Modern software applications need to run on a variety of web and mobile platforms with diverse software and hardware-level features. Thus, developers of such software need to duplicate the testing and maintenance effort on a wide range of platforms. Often developers are not able to cope with this increasing demand and release software that is broken on certain platforms, thereby affecting a class of customers using such platforms. Hence, there is a need for automating such duplicate activities to assist the developer in coping with the ever increasing demand. The goal of my work is to improve the testing and maintenance of cross-platform web and mobile applications by developing automated techniques for comparing and matching the behavior of such applications across different platforms. To achieve this goal, I have identified three problems that are relevant in the context of cross-platform testing and maintenance: 1) automated identification of inconsistencies in the same application's behavior across multiple platforms, 2) detecting features that are present in the application on one platform, but missing on another platform version of the same application, and, 3) automated migration of test suites and possibly other software artifacts across platforms. I present three different scenarios for the development of {cross-platform} web and mobile applications, and formulate each of the three problems in the scenario where it is most relevant. To address and mitigate these problems in their corresponding scenarios, I present the principled design, development and evaluation of the two techniques, and a third preliminary technique to highlight the research challenges of test migration. The first technique, X-pert identifies inconsistencies in a web application running on multiple web browsers. The second technique, FMAP matches features between the desktop and mobile versions of a web application and reports any features found missing on either of the platform versions. The final technique, MigraTest attempts to automatically migrate test cases from a mobile application on one platform to its counterpart on another platform. To evaluate these techniques, I implemented them as prototype tools and ran these tools on real-world subject applications. The empirical evaluation of X-pert shows that it is accurate and effective in detecting real-world inconsistencies in web applications. In the case of FMAP, the results of my evaluation show that it was able to correctly identify missing features between desktop and mobile versions of the web applications considered, as confirmed by my analysis of user reports and software fixes for these applications. The third technique, MigraTest was able to efficiently migrate test cases between two mobile platform versions of the subject applications.

There has been significant debate, specifically within the African context, regarding the validity of using HCT data as part of routine surveillance data for the HIV epidemic. The use of HCT data in tracking the prevalence of HIV, as well as in estimating incidence rates for HIV, has been applied in some African countries, and may offer opportunities to strengthen surveillance in the Gauteng Province, South Africa. Literature suggests HCT data are biased as a result of the high proportion of repeat testers, where repeat testing may be related to high risk sexual behaviour. (1–8) It has been suggested that HCT data be separated into first-time- and repeat tester data in prevalence or incidence estimations. (9) The aim of this research was to determine if there are demographic and HIV prevalence differences between first-time- and repeat testers, as suggested in the literature. (9) Existing mobile HCT unit data was used from the Foundation for Professional Development (FPD). The data was collected in the Tshwane Metropolitan Municipality, Gauteng Province, South Africa. An observational, cross-sectional study design was applied. A systematic random sample of 400 first-time testers and 400 repeat-testers was drawn and analyzed. The findings of this study indicated an overall 10.0% (n=80) HIV prevalence rate. When compared to the Gauteng adult prevalence (15+) of 14.4%, the study prevalence is lower. (10) When looking at the characteristics of the first-time tester and repeat tester groups, there was an HIV prevalence rate of 12.5% (n=51/407, p=0.0152) in the first-time tester group, and 7.4% (n=29/393, p=0.0152) HIV prevalence rate in the repeat tester group. Although literature suggests that repeat testers are the more at risk population, the finding in this study clearly demonstrates that there is a difference in HIV prevalence between first-time- and repeat testers. When first-time/repeat tester was used as the dependent variable, it was found that females are 0.6 less times likely to be a first-time tester compared to males (OR=0.6, p=0.001). The finding of a difference in HIV prevalence between first-time- and repeat tester groups was consistent with three other studies in Ethiopia, Uganda and Kenya. In these studies, HIV prevalence in first-time testers was slightly higher than in that of repeat-testers. (3,11,12) It was found that there is a difference in the HIV and demographic profile between those who test for HIV for the first time and those who are repeat testers. The perceived risk and vulnerability to HIV plays a heavy role in motivating individuals to test once, or repeatedly. In regards to disease surveillance, this study did not prove that the population that utilises mobile HCT are representative of the Tshwane population. This study highlighted the need to better understand the sub-groups and characteristics of those who test for the first-time and those who test repeatedly for HIV. In conclusion, this study has provided evidence that there is a difference between the HIV prevalence of first-time- and repeat testers. However, there is good reason to doubt that the prevalence rate of first-time testers is genuine. Copyright / Dissertation (MSc)--University of Pretoria, 2013. / School of Health Systems and Public Health (SHSPH) / Unrestricted

