Sun Gives Up on Java Tools

According to a story published yesterday evening on the ZDNet Web site, Sun Microsystems, Inc. is going to drop Java Workshop and Java Studio. Instead, the article says, they are shopping for an outside company that produces and supports Java tools. NetBeans is mentioned as a possible acquisition, but that's only a rumor at this point.

Comments

Re:Sun still wants to be Microsoft.

[jilles]

You are right about sun using Java as a tool to stimulate hardware sales but everything else you say is BS.

First Java is getting big on the server side. Suns vm is particularly good in running server side stuff. They are also pushing it on the embedded machine side (again good for hw sales).

Your qualification of the sunray as a Java thin client is not correct. The new SunRay1 machines do not include a clientside vm. They don't need to because all they do is display the output of apps (possibly Java apps) on the server side.

"but I stopped supporting Sun *long* ago"

That is no excuse for spreading ill informed rumours.

IDE?

[Nabuchodonosor]

Emacs, JDE and JDK. do we need something else to write and compile java code? no.

Move shows Sun's trust of Open Source

[Zico]

Way back in December, Sun said that they'd release the source sometime this year under the Sun Community Source License. Well, instead of giving the Open Source community a chance to improve the product and make it work well, they never released the source and are now dumping the whole thing to gobble up yet another company. Seems like if they just would've taken a chance on Open Source, they could've possibly saved themselves a lot of money. Message to Open Source community: Sun obviously doesn't trust you -- think long and hard before you trust them.

As an aside: Java really is a pretty nice language. I can't imagine any other company being such a screw-up when it comes to Java -- they fscking invented it, for chrissakes!

Cheers,
ZicoKnows@hotmail.com

Borland JBuilder is coming for Solaris (and Linux)

[Tom Davies]

I believe Sun arranged for Borland to port their JBuilder IDE to Solaris. This product is now in beta.

Apparently this is a 100% Java product, so it is also expected to be available for Linux, assuming the Linux JVM is up to it...

JBuilder on NT is a good product, so if the 100% Java conversion doesn't slow it down too much (and JBuilder 3 is 80% Java already) this is certainly something to look forward to.

Thank goodness for VisualAge

[Em]

The Java IDE has been something of a mess for a while, so it's good to see it get cleaned up a bit. Sensible decision from Sun too (shocker).

[cue plug-my-favourite spout] So this leaves us with Kawa (nice and simple), JBuilder (hideously unstable) and VisualAge (a proper object development environment).

I wish more IDEs for "object" languages took the VisualAge approach and didn't act like Java is really just a different C or C++. It's not! It's a different Smalltalk.

This is why VisualAge's environment, built on the VisualAge Smalltalk product, works so much better than the rest for serious object development. I know a lot of people seem to think it's weird - but I think this is the fence you have to climb over to get away from all that C-style worrying about source files and compilers and all that old stuff.

Just my 0.0^2.

Re:I'm not impressed.

[jilles]

Hmm,

You are right about the tools, not about the language. I've been using Java since the 1.02 version and I always found that IDEs did not support the latest features. Since those often were the features I relied on I often ended up using an editor and a compiler. Over the years I have become convinced that most experienced Java programmers work this way. Perhaps they use an IDE to make editing/compiling/debugging more easy, but they sure don't use an IDE to drag and drop apps together. The latter is done by less experienced programmers. I have not seen the latest generation of IDEs for Java. I heard they are much better at supporting beans and swing which is good for newbies but not for me. I always disliked GUI editors because I always end up changing stuff manually at which point the GUI editor becomes useless.

As for the application you tried to build, a heavy server app, it really depends on which virtual machine you use what the performance of your system will be. Especially the number of threads that is supported varies from machine to machine (and is also OS dependend). I think at this moment you should either use sun's hotspot vm or IBM's vm to get maximum performance.

"Anyway, it's a chicken and egg thing. Better tools mean better apps. The best Java IDE anyone could find was written in Java. That's a very scary thing to think about."

It doesn't surprise me that the best IDEs were Java programs. To test and debug a Java program you need to run a vm anyway. To read out properties and events from a bean, you need to do Java stuff. To run a Java GUI component, you need a Java vm. A Java IDE needs to do all these things at the same time so you might as well implement the whole thing in Java. Doing so also speeds up the development process of the IDE dramatically so that's good. And most developers have fast machines on their desk so performance is not a real big issue.

Re:I'm not impressed.

[jilles]

An additional advantage of this simplicity is that development time is much lower than with c/c++. Because of this development cost is lower and more features can be built into the system.

