software visualization

This is M and A mania ala RJR Nabisco and AOL/ Times Warner. This is going to destroy Sun’s legacy and cost IBM big time.

The only reason IBM would buy Sun is to acquire rights to Sun’s patents and then wield them despotically in a way Sun never did. The long story there is this will accelerate the movement to simply ban software patents the way business method patents were all but disposed of by the Supreme Court recently.

What does Sun have that IBM wants? This is not about a smaller company being bought for their technology. It’s about a larger company buying a weakened rival in order to kill off it’s competitor. This is HP and Compaq / Carly Fiorina II.

Sun could continue as it is. It has the cash. It has the vision. The future of computing / cloud based applications, PaaS SaaS is on its side. If anything, they’ve been too far ahead of their times.

If IBM buys Sun you can bet that developers will desert Java en mass since IBM has their “own” VM just they way they have their “own” GUI toolkit for Java. IN the software arena, IBM has the worst case of NIH ever seen. Their developers are convinced they can do everything “better”, with marginal results and more importantly, the creation of discontinuities in technology development and adoption. IBM broke the Java the GUI community into two camps for NO good effect when it introduced SWT. So also with Eclipse, which is a poor imitation of IntelliJ and NB. The cultural differences between IBM and Sun’s developers are where the rubber will meet the road on this M and A. Developers aren’t so many thinking cogs that you can shuffle around from company to company, like other “assets”. They’re people with a POV and an attitude about what they do. That is, to the extent they’re any good at what they do. To the victor, IBM< will not go the spoils.

Here’s a dose of reality to all my friends on Wall Street and in management at both companies. Beneath the level of anything any analyst can see or quantify, there’s little tiny social and psychological micro-events that determine how the knowledge that’s in the heads and practices of Sun’s employees - which is what gives Sun its real value - gets “transferred” (or not) to IBM. So you bought the company. So what. trust me, you didn’t buy the developers. Quite the opposite. IF you think you can walk into any part of Sun’s IP, excepting the patents, and take ownership of it, you and Wall Street have a big surprise coming. It’s not under your control, and it never was and it never will be. The culture clash between the Sun way of doing things, egalitarian, optimistic, inventive, forward looking and social is going to slam head on into IBM’s well documented culture of manipulation, mean spirited employee relations, exploitative relationships with its customers (billable hours), aggressive and opportunistic use of the broken IP system (Phelps), divisive, conceited and rank-abusing management hierarchy, forced rankings among employees where the bottom 10% are automatically fired, etc. etc.

Prediction- the best of Sun’s employee’s flee to Google and Adobe, the rest foot drag and passively resist their new-found hell, IBM destroys zfs, Netbeans and other middleware products, Swing and finally Java itself through a combination ineptitude , alienation of key developers, grandiosity and conceit and when it’s all over a huge amount of really good stuff simply no longer exists, the market is poorer and the forward momentum of software development is set back by 15 years.

I think Dmitry's response is right on; it's called picking one thing and doing it well. Who do you think is going to best service the creation-tools needs of Java developers, a company dedicated to doing that and only that or a company whose product is living off the money/time/attention froth of their other income streams? IntelliJ is the best IDE out there and the price is literally 1/10th of what the giants of the past charged for their less-than-stellar products. There are lots of plug-in platforms out there for everything from geenral programs (pico-container, nano-container) to image processing (ImageJ). Don't believe for a second that everything is going to be hosted on a generic platform like Eclipse in the future. Specialized plug-in architectures are more efficient just like specialized drag cars are more efficient; they are not meant for everything- they're meant for ONE thing.

Originally posted by Dmitry Jemerov:
Hello Dylan,

Indeed, Eclipse is a platform, which has upsides and downsides for Java developers working in Eclipse. Indeed, there are many other products that are built on the Eclipse platform. But this also means that Eclipse JDT is just one of many tools running on the platform, and the needs of the Java support are not always top priority for the platform developers. On the other hand, IntelliJ IDEA focuses on being a Java IDE, and we try to provide the absolutely best Java development experience with no compromises.

We do not have any plans to make IntelliJ IDEA a platform. This doesn't make business sense for us and won't help the users either - all the companies which have already invested in the Eclipse RCP won't suddenly switch simply because another platform became available.

Also, it is not feasible for us to support Eclipse plugins in IntelliJ IDEA. The architecture is simply too different - we would have to spend lots and lots of work reimplementing the Eclipse APIs over our own codebase.

It is important under the following circumstance- a very very large number of object are anticipated, say a million. The design of the class of these objects then becomes very important. Making them too heavyweight with many ArrayLists may cut the number of objects the VM can handle by 1/10th or more. However, sparing the member of type ArrayList at every opportunity may create a speed issue with respect to searching (for items that would naturally be in those ArrayLists). Worse, either performance issue might not surface until a customer uses the software in an unanticipated field requiring unanticipated performance and or memory demands.

A good thread here. It sees as though you ought to be able to get an upper limit by tracing the graph of the class of ArrayList the way a GC would and just totaling up everything you encounter.