AP23 — Testing, Quality, and Development Methods

• Revised / source-checked: 2026-10-21
• Exam basis: 2026 IPA Applied Information Technology Engineer Syllabus Ver.7.2. Later proposal or review material about exam-system changes is not substituted for the tested 2026 scope.

Learning index

PDF pages and printed footer pages use different numbering systems.

PDF/printed	Syllabus location	Scope in this article
67/63	Major Category 4 > Middle Category 12 > 2. Design	unit/integration test design and software quality
71/67	same middle category > 3. Implementation and Construction > (8)	unit tests, black/white box, coverage, equivalence/boundaries
72/68	same middle category > 4. Integration and Testing	software integration/verification, drivers/stubs, regression, etc.
73/69	continuation of same section	system integration and system verification testing
75/71	same middle category > 5. Introduction and Acceptance Support > (5)	acceptance review/testing, criteria, and documents
78/74	Major Category 4 > Middle Category 13 > 1. Development Processes and Methods	development models, Agile, XP, Scrum, TDD, CI, CD
79/75	same section > (3) DevOps	development/operations collaboration, CI/CD, Four Keys, observability, DevSecOps

Shared terminology table

The following table is byte-identical in all three language files.

English	한국어	日本語
test level	테스트 레벨	テストレベル
unit test	단위 테스트	ユニットテスト
integration test	통합 테스트	統合テスト
system test	시스템 테스트	システムテスト
acceptance test	인수 테스트	受入れテスト
black-box test	블랙박스 테스트	ブラックボックステスト
white-box test	화이트박스 테스트	ホワイトボックステスト
equivalence partitioning (equivalence analysis)	동치 분할(동치 분석법)	同値分割（同値分析法）
boundary-value analysis (limit-value analysis)	경계값 분석(한계값 분석법)	境界値分析（限界値分析法）
statement coverage	구문 커버리지	命令網羅
decision/branch coverage	결정/분기 커버리지	判定条件網羅
condition coverage	조건 커버리지	条件網羅
defect density	결함 밀도	欠陥密度
quality characteristic	품질 특성	品質特性
waterfall	폭포수	ウォーターフォール
Agile	애자일	アジャイル
test-driven development	테스트 주도 개발	テスト駆動開発
continuous integration	지속적 통합	継続的インテグレーション

1. Read tests by level and purpose

Let library requirement R-F01 be: “An eligible member can borrow an available book copy.”

Common study level	R-F01 trace
Unit	isolate the state transition and rule in Copy.checkout
Integration	check the collaboration/interface among LoanService + Copy + LoanRepository
System verification	exercise end-to-end borrowing against the system requirement
Acceptance	purchaser/user evaluates acceptance criteria and fitness for business use

IPA presents lifecycle test tasks. These four levels are a useful study mapping, not a claim that every organization uses identical names or boundaries. Verification asks whether the work product was built to specification; validation asks whether it fits intended use and user needs. The perspectives can overlap.

Black-box tests derive cases from specifications and observable behavior. White-box tests use internal control flow or data structures. Neither alone proves the absence of defects. Integration may use drivers in bottom-up work and stubs in top-down work. At system scope, combine functional, non-functional, performance, load, security, and regression tests according to risk.

2. Equivalence and boundaries — worked example 1

For a valid loan period of integer 1..14 days, the classes are <=0 invalid, 1..14 valid, and >=15 invalid. Representative equivalence inputs are 0, 7, 15. A compact high-value set at/just outside both boundaries is {0, 1, 14, 15}. It is a practical set for this specification, not an exhaustive universal set.

3. Structural coverage — worked example 2

Consider short-circuit decision D = eligible && available.

• T1: eligible=true, available=true → D true; both conditions evaluate true.
• T2: supply eligible=false, available=true → D false; available is not evaluated because of short-circuiting.
• Across T1,T2, decision outcomes cover 2 of 2: decision coverage 2/2=100%. Observed condition truth outcomes are eligible T/F and available T only: condition coverage 3/4=75%.
• Add T3: eligible=true, available=false → D false and available false is observed. Condition coverage becomes 4/4=100%; decision coverage stays 100%.

Statement coverage = executed target statements/all target statements ×100; decision/branch coverage = exercised decision outcomes/all possible decision outcomes ×100; condition coverage = observed atomic-condition truth outcomes/all atomic-condition truth outcomes ×100. High coverage is evidence of exercised structure, not proof of correctness or requirements coverage.

4. Quality and measurement

For 18 confirmed defects in 6 KLOC, defect density is 18/6 = 3 defects/KLOC. State the denominator and unit. Comparing products solely by this number is unsafe without context for language, complexity, phase, and counting policy. No single metric equals quality.

