Books often have great overviews of topics and can cite dozens of papers in a single section or chapter and the book author typically cites the original discoveries of many others besides himself.
Is it ever appropriate to cite a book alone, or should I always cite the original papers?
Does it matter if my use of the the material is a brief reference to a broad topic I'm building on vs a specific paper?
Is the number of years that have passed, or perhaps how common the knowledge is in the field a factor when deciding to cite the book or the original paper?
Does the answer vary by field? If so, in my case I'm typically working with robotics and computer vision papers.