Replies

java will certainly rule the world on proprietary unix machines and mainframes for a long time. I think the mono project has a chance for steam on linux.

web services are the great neutralizer. The issue will really be the continued growth of Windows servers into the enterprise environment.

.NET, plain and simple, is just a much newer and better architected platform. Microsoft knows software and software development better than anybody else out there and .NET shows a high degree of sanity and useability. The numbers support this improved performance and reduced code base. It will be hard for coders (like myself) to justify the extra effort needed to develop commercial products if it takes so much more time in Java and runs very poorly.

Well I agree and I do not agree. Net is fast, no doubt about it and has a better framework. On the minus side there are too many libraries and the dependability of those libraries in the various combinations are a real headache. It really depends on the nature of the application. Only now are we starting to see large scale, large throughput .Nets systems (and I mean very large, like the NYSE, DOD systems, FedEx, systems like that,) that have very long shelf lives and multi millipon dollar hardware footprints. SO the jury is still out in this case, and designers in this world are very conservative. They are however "thought leaders" in the corporate world so their decisions matter.

On the other hand, the J2EE people have really let that framework get way out of hand - it is much to complicated. It reminds me of what happened to C++ (I realize that a language and a transactional framework are two different things, I just mean this as a metaphor.)

It also will have a lot to do with non technical factors - MS pricing, current plant (in both hardware and software) skill sets, migration strategies - even offshoring strategies these days. People in the coding trenches tend to forget that good large scale architecture takes a in broader range of requirements than just the programming frameworks.

I speak from a experiential bias toward very large systems and a P&L responsible management perspective.

As always, it depends on the task at hand and the situation surrounding it.

Microsoft knows software and software development better than anybody else out there

Well I do not know about that, that is a pretty parochial view of the software industry. Real time programming? In aerospace? And other tools vendors: Oracle? BEA? IBM? They certainly do not have the crowd beat on the innovation end - they have been mostly followers, not leaders. Even there OS expertise is not at the highest level at least from the CS theoretical level (we will see what longhorn looks like, I may have to retract that statement.) Their enterprise integration/consulting business was a miserable failure.

Their strength is higher than middle level competence in a lot of mainstream business areas but are masters of but a few, and they do not compete in whole areas of endeavor or bodies of knowledge. That is their whole stated strategy and it works. Offer integrated, above average value to the average problem and then move up the "value stream," that is what they are all about. The danger is when a firm with a core competency (i.e. Oracle) goes after a particular segment. That is famously what plagued IBM for years, and still does to e certain extent. MS also faces the big challenge of facing the next wave, whatever that might be (my guess is grid computing.)

Software companies typically last about 20 to maybe 30 years as market leaders. Gates is sharp enough to have beaten those odd and has slipped through the noose time and time again. Time will tell if MS as an institution can keep it up. They are a good shop but you have to put them in perspective and view the industry as a whole.