Lecture 07 — Debugging as a Method

Learning objectives

• Reproduce a defect with a minimal input
• Form hypotheses and test them one at a time
• Read error messages as evidence, not noise

Case-based lecture: you attempt, we discuss, then we reveal.

The plan

• Four cases: cs-025, cs-026, cs-027, cs-028
• Each: 5-minute attempt → discussion → reveal → prompt
• Worked examples in between — ask anything
• Break mid-lecture · exit question at the end

Key definitions

• Defect (bug) — code that does not meet its specification - distinct from a mistake you have not yet observed.
• Symptom — the observed wrong behaviour; often far from the defect's location.
• Reproduction — a specific input and sequence that makes the symptom appear on demand.
• Hypothesis — a testable claim about the cause; good hypotheses predict an experiment's result.
• Isolation — cutting the search space in half (or deleting code) until the defect is cornered.
• Regression test — a test that fails if a fixed defect ever returns.

The missing first item: reproduce, hypothesise, isolate

Expected: 4.75 6.75

Cost: Debugging cost, not runtime: one repro, one minimal input, one rerun. The method's value is that it terminates.

The ghost duplicate: aliasing proved by identity

lines = ["a", "b"]
echoed = []
for i in range(len(lines)):
    echoed.append(lines[i])
print(len(echoed), echoed[0])

Expected: duplicated rows with identity True; after the fix the rows differ and the check is False

Cost: O(k) extra per call for the copy - correctness buys the copying cost.

• **Input:** items [2.00, 3.50, 1.25]; receipt printed 5.50
• **Output:** one item skipped (1.25 missing); repro input: 3 items, expected 6.75
• Solve for the worked instance first. Then ask what would change if one
• number, name, or condition were replaced by something stranger.

Pseudocode, flowchart or Python. State one assumption. Compare with your neighbour at minute 4.

cs-025 — discuss

• What did the deciding quantity turn out to be?
• Which assumption shaped your answer?
• After the solution is revealed, the class spends five minutes on this prompt: **"Write the one-line commit message that explains the mechanism, not the symptom."**

Instructor: the reveal, the reasoning section and common wrong turns are in the instructor distribution for cs-025 — not embedded in this deck.

• **Input:** invoice rows; extra row appears only when qty is exactly 10
• **Output:** boundary bug: condition qty >= 10 should be qty > 10; minimal repro: one row qty=10
• Solve for the worked instance first. Then ask what would change if one
• number, name, or condition were replaced by something stranger.

Pseudocode, flowchart or Python. State one assumption. Compare with your neighbour at minute 4.

cs-026 — discuss

• What did the deciding quantity turn out to be?
• Which assumption shaped your answer?
• After the solution is revealed, the class spends five minutes on this prompt: **"Which hypothesis did you rank first, and what evidence moved it down the list?"**

Instructor: the reveal, the reasoning section and common wrong turns are in the instructor distribution for cs-026 — not embedded in this deck.

• **Input:** tasks edited twice; list shows one task twice after edit 2
• **Output:** aliasing: the same list object referenced twice; is-check (a is b) proves it
• Solve for the worked instance first. Then ask what would change if one
• number, name, or condition were replaced by something stranger.

Pseudocode, flowchart or Python. State one assumption. Compare with your neighbour at minute 4.

cs-027 — discuss

• What did the deciding quantity turn out to be?
• Which assumption shaped your answer?
• After the solution is revealed, the class spends five minutes on this prompt: **"Write the one-line commit message that explains the mechanism, not the symptom."**

Instructor: the reveal, the reasoning section and common wrong turns are in the instructor distribution for cs-027 — not embedded in this deck.

• **Input:** defect vanishes when print(i) is added to the loop
• **Output:** candidates: loop mutates the list it iterates; set iteration order; time-dependent state
• Solve for the worked instance first. Then ask what would change if one
• number, name, or condition were replaced by something stranger.

Pseudocode, flowchart or Python. State one assumption. Compare with your neighbour at minute 4.

cs-028 — discuss

• What did the deciding quantity turn out to be?
• Which assumption shaped your answer?
• After the solution is revealed, the class spends five minutes on this prompt: **"Which hypothesis did you rank first, and what evidence moved it down the list?"**

Instructor: the reveal, the reasoning section and common wrong turns are in the instructor distribution for cs-028 — not embedded in this deck.

Common misconceptions

• Editing code before reproducing the defect - destroys the evidence.
• Experiments ordered by suspicion rather than by cost.
• 'Fixing' by deleting the failing feature (the defect remains, the symptom hides).
• Declaring victory on one green test; neighbours must still pass.

Summary & exit question

• Named technique today: Debugging as a Method
• Four cases attempted — methods before answers

Exit question (choose one, 60 seconds)

1. Order the clinic protocol steps.
2. Why fix nothing before the hypothesis is confirmed?
3. What is a regression test for?