I have to fight the urge to be the "someone is wrong on the internet" guy in the XKCD cartoon when I read stuff like this. I don't think it's being overly pedantic to say that words have meanings, and that meanings are especially important in relational database theory because they are precise mathematical meanings--not fuzzy natural language meanings.
From that perspective this document is borderline incoherent. The authors don't know what a relational database is.
yeah I think I must be missing the "domain" of problem that this is a solution for? If you have relatively non-relational static (more read than updated) data why do you need this to store it? Wouldn't flat files or berkeley db type solutions fit the bill?
Also I am not questioning the validity of the project - more hoping that an advocate can give an example of a foot that this show fits snugly.
We would be interested in having your feedback on the official mailing list where we could discuss your concerns. The problem with the first two chapters is that we're trying to address the needs of a wide range of readers. We wanted to cover the major differences between CouchDB and an RDMS from “10,000” feet.
Again, thanks for the feedback - we would love to have you on the mailing list.
Yeah I'm also not really impressed with this first draft. The first 2 chapters seemed more about telling me why rdbms is better. It's a shame too because CouchDB has a lot of potential that this book isn't doing it credit for. One of the guys working on it gave a really good interview on Twit back in April. Look at twit.tv under FLOSS from earlier this year.
From that perspective this document is borderline incoherent. The authors don't know what a relational database is.