This article announces the arrival of real time Java. Being the premier big systems language, the niggles and flaws in Java have always been writ large - the lack of access to the OS is my particular bug bear due to the dramatic expense of dealing with it in real world systems that need to be reliable. Another big problem in performance systems is getting the garbage collector under control. There's only desultory efforts for that, things like the WeakReferences allow a programmer to code defensively, but they don't give control.
Real time is another area where Java just won't cut it, and not only because of the GC issue. In my experience, real time is hard. It requires a top-to-bottom philosophy, and most after-the-fact reworks of systems end up being near-real-time but not the real McCoy. Still, the features that get added always find a use somewhere, so even near-real-time is useful for expanding the solution set.
Posted by iang at July 2, 2004 09:22 AM | TrackBackJust because the RTSJ spec. was nine years in the making does not mean that real-time Java was not available before: Esmertec, Newmonics and a few other companies have been offering real-time Java solutions for years, which did find takers, including many blue-chip ones.
And in any case there is no single definition of "real-time": this is just marketing.
-- O.L.
Posted by: Olivier at July 2, 2004 11:47 AM> .... Being the premier big systems language,
what sort of big systems do you mean there mate ?
[from the article:]
> This is for building big control systems like power systems and aircraft carriers
God Help us :)
As you know I don't like Java, for the straightforward reason that it is the Government Programming Language. Stanford Universty Networks (the most statist educational institution in the entire world, located in the world's most ultraleftist enclave -- SUN, the same people who want the government to take apart MSFT because it makes too much money) created Java, the "statist programming language". Everything about it is statist, from the general notion that it should work "Everywhere" to the fact that it never works :-)
Mind you, i still don't trust C. Gimme some machine code anyday!
> NASA
that says it all! :)
Posted by: JPMay at July 2, 2004 12:09 PM> what sort of big systems do you mean there mate ?
Well, any, I suppose. Big systems are ... ones with too many lines of code, too many modules and too many people. Big systems are ones that require teams of squabbling programmers. Big systems are ones where it is difficult to understand everything, impossible to cost-estimate big changes, and implausible to audit. Big systems are ones where you cringe when you hear about tricky code, because you know that 6 months down the track, the only programmers available to work on it will be ... average.
Big systems are ones where, when you pull it off, you know you're better than anyone else :-)
Posted by: Iang at July 2, 2004 12:41 PM