Shulou(Shulou.com)06/03 Report--

After a month of hard work, the automation of an unfamiliar project is finally coming to an end. I used to know what kind of products or projects are suitable for automation and what kind of products or projects are not suitable for automation; and because I have designed a high-end tool, I think I know more about automation. But over the past year, I have experienced two tools that are not successful in design. Recently, I have constantly reflected on how to do automated testing better.

1. Be able to use existing open source tools instead of developing a set of tools from scratch. Because open source tools have been repeatedly studied and adjusted by many people, they will have a much better advantage than starting from scratch in terms of stability, ease of use, or design philosophy.

two。 For projects or products that are not particularly urgent, it is necessary to have a certain understanding of them before doing automation. It is impossible to make good automation tools without understanding the requirements. This year, there are two tasks for automated testing tools, both of which share a common feature: I don't know anything about Business. Automation in this case will have the following problems:

First of all, it takes a lot of time to understand Business

Secondly, list the function points, how many functions need to be done, and worry that they are not complete, so make a list of what can be listed. However, if all is done, then this tool is doomed to be a failed tool. The 80-20 principle has always been followed in software testing, and I think this theory is also applicable to automated testing, that is, 80% of the functions will be used by 80% of the people, so when designing automation, the first thing to consider is how to do these 20% of the functions first. and how to make this 20% easy to use.

Third, even if it is done, it is not easy to use. Because the user's habits are often different from those of the designer, especially when the designer is not familiar with the business and the user's habits.

3. For projects with tight time requirements, keep in mind that they are big and comprehensive, especially if you don't know anything about the project. Design will gradually change with the understanding of the product, gradually optimized, if the beginning is large and complete, first, can not be completed on time; second, the late maintenance cost is relatively high

For example: the first automation tool I designed in this year is UI testing. Because of my previous work experience, I want to design a more comprehensive, one-time tool, but in fact, for unfamiliar products, this idea is wrong. And I thought I had a sense of achievement to integrate all the verification results at that time, but now I want to do case, and the tools are separated from verifyresult, so the work I spent a lot of time on integrating the verification results is in vain.

4. To achieve the separation of data, case (especially case template), testing tools and result verification, it is convenient for later maintenance. Now think about it, the previous company actually did the same, but at that time only knew that the efficiency was relatively high, did not understand this height.

5. Although it is important to integrate into one tool, the input-output ratio should not be underestimated. If the workload of integration is large, then non-integration may be the best choice. There is a moderate degree of everything, and the golden mean is recommended.

6. I think the most important core of automated testing is the ability to design an efficient automation framework, the ability to develop, and the ability to design case.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.