Hi, currently my organization is in the testing phase of migrating to COBOL 6 (CB6). We've got to the point of running all of our test data through the new version and corrected the errors found through that method. The next phase we'd like to do is execute the logic that hasn't been executed. This task is very daunting as we have programs older than myself and more numerous than the stars! /s ...Our goal is to try and trigger any NUMCHECK issues yet to be discovered by manually reading unexecuted code and creating test cases to now trigger that code. We are aware that doing this for every program and every unexecuted line is impossible, so that leads me to the following questions.

1) Is there a more efficient way to go about doing this?

2) How are other organizations handling the issue?

3) What are the realistic expectations of resolving all CB6 related issues?

This sounds like it's a problem of test coverage: your test suite doesn't thoroughly cover your codebase. There's no way around it but to write more tests if you want thorough coverage. In our experience, most COBOL customers are doing as much testing as they're able and then deploying the code. This isn't really a migration question so much as it is a test question in general, as you likely have code in production currently that wasn't executed during testing. Migration just makes the issue more prominent because invalid data can cause different behaviour. As far as I know, the only real options you have are to deploy and hope you have enough error checking to catch production errors, or write additional tests.

Hi

Were you able to complete your COBOL 6 upgrade successfully? What was the testing approach followed?

We have similar scenario with 200 applications, need to go from v4.2 to v6.2. Most of the applications do not have a Regression test cases / test bed. 

Wondering how to manage the testing activities, especially regression testing part.

Thanks

Selva