As software gets more complex, it needs to be tested even more vigorously. More so because a bug could mean the loss of huge amounts of money in lost time to market. Is fully automatic testing possible? Unfortunately, that's still a dream, says Dr Alexander Petrenko, an expert on model-based testing, who was in India recently. Still, systematic testing is possible and at the moment it's as close to the dream as it can get. This requires that computers should already be equipped with some data, they should be able to execute it and give the report.
According to Petrenko, the two problems so far have been: the complexity of information, and the lack of standards of correctness. "Even verification can have errors. To automatically correct errors, the tools need more information, which is possible only by manual testing today. Human beings can spot errors based on experience, advice, knowledge etc."