I am actively programming the PalmOS for about half a year now-and went through many beta testing cycles with mostly moderate sucess-as 1.1 versions demonstrate. However, most mistakes made on my end are avoidable-if one only knew how. This article series will cover most of the things I deem important.
Random testing is useless-structured testing rules
Given enough eyeballs, all bugs are shallow-the Linus law
This classic law of open-source programmers sounds reasonable. But: how many testers can you afford in terms of time, management effort and money? Most developers go for 10-20 people, of which about 75/100 never get active. Thus, you are left with five to ten beta testers who will probably all do more-less the same routine tasks and are unlikely to perform stress tests,… .
But-most bugs cant be discovered by just doing a quick overview. You may find quite a few-but there always is a rest risk that some obscure bug manages to slip in.
Compare this to a structured approach. A todo list covers the basic testing procedures, making sure that at least the basic functions are tested on each machine. And once beta testers know the frame of operations, they can develop their own test cases easily.
I know that test cases need to be developed, but trust me, the hour invested pays out!
Tune in soon for the next part… What do you think so far?