臺灣博碩士論文加值系統

對於軟體品質保證(software quality assurance)而言, 軟體測試
(software testing)是非常重要的步驟。到目前為止, 已經有數種測試標
準(testing criterion)被發表出來;其中有兩種較有名的標準 : all-
paths criterion 和 all-statements criterion。其他的測試標準都介
於這兩者之間。在取捨測試標準時所考慮的因素是 : 選擇強的測試標準
固然需花費較多的工夫找較複雜的例子(test case),但所得到的軟體也較
為可靠。反之,選擇弱的標準當然軟體的品質亦較差。選擇何種標準完全
要看程式的大小、測試人員的時間與預算而定。為了幫助軟體測試人員選
擇適當的測試技術,本文提出新的測試標準,並對這些標準之間的強弱關係
予以量化。由 Oviedo 所發表的 data flow dependency complexity 和
Bieman所發表的 data dependency graph, 觀念都是來自於 statement
或 block 之間的 data-dependency 概念。這兩份研究都指出 data-
dependency 是造成程式複雜化的重要因素。在這篇論文裡, 我們依據
data-dependency 的概念提出了一系列"測試途徑選擇標準"； 這些標準
包括 :all-PRTP criterion 、all-DRTP criterion 和 BADP crite-
rion。他們是根據 "愈複雜的程式需要愈強的測試" 的觀念而來。雖然有
許多的測試途徑選擇標準被發表出來, 卻很少有人對這些標準之間的差異
作計量性(quantitative)的研究；本論文對此點作了計量性的比較。本文
提供了 "衡量各標準於測試時所需花費的努力 (testing effort)" 的理
論基礎；並且提出一系列的衡量規則(metric)。 不僅對不同測試標準提
供了比較的理論基礎, 而且也是軟體測試人員選擇正確測試方法的重要參
考指標。

Software testing is an important process in software quality
assurance. A number of testing criteria have been proposed. Two
famous testing criteria are : all-paths criterion and all-
statements criterion. All other criteria can be categorized
as falling between these two criteria. There is a trade off in
selecting among existing criteria. The stronger the selected
criterion, the more complicate test cases must be used and the
correctness of software will be enhanced. On the other hand,
the weaker the selected criterion, the lower assurance of
software correctness. It depends on the size of program,
budget, and time constraint to decide which criterion should be
used for conduct- ing a test. In order to help software
engineers to adopt a appropriate testing technique, it is
important to provide a guideline for software developers to
test software. The data flow dependency complexity proposed by
Oviedo and the data-dependency graph proposed by Bieman are
based on the data-dependency between different statements or
blocks. Both of these research showed the data-dependency is
an important factor of program complexity. In this
dissertation, a family of testing path selection criteria
based on data-dependency are presented. They are : all-DRTP
criterion ,all-PRTP criterion and BADP criterion. It is based
upon the obvious that the higher the program complexity the
more intensity tested is needed. Although there are many
methodologies for selecting testing path. However, there is few
research providing scale measurement of the difference among
them. This dissertation also provides a quantitative
analysis for the comparison among these criteria. A theoretical
basis for measuring testing efforts is presented and a number
of testing metrics are proposed. It provides a basis for
comparing different testing criteria. Also it will be a good
guideline for selecting proper testing methodologies.

Abstract (Chinese)
Abstract (English)
Acknowledgement (Chinese)
Contents
List of Figures
List of Tables
Chapter 1. Introduction
1.1 Motivation of this research
1.2 The proposed approach
1.3 Organization of this dissertation
Chapter 2. A Study of Previous Works
2.1 Introduction
2.2 Basis of software testing
2.3 Baxic sruvey
2.4 Control flow oriented testing
2.5 Data flow oriented testing
2.6 Program dependences and iteration loop testing
2.7 Rapps and Weyuker''s a family of testing path selection criteria
Chapter 3. A Family of Data-dependency Testing Path Selection Criteria
3.1 Introduction
3.2 Definition of terminologies
3.3 A family of data-dependency testing criteria
3.4 Comparison of testing criteria
3.5 Summary
Chapter 4. Testing Effort measurement based on Statement Type
4.1 Introduction
4.2 Termilonogies
4.3 Quantitative Measurement between all-edges and all-statements
4.4 Quantitative Measurement between all-p-uses and all-edges
4.5 Testing Metrics and example
4.6 Summary
Chapter 5. Testing Effort Measurement based on Variable-used
5.1 Introduction
5.2 Terminologies and definitions
5.3 Quantitive measurement between all-p-uses/some-c-uses and all-p-uses
5.4 Quantitive measurement between all-p-uses/some-c-uses and all-defs
5.5 Quantitive measurement between all-c-uses/some-p-uses and all-defs
5.6 Quantitive measurement between all-c-uses/some-p-uses and all-c-uses
5.7 Quantitive measurement between all-uses and all-c-uses/some-p-uses
5.8 Quantitive measurement between all-uses and all-p-uses/some-c-uses
5.9 Quantitive measurement between all-du-paths and all-uses
5.10 Testing metrics and example
5.11 Summary
Chapter 6. Testing Effort Measurement based on Definition Affected
6.1 Introduction
6.2 Terminologies and definitions
6.3 Quantitative Measurement between BASP and all-du-paths
6.4 Quantitative Measurement between all-DRTP and all-c-uses
6.5 Quantitative Measurement between all-PRTP and all-p-uses
6.6 Testing metrics and example
6.7 Summary
Chapter 7. Conclusions and Future Works
7.1 Summary and contributions of this research
7.2 Future research
Appendix
References
Biographical Sketch (Chinese)