Source code walkthrough, similarly to inspection, is a number of processes and error-detection measures used by a code reading group. Moreover, it has much to do with the inspection procedure, but the methods themselves are somewhat different and, in addition, another error detection method is applied.
The code walkthrough, similarly to the inspection method, is an interrupted meeting, lasting 1 or 2 hours in general. The walkthrough team is composed of 3-5 people as a rule. One of the persons does the same work as the mediator in the inspection team, another person performs secretarial functions (writes down all the errors discovered), and the other person serves as a tester.
Assumptions as to who would be the fourth and fifth members diverge. Surely, one of them should be a programmer. Regarding the fifth participant, there are the following assumptions:
- a highly qualified programmer;
- dedicated programming language specialist;
- beginner programmer (to have unbiased and fresh outlook, not burdened by previous experience);
- the person who will ultimately operate the program;
- an expert from some other project;
- somebody from the same programming group as the coding expert.
Do you want to cut development costs? Do you value teamwork, intelligence, speed and clear communications?? If so, employ offshore software testing team to benefit from its unbiased software qa services! Hiring overseas partners greatly influence testing quality, efficiency and cost!
The initial procedure is identical as that of the inspection procedure: participants receive necessary materials in advance, several days before the meeting, that allow them to become familiar with the program. Nevertheless, the procedure in the meeting is other than that of the inspection meeting. Instead of simply reading the program’s text or using a list of errors, the meeting participants “play a computing machine”.
A person assigned to the role of a tester comes to the meeting taking along a small set of paper tests, representing sets of input data (and expected output data) for a program or module. Each test is performed mentally in the meeting. This implies that the testing data are walked through in the program’s logic. Meanwhile, the state of the program (namely, the values of variables) is tracked on paper or a board.
Certainly, there must be a small number of test cases which should be simple in nature, because humans run the program at the speed that is many orders of magnitude less than that of the machine.
Add Comment