The eight product-quality characteristics in the Ver.7.2/JIS X 25010 scope are functional suitability, performance efficiency, compatibility, usability, reliability, security, maintainability, and portability. Do not import a later model into this 2026 scope. Connect risk and purpose to characteristics, criteria, and measurements.

5. Development methods

• Waterfall plans and manages sequential stages and deliverables, while change cost must be considered. Prototyping helps obtain early feedback on uncertain requirements or UI. Incremental/evolutionary models repeatedly expand functionality and design.
• Agile emphasizes short feedback and response to change. Scrum is iterative, incremental, and empirical. Choose methods and controls based on uncertainty, regulation, feedback needs, and risk.
• TDD is the development practice failing test (red) → minimum passing code (green) → refactor; it is not a test level.
• CI means frequent integration with automated checks and rapid feedback, not automatic production deployment. Continuous delivery keeps changes deployable but may retain an approval gate for production. Continuous deployment automatically releases passing changes to production.
• DevOps improves development/operations collaboration, feedback, and automation. Observability, Four Keys, and DevSecOps, which includes security in the flow, are also in scope. No single tool completes DevOps.

6. Original four-choice questions

Q1

Which test level most directly checks collaboration between LoanService and the LoanRepository interface?

• ア Unit test
• イ Integration test
• ウ System test
• エ Acceptance test

Answer: イ (2). ア isolates one unit’s rules. イ is correct because it checks collaboration/interfaces. ウ considers the complete system against system requirements. エ considers purchaser/user acceptance criteria and business fitness.

Q2

For an integer valid loan period of 1..14 days, which is a compact high-value boundary set?

• ア {1, 7, 14}
• イ {0, 7, 15}
• ウ {1, 2, 13, 14}
• エ {0, 1, 14, 15}

Answer: エ (4). ア omits outside values. イ is suitable for equivalence representatives but omits an inside value at each boundary. ウ tests only the valid region. エ contains the point just outside and at each boundary.

Q3

After executing only T1 and T2 above, which combination is correct?

• ア Decision 50%, condition 50%
• イ Decision 75%, condition 100%
• ウ Decision 100%, condition 75%
• エ Decision 100%, condition 100%

Answer: ウ (3). ア misses that both decision outcomes occurred. イ mixes the denominators. ウ correctly gives decision 2/2 and condition outcomes 3/4. エ misses that short-circuiting prevented observation of available=false.

Q4

Which is the best interpretation of 18 confirmed defects in 6 KLOC?

• ア 3 defects/KLOC, with language, complexity, phase, and counting-policy context needed for cross-product comparison.
• イ 0.33 KLOC/defect is the only permitted quality metric.
• ウ 108 defects/KLOC, proving total product quality.
• エ Defect density needs no denominator.

Answer: ア (1). ア has the correct calculation and caveat. イ gives the reciprocal and wrongly claims exclusivity. ウ multiplies instead of dividing and overclaims. エ omits the density denominator.

Q5

What is the correct TDD order?

• ア Minimum code → refactor → failing test
• イ Failing test → minimum passing code → refactor
• ウ Refactor → production deployment → test
• エ Remove requirement → skip test → integrate

Answer: イ (2). ア does not begin with a failing test. イ is red→green→refactor. ウ is not the TDD cycle and production release is separate. エ removes requirements and tests, so it is not TDD.

All questions are original; no IPA past question was used or adapted.

Quick sheet

• Specification/external behavior → black box; internal structure → white box. They complement each other.
• 1..14: equivalence representatives 0,7,15; core boundaries 0,1,14,15.
• Check coverage numerator, denominator, and short-circuiting. High percentage ≠ defect-free.
• Defect density = defects/size. Record unit and comparison context.
• Connect unit→integration→system→acceptance purposes to R-F01.
• TDD is red→green→refactor. CI ≠ automatic production release. Delivery ≠ deployment.

Self-check and 1/3/7-day review

Without looking, (1) map R-F01 to four test levels, (2) write equivalence/boundary inputs for 1..14, (3) calculate T1–T3 coverage, (4) name all eight quality characteristics, and (5) explain TDD and CI/CD distinctions.

• Day 1: recall the terminology and quick sheet; retake Q1–Q5.
• Day 3: change the numbers/conditions and recalculate boundaries, coverage, and defect density.
• Day 7: connect R-F01 from test levels through development methods on one page; retest only missed items.

Official sources and notices

• IPA, Applied Information Technology Engineer (Level 3) Syllabus Ver.7.2 — checked 2026-10-21
• IPA, Syllabus overview/index — checked 2026-10-21
• Manifesto for Agile Software Development — checked 2026-10-21
• The 2020 Scrum Guide — checked 2026-10-21

This material is not affiliated with or endorsed by IPA or any other source organization. It does not guarantee learning, passing, or exam results.