mjhjmtu

I am in process of writing a POSIX shell script.

I have one function, which redirects (almost) all output to standard error stream like that:

# print something to stderr
printf "..." $var 1>&2

It has been pointed out to me that there is a way to redirect all of the function's output with:

# redirect all output to standard error stream
exec 1>&2

keeping this line in the beginning of the function.

If there were no other redirects, I would simply agree probably, but I have there a test for color support:

# check if we have color support
if command -v tput > /dev/null 2>&1 && tput setaf 1 > /dev/null 2>&1

Basically, I don't know how the exec works, and I don't need to know much in detail now, I just need to know if that exec line will interfere with the above color support test anyhow?

exec <redirection> will redirect the current shell's (or subshell's) I/O streams. And since commands started by the shell inherit the shell's I/O streams, those redirections will affect them, with command-specific redirections applied on top.

So, with something like:

exec >/some/file 2>&1
foo >/another/file

• After the exec, the shell's stdout goes /some/file, and stderr also goes to /some/file.


• When starting foo, the shell redirects foo's stdout to /another/file, but stderr is inherited unchanged and will go to /some/file.

In these commands:

command -v tput > /dev/null 2>&1 && tput setaf 1 > /dev/null 2>&1

You're not using the inherited stdout/stderr streams at all. First, stdout is redirected to /dev/null, then stderr as well. So that particular exec line won't affect these commands.