Unix Pipes in Practice: Commands into Tools

The pipe — | — is the one idea that makes the Unix command line feel like a language instead of a menu. Each command does one small thing well; the pipe wires them together so the output of one becomes the input of the next, and the combination is a tool nobody had to write.

The idea in one line #

$ ps aux | grep ^ana | wc -l

Three commands, none aware of the others, one answer. ps reports everything, grep filters, [wc` counts. Rewrite that in any GUI and it is a feature request; here it is a Tuesday.

Pipelines that earn their keep #

The ranked count, seen across this series in different clothes:

$ awk '{ print $7 }' access.log | sort | uniq -c | sort -rn | head

A cleanup report that shows before touching anything:

$ find . -name "*.tmp" -mtime +30 | tee /tmp/cleanup.txt | wc -l
$ du -ch $(cat /tmp/cleanup.txt) | tail -1

tee splits the stream — a copy to a file, the rest onward — which makes any pipeline auditable.

Why streams beat screens #

A pipeline consumes and produces text, which has three consequences worth internalising:

• Composability: any command that reads lines can join the chain — jq for JSON, awk for columns, xargs to turn lines back into arguments.
• Auditability: insert less anywhere to look at intermediate results, remove it, carry on.
• Scale of attention: you filter before you look, so a million-line log is a non-event.

Every note in this series is secretly about the pipe — from grep to sort and uniq. The whole terminal notes catalogue is on the journal.