Frequently Asked Question

What is the difference between cat file | grep foo and grep foo file?

Functionally the two pipelines produce the same lines, but cat file | grep foo forks an extra process (cat) and creates a pipe between it and grep. grep foo file skips both: grep opens the file itself, reads it directly, and prints matches. The result is identical but with one fewer process and one fewer kernel buffer in play. The pattern is famous enough to have a name; "useless use of cat" or UUOC, and people have been gently roasting each other about it on Usenet since the 1990s.

That said, there are times when the cat form is genuinely useful. It reads left-to-right, which is sometimes clearer when you're building a pipeline experimentally. cat *.log | grep foo glues multiple files into one stream cleanly. And tools that only read from stdin and don't take filenames need something to pipe into them. The general principle is: prefer the file argument when the next command accepts one, but don't lose sleep over a stray cat.