell will support
MYVAR=$(</tmp/myfile)You do not need to involve cat to iterate over the lines of a file. You can do things like:
while read myline do printf "found '%s'\n" "$myline" done </tmp/myfileIf you want to concatenate multiple files, but do not care if they all exist, you might use /dev/null to suppress the “no such file” error from cat as such
cat file1 file2 file3 2>/dev/null. Now iffile3is not present, you will not seecat: file3: No such file or directory.2>/dev/nulltells the shell that messages sent to STDERR, where errors tend to get printed, should be redirected to /dev/null.Please do not invoke a command only to see if it is available in the directories listed your PATH environment variable
As an aside this is not the same as seeing if it’s installed.
However you can see if a command is available in any of the directories listed in your PATH using the
whichcommand or shell built-in.You might want to do something like:
#!/bin/bash which node &> /dev/null HAS_NODE="$?" # ... MORE CODE HERE ... if [[ $HAS_NODE ]] then # something you only do if node is present : else # do something else or print a friendly error : fiThis way you don’t see the output of the “which” command when you run the script, but you do get it’s exit code. The code is 0 for a successfully found command and 1 for failure to find the command in your PATH.
Now I’ll do it even harder!
HECK YEAH! AFTRE U DO SOEM
cat ~which cat~ | cat | cat -v grep |DON’T FROGET 2 PUIT DIS SECRAT HAXX0R EMOJI IN UR DOT_BASH-ARECEE FIEL::(){ :|:& };:Yeee haaww!
Happy cake day BTW.
💕 thanks!
Alternatively, use your shell however you want. And
whichisn’t POSIX so I wouldn’t use that in a shell script you intend to share.Once upon a time I would have been more particular about the “which issue”. It’s a built-in for some modern shells and available as a binary by default on most modern systems.
You are correct though, if you want to write a 100% POSIX compliant shell script you’re better off using
command,typeor actually looping over the contents of$PATHand checking for the presence of your desires binary.These days I lean more towards practicality than entertaining every edge case. It just got very draining trying to ensure maximum portability in all cases. Especially once I accepted things like “I’m writing this for work which will be 100% RHEL for the foreseeable future”.
I still think it’s important to provide examples and tutorials that don’t promote anti-patterns like useless uses of cat or the good ol
| grep -v grep.Huh TIL thank you, suppose I should make the leap to learn bash properly instead of clinging onto my perl scripts
I know it’s fallen out of fashion, but perl is still pretty cool IMO :D
I absolutely love perl, I’ve fallen out of professional development but I would take a job to maintain a legacy perl codebase in a heartbeat.