Sed Lesson 2 – Execution Model & Addressing

The Sed Execution Cycle

sed processes input one line at a time.

For each line:

Read one line
        ↓
Store it in Pattern Space
        ↓
Execute all commands
        ↓
Print the Pattern Space
        ↓
Discard it
        ↓
Read the next line

The temporary buffer holding the current line is called the Pattern Space.

Each line is processed independently unless advanced commands (N, H, G, etc.) are used.

---

Mental Model

Think:

For every line:

• Does the address match?

• If yes, execute the command(s).

This explains most sed behavior.

---

Addresses

General syntax:

address command

Example:

sed '2s/a/A/g'

Only line 2 is modified.

---

Line Number Address

Modify line 3:

sed '3s/.*/FOUND/'

Delete line 5:

sed '5d'

---

Line Range

Operate on lines 10–20:

sed '10,20d'

---

Last Line

$ represents the last line.

Delete the last line:

sed '$d'

Print only the last line:

sed -n '$p'

---

Pattern Address

Execute a command only on matching lines.

Replace text:

sed '/banana/s/a/A/g'

Delete matching lines:

sed '/ERROR/d'

---

Negation

! means “all lines except.”

Example:

sed '/ERROR/!s/o/O/g'

Every line except those containing ERROR is modified.

---

Multiple Commands

Using multiple -e options:

sed -e '1d' -e '$d'

Or in a script:

sed '
1d
$d
'

Commands are executed in order.

---

Vim Comparison

Many sed commands are identical to Vim Ex commands.

Vim	sed
:%s/a/b/g	s/a/b/g
:5d	5d
:10,20d	10,20d
:/ERROR/d	/ERROR/d

The major difference:

• Vim edits the whole file interactively.

• sed processes one line at a time automatically.

---

Key Takeaways

• sed works on one line at a time.

• The current line is stored in the Pattern Space.

• Addresses decide where commands execute.

• Commands execute sequentially for each input line.

• Most everyday sed usage combines addresses with s, d, p, i, a, or c.