Many developers today have difficulties testing their applications on mobile devices. This is due to a number of factors, such as the fact that the mobile phone market has become even more fragmented with the introduction of touch screen technology. Existing software that was designed for traditional mobile handhelds is not necessarily compatible with the newest models and vice versa. For developers this incompatibility increases the difficulty when creating software. Lack of resources for testing the application may lead to the application being limited to either just a specific model or in some cases to only one specific version of the operating system software. Without providing support for a large number of models the product may have difficulty attracting customers, and hence fail to gain the desired market share. The challenge is to find a way to make testing simple, effective, and automated on a large number of mobile devices. To achieve this test automation applicationsare needed and a test strategy must be devised. Additionally, testing is often described as never-ending since testing generally reveals errors rather than demonstrating when errors are absent. Because of this some limitations of testing are justified. In order to limit the scope of this thesis I have selected some of the most appropriate methods for testing, and will only examine these specific methods.  The focus for the testing is not specifically to find errors, but rather to confirm that the product offers the specified functionality. This thesis describes an approach to functional testing of an application for Symbian mobile devices based upon log analysis. Unfortunately, testing applications on mobile devices is still not straightforward, and this thesis does not shed any light upon how to lessen this complexity. However, I believe that both testing and development will be more and more built around use cases in the future. Unfortunately, automation of testing based uponthese use cases will be further complicated by the increasing use of touch screens and physical input (such as gestures). / Idag har många utvecklare problem med att testa sina applikationer på mobila enheter.Detta har många orsaker, exempelvis att den globala mobila marknaden har blivit än mer fragmenterad i och med introduktionen av pekskärmstekniken och de snabba förändringar som sker idag. På grund av de många telefoner som idag finns så finns det ett behov för en automatiserad testprocess då det tar för lång tid att göra manuellt. OptiCall Solutions AB har utvecklat en applikation för Symbian S60 som behöver kunna köra på många olika telefoner. Denna masteruppsats har målet att hitta ett sätt att automatisera testning av mobilapplikationer på olika enheter, mer specifikt enheter som kör Symbian S60.  OptiCaller är målet för testerna. Testmetodologier och verktyg har analyserats och kraven har samlats in på den önskade lösningen. Lösningen består av ett program som kör testskripten direkt på telefonen, mjukvara som analyserar testresultaten och presenterar dem i ett GUI, ett teststrategidokument, samt ett felrapporteringssystem. Med hjälp av dess kan testaren skapa sina egna skript för att automatisera och sedan samla in resultaten för analys. Detta eliminerar behovet för manuell testning och gör testningen effektivare, speciellt när man kör många tester. Analysmjukvaran är även integrerad med Felrapporteringssystemet för att underlätta felrapportering.

The rapid advancement of web and mobile application technologies has recently posed numerous challenges to the Software Engineering community, including how to cost-effectively test applications that have complex event spaces. Many software testing techniques attempt to cost-effectively improve the quality of such software. This dissertation primarily focuses on that of hybrid test suite prioritization. The techniques utilize two or more criteria to perform test suite prioritization as it is often insufficient to use only a single criterion. The dissertation consists of the following contributions: (1) a weighted test suite prioritization technique that employs the distance between criteria as a weighting factor, (2) a coarse-to-fine grained test suite prioritization technique that uses a multilevel approach to increase the granularity of the criteria at each subsequent iteration, (3) the Caret-HM tool for Android user session-based testing that allows testers to record, replay, and create heat maps from user interactions with Android applications via a web browser, and (4) Android user session-based test suite prioritization techniques that utilize heuristics developed from user sessions created by Caret-HM. Each of the chapters empirically evaluate the respective techniques. The proposed techniques generally show improved or equally good performance when compared to the baselines, depending on an application under test. Further, this dissertation provides guidance to testers as it relates to the use of the proposed hybrid techniques.

