Build it without a mouse

You are standing in ~/projects/first-site. Type this.

mkdir -p ~/projects/first-site

Nothing prints. The folder is already there. That is the point.

-p means parents. A parent is a folder that must exist before a child can sit inside it. mkdir -p creates every missing parent on the way. It also stays quiet if the path already exists. That second part is called idempotent: run it twice, same result. The path starts at ~, so the command works from any folder. Read it from the left. ~ is already there. projects may be there. first-site may be there. -p fills any gap. That is why one line rebuilds the project from nothing.

Without -p the same idea fails in two ways. A missing parent prints No such file or directory. A folder that already exists prints File exists. You will rebuild this path later. Keep the one-liner with -p. You do not need the failures.

Now make a scratch file. Leave index.html alone.

touch scratch.txt

touch creates an empty file. It does not open a window. The file just appears. Practise copies, renames, and deletes on scratch.txt. The real page stays put until the project asks you to tear the folder down. Run ls. scratch.txt should sit next to index.html.

Copy keeps the original. Move does not.

cp scratch.txt scratch-copy.txt

cp needs -r before it will copy a directory. A directory is a folder: a named place that can hold files. You are copying a file, so the plain command is enough.

Rename the copy.

mv scratch-copy.txt notes.txt

You now have scratch.txt and notes.txt in this folder. The copy left the original. The rename only changed a name.

mv is two jobs in one command. If the destination does not exist, it is a new name. If the destination is a directory, the source moves inside it.

Try the second job. Make a folder, then move notes.txt into it.

mkdir -p scratch-dir
mv notes.txt scratch-dir

notes.txt now lives inside scratch-dir. It is no longer sitting next to index.html. Same command, different destination, different job.

Tab and the up arrow

You are not supposed to memorise long paths. Type ~/pr and press Tab. The shell should finish projects if that folder exists. Press the up arrow to recall the last command. Edit it. Run it again.

A space in a name splits it. The shell reads my file.html as two words, my and file.html. Quote the whole name so it stays one word.

touch "my file.html"
mv "my file.html" "my notes.html"

Double quotes are enough here. Skip them and the command looks for a file called my that you do not have.

Safety flags exist for the nervous moment. Plain mv will overwrite a file that already has the destination name. cp -i and mv -i prompt before they overwrite. cp -n and mv -n refuse to overwrite. Use them when two names might collide.

Delete is the one that does not forgive.

rm scratch.txt

rm has no trash. The file is gone. There is no undo. The file manager’s trash is a different tool. rm does not use it. An engineer at GitLab once ran rm -rf on the wrong database host and destroyed live production data. Your scratch file is smaller. The rule is the same.

By default rm does not remove directories. rm scratch-dir refuses. The flag -r means recursive: delete the folder and everything inside it. -i prompts before every removal. -f never prompts.

The slash is the other trap. rm -r home/ starts from where you are standing. rm -r /home/ starts at the top of the machine. That leading / is the difference between a local folder and every user’s home. Do not run the second one.

Clean up the rest of the scratch work.

rm "my notes.html"
rm -r scratch-dir

index.html is still here. Check with ls if you want. You did not touch it. The scratch names are gone. The folder is the same one you opened as file://.

You can now rebuild the folder from the keyboard:

mkdir -p ~/projects/first-site

The project will delete this path and ask you to bring it back for real. rm is how it disappears. mkdir -p is how it returns. The page itself is still the file you double-clicked. Next you will put words into it without a mouse.

Mark as read