June 2014 Summaries

Gilt has adopted Appium as their mobile testing framework and has made significant progress in addressing stability issues that plagued them in earlier versions. The latest release implements the defined part of the Mobile JSON Wire Protocol, providing support for mobile-specific actions, and has improved stability, including fixing XPath-related issues. While some pain points remain, such as error reporting and integration with Xcode workflows, Gilt is overall very happy with Appium's performance and sees it as a crucial tool in their testing process, enabling them to focus on writing more tests and improving the reliability of their automated tests.

The author of the text argues that traditional release testing is slowing down the pace of software releases and increasing cycle time. He proposes a new approach to bridge the test divide by automating large numbers of checks on every build, deploying to production continuously, monitoring traffic and logs for errors, and building the capability to rollback on failure. This approach transforms the role of testing from a manual process to a strategic one that balances risk, opportunity, reward, and time invested in release-testing. The author suggests separating components of the system, developing strategies that change the test effort based on which parts were changed, and using exploratory testing to vary the test approach to balance risk with cost.

The concept of software quality is complex and ephemeral, making it difficult to define. It can be understood as the opposite of what causes unnecessary aggravation in users. The author proposes a dialectical theory of software quality, where opposing imperatives exist between developers, architects, and users. Independent QA engineers play a crucial role in enhancing product quality by identifying points of aggravation that developers may overlook. They use black-box testing to exercise the software's functionality without knowledge of its internals, focusing on user experience. This approach is complementary to white-box testing, which verifies individual pieces work, and together they form a dialectical relationship. The author argues that even in Agile development with continuous integration and deployment, independent QA remains essential for defining and executing user scenarios, identifying usage issues, and serving as an "entry point" for automated tests.

In Madrid, Spain, Matthew Heusser, a software delivery consultant and writer, attended the ExpoQA conference where he was surprised by the culture shock of speaking English in a predominantly Spanish-speaking audience. He observed two major categories in testing: day-to-day feature testing and release-testing after code completion, which is often compressed to reduce time. This compression can slow down delivery. He also noted a significant gap between legacy teams that have not modernized and newer teams starting from scratch. The use of techniques like DevOps can help bridge this divide. In his tutorial on lean software testing, he emphasized the importance of reducing batch size and variability in work to improve flow and reduce queues. He also discussed performance testing types, including front-end and back-end testing, synthetic monitoring, and application usage patterns. The panel discussion touched on topics such as whether testers will need to learn coding skills, how to convince management of QA's importance, the future of testing, and the lack of testing education worldwide. Matthew highlighted the limitations of certifications like ISTQB, which he believes can only teach a few technical tools in a short period.