You're viewing a single thread.
You know, I've always read that COBOL projects still get maintained to this day because the costs of rewriting these projects just are too high. I wonder if there's a cutoff point where maintaining them starts costing more than the rewrite. I just don't see how organizations can justify maintaining these projects without these kind of changes forever.
Mission critical code. There are decades of bug fixes. The biggest cost of rewriting it is a risk of errors in the logic.
I can understand that, the fear of moving and the logic being ruined. I wonder how much modern frameworks could cut down the codebase though
Modern frameworks don't help with business logic corner cases. You would want to carefully analyze the algorithms of the legacy code and rewrite same logic in a new language. Even then, the same logic operators don't work the same in every language (automatic type conversions, truthiness of non-boolean types).
Outside of looking a Cobol once or twice I have almost zero working knowledge of the language. But still this feels like something a transpiler could handle. Or maybe a next gen LLM if direct translation of the source isn't desirable but just the core logic
My state’s unemployment system is still COBOL. They did not have a fun time in 2020.