Further to this question, what is the philosophy behind adding an IP to the bridge device br0 and not to eth0 that is a part of the bridge. None of the answers to that question satisfied me.
The way to think about a Linux bridge is that it consists of two parts:
Confusingly, they both have the same name.
When you create a bridge and place a physical NIC into the bridge, that physical NIC no longer acts like a NIC, it is, instead, just a network cable (at least, that's the easiest way to think about it). So, since you need to have a NIC to associate an IP address with, you use the "virtual NIC" that's a part of the bridge.
I'd draw some ASCII art, but I suck at that, so you'll just have to visualise it...