In the 9CBS software engineering forum, someone asked this question, original post: http://community.9cbs.net/Expert/topicView3.asp? Id = 3749711.
I first rushed in the first time. Later, there were many people express their views on this issue. But I am quite uneven, I wrote a relatively long time. Below is the content I reply.
First interviewer's question, mainly explained two points: First, there is a fatal issue. Second, what is the test personnel? Secondly, the problem of Redvale (Autumn and Autumn), how to "perfect" answer. Let's analyze.
First, fatal problems
To answer how to deal with deadly, first, what is the fatal problem, followed by where the fatal problem is from.
From the language, "deadly problems" this is that it is not rigorous, because there is no recognized standard, everyone may differ from the understanding of lethal problems, and with the different stages in the software process, understanding of deadly problems May be different. From my personal understanding, lethal issues are defects that the software cannot continue to test. For example: crash, data loss, main function group is completely lost, system suspension, etc. Maybe someone else's understanding is different, I will not argue with others.
In a company, there should be a standard for the severity of defects, clearly explain what various defects should belong to what severity, in addition, there should be a special person (test team leader, test director, test manager, etc.) Handling the wrong level of all people, multi-person responsibility is easy to have different standards (especially the defects and work, evaluation, etc. Did the introduction of the program? It turns out that Microsoft also did this, the number of defects and the assessment, and the testers all over the country grabbed the bug.
The source of fatal mistakes is the program, in other words, the program that needs to be tested has problems. From this perspective, fatal errors and serious errors, slight errors, etc., there is no essential difference. Maybe change an angle, slight error can become a fatal error.
Second, the test personnel
This problem is actually not the relationship between testers. The processing of error issues, and the company's system has the most close relationship.
I want to use the test order, I have found the problem, I will fill the form and finally give the project manager.
It is generally very advanced, there are various BUG management tools, so it is very convenient to submit, management, tracking, statistics for defects.
Regardless of the problem of the type of level, the probably process process should be similar to the support of the BUG management tool.
In this case, you will ask, since it is a fatal issue, always should always be different from others. Otherwise, "deadly" is not a waste?
Still old, does the company have developed the processing of lethal problems? If there is, it is very simple, according to the company's process; if not, you will see yourself.
I said above, my personal understanding is that fatal defects are unable to continue testing, such a priority should be the highest, because there is a blocking effect, not processed, the following work cannot be continued. At this time, the test person in charge should be reported, contact the person in charge and the development manager, and finally let the project manager find relevant personnel.
Here, unless the tester is directly affiliated with the development department, it is not recommended to find a programmer directly by the tester. A company, Yue Yue is a bigger thing, if possible, it is best to negotiate between leaders, and the tester is just testing.
Corresponding to this, what progress, market, demand, defect solution, etc., and what is the relationship with the test staff? Maybe you can think about it yourself, but if you don't have a support, you may have a good job in this may have a promising job. Third, the perfect solution
It is a very important principle that there are not many, but at the end of it. The bug in the program is not finished. In a limited time, it is necessary to do an unlimited time, which requires a process of measuring. There is less test, there are too many remaining BUG in the program; too much test, the project resources will be wasted. Since the testing process itself can't be perfect, why is a perfect answer to a specific problem.
Microsoft's classic test topic is just a thing to find something, how should you test, no one demands the perfect answer, can only say that it is trying to do its best to explain.
Said a lot, and I didn't say it, I was bleached.
