object databases

I've owned this book for 3 days and cannot put it down! I have not been a big fan of O'Reilly over the years, but this book is "incredible". It goes into the detail that most books seem to omit and it's those details that we "developers" desperately need.

I would recommend this book to ANY and ALL .NET developers who are writing code that hits against a database. There are topics covered in this book that Microsoft's own MSDN and VS.NET help system seem to fail at properly explaining. It's because of this that I've found myself (all too often) going to codeguru.com and google groups to get answers to ADO.NET questions that this book actually covers.

My hat goes off to authors Matthew MacDonald and Bill Hamilton on a job well done.

Coming from a Visual Basic background and now working in VB.NET and C#.NET, this book should satisfy both the VB.NET and C# developer.

This book is classic O'Reilly. It's separated into three sections. The first being a thorough but brief introduction to all aspects of the API. The second section is an API reference. And the third a quick reference. This third section is included on the CD that comes with the book and will integrate into Visual Studio.

Don't expect the first section of the book, which is an introduction to ADO.NET to give you a gentle introduction to the subject. That's not the Nutshell form. If you don't know ADO at all you will want to buy both this book and an introductory book. If you know related APIs, or you know ADO.NET and you need a refresher or have weak spots you will find some new things in the first section. For me it was the support for disconnect access and also the integration with the XML features of SQL Server 2000.

Although this book stays true to the Nutshell form it is a little longer in the introduction than the usual. The introductory section is seventeen chapters and is almost half of the length of the book. So if you are an intermediate or advanced engineer I think you could probably learn enough ADO directly from this book without any other introductory book.

The first third of the book discusses, with code samples, how the component parts (connections, data providers, data readers, data sets etc.) fit together. This is interesting.

The next third analyses the classes themselves but can be seen as a help rehash (altho' this is a little harsh since it does go into more detail than the standard help).

And the last third lays out the namespaces themselves and shows the classes within them, tabulating their members without further comment. Not useful.

In the end, I still didn't feel confident enough to develop even a simple application (except with the data grid, as in many samples). Why would I want to merge two data sets, and why do I get duplicates in the data set but not the underlying data base? What exactly happens when I Update, or EndEdit()? When I do a Fill() on a dataset, am I really sucking the entire data into memory?! Is this efficient, what is Best Practice on this?

ADO.NET puts a large, complex, flexible abstraction between the developer and the database itself, and what is needed is someone to show the "correct" way to use all these bits in real-world programs.