Submitted by Josias Lima (josias@icomp.ufam.edu.br) on 2018-10-01T22:10:08Z No. of bitstreams: 5 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) dissertação_josias-vfinal.pdf: 2426702 bytes, checksum: 608ad25d6314e5cbb942757762543ca1 (MD5) 313 ATA de Defesa - Josias Gomes Lima (Assinada).pdf: 621386 bytes, checksum: 7ecfd9469b78fb1425568cd0a2e5b066 (MD5) 313 Folha de Aprovação - Josias Lima (Assinada).pdf: 483409 bytes, checksum: 97c624af46d2aef955b7f2f2930fd1ad (MD5) CartaEncaminhamentoAutodepósito_Josias.pdf: 120108 bytes, checksum: fd11c4a58dfbac511b1e7e65ed95b59a (MD5) / Approved for entry into archive by Secretaria PPGI (secretariappgi@icomp.ufam.edu.br) on 2018-10-01T22:11:19Z (GMT) No. of bitstreams: 5 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) dissertação_josias-vfinal.pdf: 2426702 bytes, checksum: 608ad25d6314e5cbb942757762543ca1 (MD5) 313 ATA de Defesa - Josias Gomes Lima (Assinada).pdf: 621386 bytes, checksum: 7ecfd9469b78fb1425568cd0a2e5b066 (MD5) 313 Folha de Aprovação - Josias Lima (Assinada).pdf: 483409 bytes, checksum: 97c624af46d2aef955b7f2f2930fd1ad (MD5) CartaEncaminhamentoAutodepósito_Josias.pdf: 120108 bytes, checksum: fd11c4a58dfbac511b1e7e65ed95b59a (MD5) / Approved for entry into archive by Divisão de Documentação/BC Biblioteca Central (ddbc@ufam.edu.br) on 2018-10-02T12:58:51Z (GMT) No. of bitstreams: 5 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) dissertação_josias-vfinal.pdf: 2426702 bytes, checksum: 608ad25d6314e5cbb942757762543ca1 (MD5) 313 ATA de Defesa - Josias Gomes Lima (Assinada).pdf: 621386 bytes, checksum: 7ecfd9469b78fb1425568cd0a2e5b066 (MD5) 313 Folha de Aprovação - Josias Lima (Assinada).pdf: 483409 bytes, checksum: 97c624af46d2aef955b7f2f2930fd1ad (MD5) CartaEncaminhamentoAutodepósito_Josias.pdf: 120108 bytes, checksum: fd11c4a58dfbac511b1e7e65ed95b59a (MD5) / Made available in DSpace on 2018-10-02T12:58:51Z (GMT). No. of bitstreams: 5 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) dissertação_josias-vfinal.pdf: 2426702 bytes, checksum: 608ad25d6314e5cbb942757762543ca1 (MD5) 313 ATA de Defesa - Josias Gomes Lima (Assinada).pdf: 621386 bytes, checksum: 7ecfd9469b78fb1425568cd0a2e5b066 (MD5) 313 Folha de Aprovação - Josias Lima (Assinada).pdf: 483409 bytes, checksum: 97c624af46d2aef955b7f2f2930fd1ad (MD5) CartaEncaminhamentoAutodepósito_Josias.pdf: 120108 bytes, checksum: fd11c4a58dfbac511b1e7e65ed95b59a (MD5) Previous issue date: 2018-08-31 / FAPEAM - Fundação de Amparo à Pesquisa do Estado do Amazonas / The unit test is the level of software testing by which individual parts of the source code are tested. Implementing this type of test brings some benefits such as reducing failures in existing resources, improving code structure, decrease side effects, and reducing fear of code change (Burke and Coyner, 2017). However, the test activity for mobile applications is time-consuming, causing some developers to choose not to create unit tests. Reduced time makes testing automation a necessity. In this context, this work proposes a plugin to assist developers in selecting components that have a greater value in relation to the cost x benefit of the unit test in mobile applications of the Android platform. To measure the value of cost and benefit of components, the following metrics were chosen: halstead effort (HE), future maintenance cost (FMC), code smells (CS), call frequency (CF), risk of failures (RF), market vulnerability (MV) and business value (BV). The proposed plugin has three main processes: (1) Extraction of static metrics; (2) Extraction of dynamic, market and business metrics; and (3) Execution of the genetic algorithm to select the components to be tested. The proposed plugin called SCoTUAM can be added to the development interface of the Android Studio IDE. In this work two empirical studies were carried out. In the first study, the purpose was to analyze the correlation