Note: 本篇仍在未完成狀態

期中考: 第 5,6,7,8,9 章 (boundary value, equivalence class, decision table, path testing, data flow testing)

Intro.

white & black

測試大概可以分成 白箱測試 (white-box testing) 與 黑箱測試 (black-box testing)

white-box testing 又可以稱為 透明箱測試 (glass box testing)， 因為可以看到內部結構，因此又稱為 結構測試 (structural testing) 或 邏輯驅動測試 (logic-driven testing)

test case adequacy criteria

test case adequacy criteria 討論的是 “需要多少的測試才足夠”， 這可以由涵蓋的 control-flow & data-flow 比例來衡量

  • control-flow coverage (測量可以經過多少 execution path)

    • statement coverage

    • branch coverage

  • data-flow coverage (definition & use & kill)

    • all-du-paths (definition & use)

    • all-uses

    • all-defs + 如果每條 variable 的 dc-paths 都包含的話

    • all-c-uses (computation)

    • all-p-uses (predicate)

    • all-c-uses/some-p-uses

    • all-p-uses/some-c-uses

    • dc-paths (definition & clear)

test case 難寫通常也代表著程式太複雜，因此要降低結構複雜度，再來是避免 non-determinism

Rapps-Weyuker hierarchy of data flow coverage metrics :

                    +-----------+
                    | All-Paths |
                    +-----------+
                        |
                    +--------------+
                    | All-DU-Paths |
                    +--------------+
                        |
                    +----------+
                    | All-Uses |
                    +----------+
                        |
                        |
         |¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯|
         |                              |
+------------------------+      +------------------------+
| All C-Uses some P-Uses |      | All P-Uses some C-Uses |
+------------------------+      +------------------------+
                                        |
                                        |
                        |¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯|
                        |                        |
                    +----------+            +------------+
                    | All-Defs |            | All P-Uses |
                    +----------+            +------------+
                                                 |
                                            +-----------+
                                            | All Edges |
                                            +-----------+
                                                 |
                                            +-----------+
                                            | All Nodes |
                                            +-----------+

P-use

use in predicate

C-use

use in computation

O-use

use for output

L-use

use for location (pointers, subscripts)

I-use

iteration (internal counters, loop indices)

I-def

defined by input

A-def

defined by assignment

Spyder

ANSI C

Dynamic

http://spaf.cerias.purdue.edu/Students/spyder.html

Unravel

CodeSonar

Indus/Kaveri

JSlice

SeeSlice

Programslice

decision-to-decision path (DD-path)

coverage.py

coverage report -m

$ coverageg report -m

Name    Stmts   Miss  Cover   Missing
-------------------------------------
hw1        58      2    97%   49-50
test       21      0   100%
-------------------------------------
TOTAL      79      2    97%

Ch5 - Boundary Value Testing

Ch6 - Equivalence Class Testing

  • weak normal

  • strong normal

  • weak robust

  • strong robust

Ch7 - Decision Table-based Testing

Ch8 - Path Testing

Program Graphs

用 imperative 寫出來的程式的 program graphs 會是

  • directed graph

  • nodes are statement fragements

  • edges are flow of control

DD-Paths

Test Coverage Metrics

Miller’s Coverage Metrics

Todo

DO-178B

Ch9 - Data Flow Testing

  • Define/Use Testing

  • Slice-Based Testing

  • Program Slicing Tools

重點: 變數在哪裡拿到 values、在哪裡被使用

  • unifying structure of test coverage metrics

  • program slice (1979 - Mark Weiser)

define/reference anomalies

檢查以下 issue

  • 只有定義但沒有被使用的變數

  • 變數未定義就使用

  • 變數定義了多次

  • 變數在使用之前就被 deallocate

defining node

usage node

P-use & C-use

Ref