It always depends on the definition of OOP. Typical enterprise OOP is grounded of the idea of creating black boxes that you can't misuse. That creates silos that are hard to understand externally (and often internally as well, because their implementation tends to be composed of smaller black box objects). That practice may prevent some misuse but it creates a lot of problems globally because nobody understands what's happening anymore on the global scale. This leads to inefficiencies, both performance wise as well as development wise. Even with some understanding, there is typically so much boilerplate that changing things around becomes extremely tedious.
Actually, I have some similar concerns about powerful type systems in general -- not just OOP. Obsessing about expression and enforcement of invariants on the small scale can make it hard to make changes, and to make improvements on the large scale.
Instead of creating towers of abstraction, what can work better often is to try and structure things as a loosely coupled set of smaller components -- bungalows when possible. Interaction points should be limited. There is little point in building up abstraction to prevent every possible misuse, when dependencies are kept in check, so module 15 is only used by 11 and 24. The callers can easily be checked when making changes to 15.
But yeah -- I tend to agree with GP that immutability is a big one. Writing things once, and making copies to avoid ownership problems (deleting an object is mutation too), that prevents a lot of bugs. And there are so many more ways to realize things with immutable objects than people knew some time ago. The typical OOP codebase from the 90s and 00s is chock-full with unnecessary mutation.
Actually, I have some similar concerns about powerful type systems in general -- not just OOP. Obsessing about expression and enforcement of invariants on the small scale can make it hard to make changes, and to make improvements on the large scale.
Instead of creating towers of abstraction, what can work better often is to try and structure things as a loosely coupled set of smaller components -- bungalows when possible. Interaction points should be limited. There is little point in building up abstraction to prevent every possible misuse, when dependencies are kept in check, so module 15 is only used by 11 and 24. The callers can easily be checked when making changes to 15.
But yeah -- I tend to agree with GP that immutability is a big one. Writing things once, and making copies to avoid ownership problems (deleting an object is mutation too), that prevents a lot of bugs. And there are so many more ways to realize things with immutable objects than people knew some time ago. The typical OOP codebase from the 90s and 00s is chock-full with unnecessary mutation.