of the metrics, where the result showed the possibility of using the combined FMC, CS, CF, RF, MV and BV metrics in a multiobjective solution. In the second study, the objective was to analyze the plugin’s effectiveness in selecting components with error compared to the manual selection performed by unit test specialists in Android mobile applications, where the result showed the feasibility of the proposal in assisting the developer in the selection of components for the unit test. / O teste de unidade é o nível de teste de software pelo qual partes individuais do código fonte são testadas. A realização deste tipo de teste traz alguns benefícios, tais como redução de falhas em recursos já existentes, melhoram a estrutura do código, diminuem os efeitos colaterais (side effects) e reduzem o medo da alteração do código (Burke e Coyner, 2017). No entanto, a atividade de teste para aplicações móveis tem o tempo reduzido, fazendo com que alguns desenvolvedores optem por não criar os testes de unidade. O tempo reduzido faz com que a automatização dos testes se torne uma necessidade. Nesse contexto, este trabalho propõe um plugin para auxiliar os desenvolvedores na seleção de componentes que tenham um maior valor em relação ao custo x benefício do teste de unidade em aplicações móveis da plataforma Android. Para medir o valor do custo e benefício dos componentes, foram escolhidas as seguintes métricas: halstead effort (HE), custo de manutenção futura (CMF), cheiros de código (CS), frequência de chamadas (FC), risco de falhas (RF), vulnerabilidade de mercado (VM) e valor de negócio VN. O plugin proposto possui três processos principais: (1) Extração de métricas estáticas, (2) Extração de métricas dinâmicas, de mercado e de negócio e (3) Execução do algoritmo genético para seleção dos componentes a serem testados. O plugin chamado SCoTUAM pode ser adicionado à interface de desenvolvimento da IDE Android Studio. Neste trabalho foram realizados dois estudos empíricos para avaliação do plugin proposto. No primeiro estudo, o propósito foi analisar a correlação das métricas, onde o resultado mostrou a possibilidade de usar as métricas CMF, CS, FC, RF, VM e VN combinadas em uma solução multiobjetivo. No segundo estudo, o objetivo foi analisar a eficácia do plugin em selecionar componentes com erro comparado com a seleção manual realizada por especialistas em teste de unidade em aplicações móveis Android, onde o resultado mostrou a viabilidade da proposta em auxiliar o desenvolvedor na seleção de componentes para o teste de unidade.

This dissertation introduces a hybrid strategy for automated testing of Android applications that combines reinforcement learning and test suite prioritization. These approaches aim to improve the effectiveness of the testing process by employing reinforcement learning algorithms, namely Q-learning and SARSA (State-Action-Reward-State-Action), for automated test case generation. The studies provide compelling evidence that reinforcement learning techniques hold great potential in generating test cases that consistently achieve high code coverage; however, the generated test cases may not always be in the optimal order. In this study, novel test case prioritization methods are developed, leveraging pairwise event interactions coverage, application state coverage, and application activity coverage, so as to optimize the rates of code coverage specifically for SARSA-generated test cases. Additionally, test suite prioritization techniques are introduced based on UI element coverage, test case cost, and test case complexity to further enhance the ordering of SARSA-generated test cases. Empirical investigations demonstrate that applying the proposed test suite prioritization techniques to the test suites generated by the reinforcement learning algorithm SARSA improved the rates of code coverage over original orderings and random orderings of test cases.