C/C++ is nice if you need to deal with hardware directly since it gives a lot of flexibility in this area. The same flexibility backfires if you apply these languages in domains where hardware interaction is not needed (for instance an object oriented ecommerce server application). Generally an expert C/C++ programmer is needed to deliver clean code. Expert C/C++ programmers are scarce and it is a waste of their skill top let them chase stupid errors in pointer arithmetic. You could benefit more from their skill if they applied their knowledge to the actual problem that needs to be solved: how to meet the requirements efficiently. Java allows this. because of its simplicity even newby programmers can deliver nice applications. Expert Java programmers can be very productive in Java.

When I started my CS study, the language we had to learn was C. I had to learn the hard way about pointers and stuff like that. Then Java came along. Within weeks I was producing programs more quickly then I ever could in C. Generally you need only a fraction of the LOC needed to implement the same stuff in C.

"I cannot see any advantages in using Java (beside of attracting programmers that cannot handle pointers and dynamic allocation :)) ) "

Boy you must be blind. Perhaps handling pointers and static allocation is special hobby of you (it bores me since it slows me down). You should realize that each time you spend time on that you are wasting time because these are tasks that can and should be automated (like Java does). The only reason not to automate them has always been performance. And that argument is becoming of less relevance with each new release of Java. The garbage collection algoritms in the current generation of VMs are pretty good and the performance penalty of using dynamic allocation is not so big as it used to be. Further improvements in that area are on the way.

Re:most die-hard Java supporters...

[jilles]

Its even more fun to watch c programmers become obsolete like Cobol programmers are these days. In a few years there will be a huge amount of legacy of code written in C/C++. It will no longer be the default choice for most programming tasks with the sole exception of very low level hardware stuff. Of course legacy code needs to be maintained so there will be plenty of jobs for these fellows.

Java has been closing the infamous speed gap during the past few years. Most whining about Java performance in this tread seems to be obsolete. Guys this is slashdot, news for nerds stuff that matters. You should know better. True some pioneer projects like corel's wordperfect for Java failed miserably, but there are commercial and succesfull Java programs all over the place nowadays (or millions of Java programmers are wasting their time each day). Netbeans, which is mentioned in the article is a nice example.

As for myself, I think I can counter the claim about not knowing about other technologies. I programmed a lot of languages ranging from logic languages like prolog to functional languages like Gofer and Lisp. I also did C, Pascal, basic, C++ and smalltalk. I even spent some time with scripting languages like Javascript and perl (puke, if you've seen a decent language before you just have to dislike it's syntax).

Makes sense.

[dkh2]

Sun has a clue that Microsoft hasn't. If you're good at developing a language you should do that. If you're good at developing tools you should do that instead. Go with your strength and not what you dream of your strength being. (Unfortunately, it looks like Microsoft's strength is in bullying the competition in all arenas.)

Just as Microsoft got it's true start by developing BASIC (read your history kiddies, that's where Billion-dollar Bill got started), Sun developed Java from the ground up and should stick with the language development aspects of Java. Sun is wise to contract out or otherwise semi-divest themselves of the development of tools for Java.

Additionally, while some would argue that only the people who truly know Java from the inside out, from the ground up, would know how to build the best tools... I don't see the masses beating down doors for the Microsoft development tools. In fact, a fresh set of eyes that is NOT completely steeped in the language development hurdles works without the encumberance of that knowledge. They work on the language in it's existing state of development.

As I heard Fats Waller (Jazz/Blues LEGEND) say in an interview once: "Be what you is."

D. Keith Higgs
CWRU. Kelvin Smith Library

Re:I'm not impressed.

[Masem]

From my point of view, Java is completely obsolete. If you want to build fast and stable solutions, either on the client or server side, you should use proven technology like Perl, ANSI-C, ANSI-C++ (or Delphi/VB on the client side). I cannot see any advantages in using Java (beside of attracting programmers that cannot handle pointers and dynamic allocation :)) )

So, how do you write a function that will print to the system printer in C++ that works across all systems?

Or do anything that involves graphic displays?

The point is while there are differences in the speed and memory requirements of the various VMs, they (At least to be java certified) will all perform the same functionality. Yes, there's still clutter between differences of M$'s VM and Sun's VM, and the like, but that's why Sun insists that only programs that have been tested to be fully compatible to be able to display this logo.

Now, I won't argue that there are C and C++ toolkits that work across platforms. But too many of the times, these toolkits are limited to Unix and WinXX. And they are closed source, so that if you aren't one of the priveledged many, you're screwed.

While Java's Open source-ness is still not for sure, there is nothing stopping anyone from making a new VM for their system, as they know what calls they have to put in.

I would agrue that Java's dream never materiallized... but in the sense of being the windows killer. But Java's found other niches that they are easily killing. Any device with WinCE, for example... Java's footprint is much smaller, and more open and robust.