Interface Segregation principle aims to avoid creating useless dependencies between components by splitting groups of cohesive features through new smaller interfaces.
One important benefit of adhering to is the removal of unnecessary compilation time forced when components were tight coupled with unnecessary part of fat interfaces.
But I wonder whether that benefit really exists in Java...
Indeed, in Java, contrary to C++ world for instance, as long as one used method doesn't change its signature, client component can be left untouched even if dependent module(interfaces generally) has grown, avoiding the need to recompile this latter.
So, is the benefit of compilation coupling really significant (maybe even non-existent) in Java?