Maintainability -- which depends upon legibility -- trumps performance. The article acknowledges (however reluctantly) that the legacy code (A) is the only repository of the business processes and (B) is not legible enough to reverse-engineer. Never mind migration: this is an unacceptable situation right now, and the auditors ought not sign.
I've done migrations off my. Savings were 8+ figures/year, so worth it, but, the application didn't have a lot of development. Several companies tried and failed; in addition to all the issues described, you find random bugs in the source coffee that just happen to work.
Cobol also has goto statementsv and unions, which lend themselves to really hard to follow code.
Maintainability -- which depends upon legibility -- trumps performance. The article acknowledges (however reluctantly) that the legacy code (A) is the only repository of the business processes and (B) is not legible enough to reverse-engineer. Never mind migration: this is an unacceptable situation right now, and the auditors ought not sign.
I've done migrations off my. Savings were 8+ figures/year, so worth it, but, the application didn't have a lot of development. Several companies tried and failed; in addition to all the issues described, you find random bugs in the source coffee that just happen to work.
Cobol also has goto statementsv and unions, which lend themselves to really hard to follow code.
Makes me wanna start working on mainframes. Where do I apply?