Microsoft seems set to announce the invention of a cool new technology to allow mapping of objects to persistent storage in databases. Whidby, the next iteration of .NET is slated to include ObjectSpaces which looks like an O/R tool. Scroll down for the ADO.NET section. Congrats to MS, but... why did it take so long?
I have had an ongoing discussion for the last several months about Microsoft, Java, Free/Open software, etc. He is an admitted Microsoft fanboy and makes pretty good money being a Microsoft fanboy (more than I do being a Ruby fanboy and Java user) so I cannot fault him for it ;-). Over the course of the discussion I have come to a few conclusions about Microsoft. First, their marketing ability extends into the realm of MS developers and admins. If a technology is not provided my Microsoft it is probably unreliable, irrevelvent, not mature enough for real use, or simple doesn't exist in the view of a lot of MS developers. I think this is accomplished by providing so much material to their developers that there really is no time to realize other tools exist.
O/R mapping is a perfect example of this. In a case study sense (I know this isn't a proven assertion but the ideas have survived discussions and observations with a wide range of admitted "Microsoft Developers," meaning developers developing for MS platforms exclusively not ones working at MS, who I fully believe are some of the smartest people around (more later)). It is an almost mature technology to Java developers having been born (in the minds of the body of Java geeks out there, not in terms of the real research on it) in Cocobase and Entity Beans, grown through Toplink, Apache OJB, Hibernate, JDO, and scores of proprietary solutions. Big name consultants writing hardback books tell you to do it and provide pretty UML showing grossly simplified models of it. However, very capable developers on MS platforms have never heard of the ideas. It is bizarre. You cannot escape it in J2EE land (just because of Entity Beans, ick), but I can point at .NET developers who are very skilled, writing very big, very enterprise applications who have never heard of the concept even under differnt names. The big thing to them is "detached datasets."
MSDN, various free mags, sycophantic "developer" trade mags, etc all create an MS-sphere that very effecitvely shields a lot of developers from outside ideas. This is coupled with a background difference between a lot of people in the Java world right now, and people in the MS world. In Java it is very difficult to escape interacting with Free/Open software. There is simply too much, too good, too well supported, and too well priced free/open software available and in use seemingly everywhere to avoid it. There is a strong exploratory and self-instructive current in the free software culture that encourages people to go find new things and try them, and to use silly tools like Google to find them, instead of the MSDN search tool. What is more, once you start trying to explore and learn this stuff you find hordes of smart people who A) want to help you, and B) want to show you scores of other new things. There are also scores of helpful people in the MS developer camp, but they too tend to be insulated from new ideas so the same ideas get recirculated.
This has a whole lot of effects. First, every windows application is built like every other windows application. This isn't completely true, but it is very close. .NET web applications all use aspx, pass around datasets (mostly live so you can update directly back to them), use webforms to handle updating their datasets, etc etc. Every desktop app was built in Visual Studio, uses theit Document-View model, etc, etc. There is a big benefit to this -- developers are much more modular. If you know the standard library for .NET, you know every library you need to know. I exagerate, of course, but it is to a large degree true. The Free/Open background imposed on Java developers inflicts with with craziness like trying new architectures. Very similar applications can be built using classic J2EE ( JSP - Servlet - EJB ), servlet only using lots of servlet containers (go affinity) and OJB as a data access tool, and even probably a Prevayler (cool project!) based implementation now for the bleeding edge types who hate their customers ;-).
I specified a definition for "MS Developer" earlier that excluded the people working at MS itself. MS has the great fortune of having more brilliant people per square foot than any other unclassified organization on earth. Again, I don't have the statistics to back this up, but I'd bet a pint or two on it. MS Research does some brilliant things -- they published articles on Bayesian spam filtering in the 90's, for instance. There is a huge disconnect between Microsoft the Business and Microsoft the Researchers however. The business unit seems primarily focused on monopoly preservation and extension. This necesarily excludes a lot of the brilliant things that their research people come up with from seeing any type of commercialization much less readily available usage. No other company is in a position even close to Microsofts in terms of being able to take computer science and IT to new levels. They have enormous respect in the business market, grudging respect even from their most vocal enemies in terms of technical capability, an amazing brain trust, and the financial resources to be able to ride out bad ideas and still execute the good. I firmly believe that their monopoly is dragging them down in terms of what they could achieve. By having it they have a far too lucrative cow to milk to want to risk anything else. This is a shame.
Previous Topic
