No software product and developer base can survive for years with a product up in the air, effectively unusable due to the risk that it won't work at some arbitrary point in time.I see your point, but the issue with MSJVM was always just 'for MS' platform. It begs the question; is Gates/MS to be condemned for creating the only platform that 'no software product and developer base' can long live without being stable and constant and consistent from box to box(yikes. If so, condemn away!) There is irony in that. The markets offer alternatives, including free alternatives, that are steadfast and that have always run only Sun JRT, pure to the last drop. The markets condemn via punishment, not reward, and they do so in total, not just on one issue.
In the specific case of marketing 'not Java on MS' as 'Java on MS', MS rightly paid a penalty for violating an agreement. They took from Sun outside of the agreement with Sun. As well, even if Visual J++ muddied the water for a while (and, not many folks bit on it as well, exactly for the reasons you outline), in general, should MS/Gates be prohibited from trying to offer what they believe(right or wrong, its theirs to try and fail or try and succeed with,) is the best enhancements to whatever they make available on their platform? I totally agree they blew their 'Not Java explicitly targeted to take advantage for MS platform' concept, which eventually became .NET/CLR, but I don't believe they should be prohibited from offering .NET/CLR (Not Java) as an alternative. They should have, and did not, market their initial forays into "Not Java for the MS Platform" as "Not Java for the MS Platform". By being cute with the marketing/packaging, they rightly got punished. But most folks knew what it was and wasn't, and that is why so few folks bit on MSJVM. Its not just RIP because of the lawsuit. It has been supplanted by .NET/CLR, ie, what it should have been all along.
I never looked at MS/DOS/WIN legally as a 'common carrier,' and yet the markets train them to act that way by a series of punishment and reward. I have kernel mode drivers written for NT 3.51 that survived unscathed through NT 4, the WIN2K merge, and through today. Not because of anything I did(in fact, several of them violate several of their 'should dos'), but because of the remarkable commitment to backwards compatibility that always slows down their releases. Yes, I know that doesn't apply 100% across the board, even recently with .NET 1/1.2/3.0, and yet even with that(I imagine drive mfgs will rescue us if/when we are dealing with .NET/25.0), it is possible to side-by-side all of them.
And of course, there is ROTOR/MONO. That hardly seems like MS clinging to the last penny to be squeezed out of the market. What it looks like is, a competition in the arena of ideas. Developers should wake up happy in this world of choices, especially when there are so many levels of interop between pure Java and CLR/.NET.
Your point goes to the entire history of software development/computerdom, not just MS Marketing lurches hither and yon. Little is stable. And yes, I'm glad the WIN9x/Me path died, but that painful history was their attempt at a legacy transition/upgrade path.
20 years ago, I had a DOS based product that did dual independent satellite channel ingest in the background, with independent foreground analysis processing. Two TSRs processing continuous streams from sat channels, and a foreground display/analysis app. It was more than a little fringe. But, I sold a lot of it to our military and to militaries all over the world, because it ran on a cheap platform and performed like much more expensive equipment. Then, MS came out with Windows, and made this much easier to do on cheap platforms. By doing so, the investment in the DOS version immediately became obsolete, but it created other opportunities under Windows(such as, the Windows upgrade.) I don't mean 'creative' opportunities, I might have preferred to toil elsewhere. I mean 'market' opportunities. So, one door closed, a much bigger one opened. That's often what happens with 'change.' There were nothing but more choices as a result of what MS did. I was not prohibited from targeting Free BSD Unix or Apple or later alternatives or even DOS, those choices remained.
You talk about Java like it died. I'm just not seeing that. It's still thriving as far as I can tell. But "a product up in the air?" I really enjoy Ray Kurzweils books, he's got an interesting series on change, the latest "The Singularity is Near." Its mostly empirically focused on hardware change, I think the drag on his premise is software. Its somewhere wrapped up in your observation, which I think is both true and not true at the same time. If it is absolutely true, then no software product and developer base can survive for years, period.
But, maybe they shouldn't, as is. That is kind of implicit in Ray's observations. Maybe they won't, as is.
I'd recommend taking a look at his 'Age of SPiritual Machines" and "Singularity is Near" books. I am not saying they are prophetic, that is a dangerous business to be in. I am saying they offer some insightful observations by a smart guy.
I used to teach a course in the mid 80s at night, in a little local business school. (Visicalc/Lotus 123/DBase/WordStar some intro to simple programming, basic care and feeding computerdom for business folks. ) At the first class, I always said ""Folks, this is all going to change, so don't look at this as purely vocational. Its details, but relax on the details. Look for the forest, not the trees. RTFM the trees. " I used an analogy that was floating around back then about "Kitty Hawk to 747/Moon in about half a century." In some ways(hardware)applies to computerdom. But the software side is way lagging that promise. 20 years later, faster hardware is running essentially the same applications, V15, but there is a new revolutionary knowledge distribution/dissemination channel unfortunately based early on on the concept of GOTO incarnate, fortunately since resolved. It has created tons of new opportunities, and has eliminated others, like all revolutionary change. In balance, it has been a huge plus. But, has it significantly helped this lag in software (lagging relative to advances in hardware) that is keeping us from fulfilling the "Kitty Hawk to..." growth path? In theory, it should help(and does, in many instances)big time, but I don't know, something else is afoot. What it has objectively done is, divert 90% of all US engineering talent to the focused mission of making bitmaps dance in an entertaining fashion, in one way or the other, either directly or in support of. Websites, games, video, simulations, movieFX, TVGraphics... the advertising beast needs to be fed. All those channels need to be filled, 24/7/365. While so much of our resources are busy making bitmaps dance... at least the hardware is getting better.
I'm not complaining. It has created a ton of opportunities. But you got to wonder some time, what are we doing?