Google could have chosen different different packages, different method names, and different arguments (in number, class and name) and return value types. They could have chosen a different way to allocate methods across packages.

Google could have had a totally different look to the APIs and still retained the syntax of the Java language. OK, the JLS does indeed place a limit on a very few classes & methods (66, wasn't it) - but G could have made an ALS, and defined their own restrictions.

All-in-all, you can't get away from the fact that Google made the choices they did because they wanted the Android programming environment to be as close to a Java programming environment as they possibly could make it. They didn't have to, but they chose to.

If they had chosen to make their API's independently, then this "Android" language would have looked like Java notionally, but would not been so intuitively accessible to existing Java programmers, using their (our) geeky memories to just know what the API methods were. This would have split the Java world if Android had taken off subsequently - the kind of split Sun say they fought to avoid, and may have fought to avoid here too.

But would Android have taken off with a language that was Java in syntax only?

As a Java programmer myself, coming from the C world before that (with only a fleeting glance at C++), I believe Schwartz's answers in this - that the API was key to take-up of the language, and key to they usefulness of Java, but inherently impossible to protect. The same points were true of C.

A lot of the supposedly "difficult thinking" behind the APIs (to me) wasn't exclusive to Sun & Java - Java was merely the pinnacle of development work that went into both C and C++ and probably many other object-oriented languages and toolkits.

I semi-agree with Gosling that Google slimed Sun over the API, but you have to admit that they put a lot of manpower into it. Failure to licence the code from Sun meant they had to do their own full implementation of *whatever* API was chosen - and that resulted in the investment of a lot of time & money.

But while I think Google could have acted better, I do actually believe that Google did the right thing - that they stuck to the APIs but with their own implementation. It feels right that the law should allow the APIs to be a "compatability" item and allowed through the copyright laws - and would class as unproctectable, functional *and* de- minimis.

Overall, it seems that turning a computer language into a successful worldwide phenomenon is a hard task, and not necessarily commercially viable. Getting volumes of programmers really requires you to either open everything up with limit-free licencing (like C or Java) - especially into educational arenas - or by already having a near monopoly on a whole branch of computers (like M$ and the desktop).

[ Reply to This | Parent | # ]