Why should piping be avoided in bash scripts?

Question

Most of the time I see someone suggesting using a pipe in a bash script there is someone pointing out not to use it and instead use only one command.

Example:

find $dir -name $pattern

instead of

ls $dir | grep $pattern

Is there another reason than look to avoid pipe?

Answer 1

There is nothing wrong with piping per se. What should be avoided is useless fork()ing, meaning that starting a process is a relatively time-consuming thing.

If something can be done in one process, that is usually better than using two processes for the same result.

Answer 2

Because pipe create a new process. In your example, ls and grep are two processes and find is one. One or more pipes makes command slower. One trivial example:

$ time find Downloads -name *.pdf &>/dev/null

real    0m0.019s
user    0m0.012s
sys 0m0.004s

$ time ls Downloads | grep pdf &>/dev/null

real    0m0.021s
user    0m0.012s
sys 0m0.004s