The enterprise journey to decompose the COBOL banking core into Java — The developer perspective 3/5
Facing the challenge to translate code automatically and not to die in the attempt. Part 3 of 5. — Failure ratio and automatic translation.
2 months have passed since the project started, we have explored several alternatives including trying to track each variable in Cobol to understand the program logic and also we have extracted SQL sentences in each file to fully emulate the original bank transaction despite this some strange feeling continues stalking the development team. We are rewriting transactions and the failure ratio per written code it’s about 15 to 50 lines per 1000 lines of code, so, what would happen if we are making mistakes in the programming of the new transactions or if we are forgetting something about the business logic? and the most important question, should we take that risk?
If only we were able to take holistically the cobol logic and transport into Java, something similar to translating from spanish to english, then we could be able to get closer to our goal, but, could that be possible?.
Yes, it’s possible, there are in the market some tools that promise they can reach that goal painlessly and promise excellent results but the vast extension of them are disgustingly expensive or have lack of support.
Could the team develop a tool able to fulfill our requirements translating automatically code and reducing the pain of our migration?
