Dynamic black-box testing
The code is not in-depth details of software testing methods. Often referred to as behavioral testing, because testing the software in the course of actual behavior.
First of all, learn from the product manual software test object should be the input and output.
Next, start the definition of software test cases. Test Case: Referring to the experimental input, and test software programs used.
Select test cases for software testers most important task. Incorrect choice may lead to too large or too small to test, or even testing the wrong objectives. Accurate assessment of risk, the possibility can not be reduced to near poverty can control the range of successful know-how.
Test basic methods: the test vs test failure
Test: Make sure the software can do at least, not test its capacity.
Failed test: simply to undermine the software design and implementation of test cases, also known as forced error testing. Deliberate attacks on software vulnerabilities.
In the design and implementation of test cases, always the first to pass the test. Before the destructive testing to see whether the basic function of the software implementation is very important, otherwise the normal use of software will be wondering why there were so many bugs.
Common test case is to try to force the software error message. Product specifications may be given functional requirements such as, for this test may fail to pass the test may also be tested. May be both. Do not have to deliberate distinction is important is to find software defects!
Select test case: equivalent distribution
Equivalent distribution: is the step by step to too many (infinite) of test cases is reduced to a small range of equally effective process. Also known as the equivalent division.
Equivalent distribution technology provides a choice of values, which values the systems approach abandoned.
Equivalence class or equivalent test interval is the same exposure to the same target or a group of software defect test cases. In the search for equivalence interval, to keep the software, similar to the input, output, operation divided into groups. The group is the equivalent interval.
Equal distribution objective is to reduce the possible combination of test cases to test software is still enough to control. Chose not to fully test because, it is necessary to take certain risks. If in order to reduce the number of test cases over an equivalent distribution, test the risk will increase. In addition, the division interval is not equivalent to a certain standard, as long as enough to cover the test object on the line.
Software from the data (including keyboard input, mouse click, disk file, print output, etc.) and procedures (executable process, conversion, logic and operation) of two basic elements.
Software testing on the data, that is, check the user input information, return results and intermediate results are correct. Mainly based on the following principles to be equivalent to the allocation of a reasonable reduction of test cases: boundary conditions, boundary conditions, and invalid data.
This article is reproduced from 51Testing Software Testing Network (see text): http://www.51testing.com/html/32/n-7332.html