OpenJDK is the default for Linux and soon to be OSX and most development projects are still targeting Java 6. The innovation happening on the Java platform is not going to come from Oracle.
The Developers
It's not a question of whether one particular language is better, or more "hip", than another; or whether Google could simply run-off and create it's own new language or repurpose someother language instead of Java.
It's the developers Google is after, since it's the developers that will create the value (apps) for their marketplace and give them the competitive advantage over other smartphones & platforms.
There are a ton of Java (and its variants like AS3) developers out there, and Google wanted to launch and gain traction in its new platform with as little friction as possible.
If Google had launched Android as a Golang SDK, do you really think the uptick in new developers coding for Android would have been that great?
Edit: Yep, AS3 isn't Java.
Of course, it's easier for them not to have to, but I'm imagining a possible nice future, even if it's far from the most likely.
But, indulging the leap :), why would you want it to be a new language? Phones are getting fast - why not python or ruby? (Or Scala, Dart, ...)
However, I don't think Java is a bad language for many programs--and doubt it's headed to a retirement home anytime soon. (I'm frankly thrilled with Java for moving some many developers to a "safe" language/environment rather than C/C++ and its buffer overflows :)
(Now, before anyone starts a big project in C# remember you are going to get punished all the way from start to finish. License for OS, for dev environment, no choice of dev environment, when you have bought VS you'll have to add resharper. Then you hopefully come to deployment where once again you have no real choice, Azure or self-hosting on, guess what, MS servers which again has to be licensed. )
If you are a start-up, the right way to do it would by joining the BizSpark [1] program. .Net does provide an excellent development and scaleable environment. This program grants you access to all of its products, including Office, SQL server, Visual Studio and a multitude of other applications/tools. There was a recent post on HN [2] that spoke more to this.
I'm still not sure why Google went with Java over C++, other than thinking "phones are fast enough."
If it doesn't matter because you can use C or C++ for critical parts... well you can do that from Objective-C too, and it's trivial. It's far simpler than working through a JNI bridge.
I'm still puzzled by the decision, given the horror of J2ME and the failure of Java on the desktop. Why not C++?
I agree with you on the garbage collector, though. It really slows down things by thrashing on multilevel caches, despite generational GC. I would really like to perhaps see something that gave you a choice of either reference counting + destructors, or, garbage collection. I can't see how you could make mixing libraries written for one or the other style work with an arbitrary ref-cnt / GC selection, though.
I'd probably miss exception handling in any event, as well.
Damn trade offs. I want it all, and I don't want to pay for it!
While it was painful to move from a nice ISAM product to SQL syntax, it was good to get away from a product that provided all the data integrity features of circa 1999 MySQL (i.e. - "none").
Anyway, it's not dead, but it's definitely looking at a stay in the old programming language home. :-)
The majority of the world's enterprise web apps run on Java. There is ZERO chance that Java is going to be retired anytime soon if you look around at the languages available today. There just simply isn't anything available that can compete with the breadth of libraries or developer base.
The fact is that if Android no longer runs Java it is going to be a big blow no matter which way you spin it.
The scenario was meant to be a bit "over the rainbow". While it's unlikely, people do sometimes switch platforms given motivation. How many people really were itching to learn Objective C before Apple starting pushing it? (you few "NextStep" old timers in back, I'm not talking about you)
Given a way to do user I/O and data persistance on a sane platform, people are pretty good at making do on just about any platform without the "gift" of a billion lines of other code.
Honestly the days of writing a billion lines of Java code were over a while ago. Platforms like Play, Grails etc which are quite light are becoming more common.
Realistically there is that only one other platform exists that they could switch to: Javascript.
Given sufficient motivation, I suspect a team of 20 part time volunteers could create an amazing application framework for a new language (or, newly popular language) in about a year, if the platform was enjoyable and perceived to be something that would actually be used for several years and several possible employers / business opportunities. As an example, the early versions of RoR seemed to come out of nowhere. (not saying we should run Ruby on phones, it seems kind of piggish, just as an example of a few people scratching an itch)
Wasn't the Apple switch from Pascal to Objective C for OSX exactly that kind of "we'll make it worth your while" change-over?
Not to mention that Google's V8 would make a great foundation.
Many so-called Enterprise Java apps often don't work at all unless you've installed the /exact/ JRE version that the particular app supports (Looking at you, Citrix). The fact is, that it was a dying language before Android. Hopefully it will resume dying, and Google will either release something better, or embrace an existing Open Source community like JavaScript, Ruby, or Python. (Edit: Now that I think about it, Google's own V8 JavaScript engine would be well suited for this. It already supports ARM Linux.)
Personally, I was forced to take 2 Java classes in college. The examples in our labs never worked. I had to Google everything in the book to make Java code that actually ran. Swing looks the same as it did in 1996. Everyone at the local JUG was considerably older, and spent most of the time complaining about what a pain SVN was. (A clue: You're doing it wrong.)
Despite all of this, I was still learning to program, and actually sortof enjoyed it. But then Sun got acquired by Oracle, inventors of the only database system that makes me want to kill myself. After that, I started seriously considering other languages. But the real nail in the coffin was when Oracle decided to cockblock Apache Harmony, and Apache stepped down from the Java Committee. For shame, Oracle. After all nice things Apache made for the Java Community. How fitting, that Google used Harmony code to create Dalvik, the Android VM.
Those facts don't make COBOL something you'd want to write apps in. I hear there's a new COBOL DI framework so you can write XML instead of COBOL.
Imagine if:
Google looses the right to use the Java API in Android.
Google makes a conversion tool to convert the Android API to another API / VM format. (maybe based on Parrot???)
"" releases an ANSI / ISO spec for an improved high level language (not just a C++ replacement like Go), and makes the spec open to encourage "safe" adoption.
"" provides reference implementations for mobile (Android) and server (GAE) type uses.
"" creates a subsidiary to sell and support the SDK to encourage corporate adoption. I'm not exactly sure how this last step supports their core business, other than making sure there is a talent pool available. But as long as I'm dreaming...
THAT is a result I'd like to see. Send Java off to the old programming languages retirement home, where it can sit and trade war stories with Visual BASIC and dBASE.