Question

There are two code snippets:

someObject.getBla1().getBla2().performBlah();

And the second one:

bla1=someObject.getBla1();
bla2=bla1.getBla2();
bla2.performBlah();

I am always told to avoid the first one, and use the second one, where as I feel that The second one is just a pain.

What’s the benefit in the second one?

Answer 1

Avoid both forms. The law of demeter is an important guideline when writing software that states

Each unit should have only limited knowledge about other units: only units "closely" related to the current unit.

Each unit should only talk to its friends; don’t talk to strangers.

Only talk to your immediate friends.

You’re violating this law because someObject knows that bla2 has a performBlah method. If it really does need to have this kind of method them move it to someObject and hide bla1 and bla2 from the outside world.

someObject.getBla1().getBla2().performBlah() would then become someObject.performBlah() and the internal implementation of someObject.performBlah would probably delegate to bla1 which in turn would delegate to bla2 to get the implementation. The important thing is to make sure that each object only knows its immediate friends and doesn’t reach out into the wider world (e.g. someObject only knows about bla1, bla1 only knows about bla2).