S20 tested their proposed technique focusing on shopping cart based applications. In S27 , a fault seeding mechanism was used instead of real-world faults. This is because of a lack of access to real refactored web pages related to visual symptoms implemented in the technique. RQ 4 - What challenges are presented when implementing the tests approaches?

Variable elements detection is one of the most mentioned. It refers to web application regions that are not static. It includes animations, statistics and publicity that change when reloading the website. In S1, S2, S20 , this is handled with a variable region detection strategy. The website is loaded multiple times in the same configuration, and any section of the screen that presents variation is detected.

Then, these sections are discarded properly in the tests. False positive sometimes called as flaky tests refers to positive tests that should have resulted negative, because the site is cross-browser compatible. False negative refers to tests that failed despite the presence of incompatibilities. This issue affects all the testing techniques. According to S6 , information gathered from DOM models may produce false positives.

Therefore, this technique is used together with visual analysis. In S12, S22 , a classification module for potential incompatibilities in screenshots is detailed. Tests are classified in true positives and false positives. This classification is supported by a machine learning technique using neuronal networks. Machine learning is an artificial intelligence branch that allows computers to learn behavior based on empirical data [ 16 ]. Machine learning techniques are used in S6 as well, to build a more precise visual difference detector. In S29 , it is outlined that navigation model analysis is prone to produce false positives and negatives as well.

By crawling the application with web crawlers, non-deterministic actions are ignored. Triggering state changes is related to navigation model analysis.

Web Service Testing Techniques: A Systematic Literature Review

The articles S7, S14 , mention the difficulty of changing states in the navigation model. When the events are numerous, any click can trigger a new state. In S16 , it is highlighted that performing certain actions in different order may conduct to different states. Interactive elements are a recurring issue in the selected studies S7, S Web 2. The need for a tool to be automated is evident in several articles S6, S9, S It is mentioned the high cost associated with the need of a user performing manual actions S6.

However, S8, S9 propose entirely manual testing techniques.

• What Is It?!
• GitHub - clbustos/buhos: Web based systematic literature review software;
• A Systematic Literature Review in Cross-browser Testing.
• nba jam android sd data apk.
• Design and implementation of a web portal on sustainable development in Cameroon.
• virgin mobile head office address.

Unreachable states are related to navigation model analysis. According to S14, S29 , there are states that cannot be reached from the website links.

Design and implementation of a web portal on sustainable development in Cameroon

Thus, the produced model will be incomplete, leaving part of the system untested. Table 7 displays the relation between the most mentioned challenges and the testing techniques discussed. Table 7 Documented challenges by the selected studies related to the testing techniques. Besides the more common challenges mentioned, other issues have been identified. S2 states that is difficult to take screenshots in visual analysis. This article also exposes the existing issues related to the presence of embedded objects when trying to generate DOM models. For S3 , web browsers that implement security measures constituted a challenge when extracting information to create DOM models.

According to S4 , there are cases when an incompatibility is not shown in a DOM model, which leads to false positives. In S12 , it is described how the variation of parameters needed for visual analysis resulted unproductive, since a reduction in false negatives conducted to an increment of false positives. The authors tried to solve this issue with a neural network model, which achieves high precision at the expense of lower recall. Recall is a term used in machine learning, referring to the proportion of real positive cases that are correctly predicted [ 44 ].

Certain CSS properties can cause incompatibilities S16 , because these may not be fully supported by the browser.

Highlights

A given example is the expression property, which only works on Internet Explorer. In S22 , a challenge which may affect any test regardless of the testing technique is discussed: when classifying a potential incompatibility as a real incompatibility opposite of a false positive , there is a certain degree of subjectivity in which critical and non-critical differences may vary on different applications.

They affirm that the proposed tool Browserbite is just for static pages, needing modifications to work on adaptive webpages. Finally, in S27 , the use of visual symptoms is proposed when testing. The challenge discussed is the need for a visual symptom to be independent of any webpage. Due to the big number of similar properties, the authors argue that this idea is doubtful. There are few secondary studies related to the SLR developed in this paper.

Even though they do not study cross-browser testing specifically, they focus on other types of testing techniques on the UI layer. We mention the studies as follows. Two studies on web application testing are presented by Garousi et al. The first one is a systematic mapping, followed by a deep analysis using a systematic review.

• t mobile logo vector download.
• ibooks 2 app free download.
• Buhos: A web-based systematic literature review management software - ScienceDirect.

These studies were conducted by the same research team. Paz and Pow-Sang [ 46 ] present a systematic review to identify evaluation methods which are employed to assess the usability of applications. The preferred method is usability testing.

nails project

In the same context, Al-Ismail and Sajeev [ 47 ] present a SLR on primary studies related to mobile web usability, user experience and usability challenges for mobile web browsing. Mascheroni et al. The same authors also study web incompatibility detection using digital image processing [ 49 ]. Saleem et al.

• Organizing Your Social Sciences Research Paper: 5. The Literature Review?
• GitHub - clbustos/buhos: Web based systematic literature review software.
• free mobile ringtones for my phone.
• ipad mini defender case best buy!

The findings have proven that the major contribution for quality assurance of web services, is done by maintaining compatibility issues, the effectiveness of services and traceability of operations. Garousi et al. This study looked for models, challenges and benefits of software test maturity measured with test maturity assessments TMA and test process improvement TPI.

Therefore, to our knowledge, this is the first SLR which studies cross-browser compatibility testing. RQ 1- Techniques for cross-browser testing: the visual analysis is the most used technique, as 16 studies selected it. Eight of the authors developed tools that applied this technique, four of those used it in combination with other techniques, and the rest of them used it exclusively. Three studies explored and validated this technique without the development of a proper tool. DOM model analysis is the second most used technique, although there is a decline in its use after This technique was always implemented in combination with other techniques.

The navigation model analysis facilitates the testing of entire websites, and it is also used in combination with other techniques. RQ2 - Tools: we classified the studied tools in 1 tools developed by the article's authors 12 tools, Twelve tools were developed and proposed by studies' authors. From these, 7 tools used a combination of testing techniques, while the rest have implemented a single technique. None of the developed tools used static analysis, attribute comparison or heuristic evaluation. Also, 14 commercial tools were listed, although 6 became obsolete or are no longer supported.

RQ3 - Validation of developed tools and techniques: 20 articles provided validation data, from those, only 14 listed a proper number of test artifacts used on the process. Four studies validated a technique and ten validated a tool. The selection of tested sites and web browsers seemed to be based on popularity. Some authors highlight the difficulty in validation due to the need of unavailable website information.