Shell from zero: a function is the reusable piece

A function is a name for a few commands. You call it like a program. $1 is the drink. $2 is the price in cents. The function does not have typed parameters. It has positions, and the quotes from lesson 2 still apply. printf turns 450 cents into $4.50.

A function is names for the arguments#

A shell function does not declare parameter types. It receives positional parameters. label latte 450 sets $1 to latte and $2 to 450 for the duration of the call. printf keeps "$1" as one word and formats the cents with integer division. Quotes at the call site matter too: the function cannot unsplit an argument that was split before it started. The function runs in the current shell, not in a subprocess. That is why the next lesson needs local.

1. $1textlatte
2. $2text450
3. printflinelatte $4.50

label() {
  printf '%s $%d.%02d\n' "$1" $(($2 / 100)) $(($2 % 100))
}
label latte 450

latte $4.50

The call sets the numbered names#

label latte 450 runs the function with $1 equal to latte and $2 equal to 450. $(($2 / 100)) is the dollars. $(($2 % 100)) is the leftover cents. %02d prints them as two digits, so 50 stays 50 and 5 would become 05. The line is latte $4.50.

label "oat latte" 450 keeps oat latte as $1 because of the quotes at the call. Inside the function, "$1" keeps it as one value. Drop either set of quotes and the space splits the argument. The function cannot repair a split that happened before it was called.

A function is not a new process#

The commands in label run in the current shell. A variable it sets without local is still set after the function returns. That surprise is the next lesson. For this one, the function only prints, which cannot leak a value into the caller except on the screen.