There is no doubt birds descended from reptiles: we have good transitional fossils, a host of skeletal characteristics, and so on. Yet a few scientists have wondered: could we have it all wrong? Could some characteristics birds share with mammals (most notably the fully divided four-chambered heart) indicate the birds and mammals had a common ancestor?
As Dr. Darren Naish argues here in his marvelous Tetrapod Zoology blog, we certainly do not have it all wrong, but the idea of a bird-mammal clade (originally called Haematothermia by 19th-century naturalist Richard Owen, who seems to have originated the idea) isn't quite 100% dead. There have been some modern papers published on it. As Naish notes, you have to pretty much ignore all the fossils to support this conclusion. As I note, it does demonstrate that radical papers do get published in the peer-reviewed literature.
There's even a cool drawing here of what a bird-mammal ancestor might have looked like.