Eclipse Foundation Celebrates 10 Years 155
msmoriarty writes with news that the Eclipse foundation is ten years old this week. Although Eclipse was released in 2001, development was controlled by IBM until the creation of the independent Eclipse Foundation in 2004. "According to Eclipse Foundation Director Mike Milinkovich, that's a major reason Eclipse was able to thrive: 'IBM....did an exemplary job of setting Eclipse free ... We became the first open source organization to show that real competitors could collaborate successfully within the community.' He also talks about misconceptions about Eclipse, its current open source success, and what he sees for the future."
Still using it (Score:5, Interesting)
Even though I've owned a copy of IntelliJ IDEA for over a year, I still use Eclipse everyday for Java development. Latest version is great and the extensions available for it make it even better.
Re:Java (Score:3, Interesting)
Java can be pretty quick these days, lots of hard work went into optimizing the runtime.
There are many things a VM like Java can do that you can't do in C++ (although C++ is inherently faster and lighterweight). But it can't optimize virtual calls away like the JVM can.
Giant contribution (Score:4, Interesting)
Eclipse and Java make a bit of a unique pair. Java is massively verbose by today's standards, but it's strict typing and highly declarative approach allows your IDE to do amazing things when it comes to refactoring or code analysis. Then there's the fact that Eclipse is by no means just a Java IDE, but that's just part of its giant eco-system.
Eclipse is one of the reasons I was super sad that Oracle bought Java instead of IBM. IBM at least proved they can make a good product using Java, using its strengths and subverting its weaknesses.
For the haters (Score:5, Interesting)
I've been using Eclipse on for pretty much 10years now and by and large, the tool has been pretty darn soliod. its a memory pig so get over it. I throw 1.5G at the heap and though it rarely if ever gets close to it, the amout of speed it performs mosdt operations is amazing.
There are warts which I find personally lousy (like Mylyn of the built-in profiler, and much of the built-in text validators), but thankfully most of those can be trivially turned off and tweaked to speed up usage even more. With a few choice plug-ins, you can do a lot of the hard lifting without effort.
I've only had cursory usage of Netbeans/Idea, but Kepler is really a dream to use. Note, almost every first few months of a new release are generally ass, and Juno was entirely ass so be warned. Just because one version of Eclipse may be a flake, don't discount the platform.
and yet... (Score:4, Interesting)
Re:Java (Score:3, Interesting)
There are many things a VM like Java can do that you can't do in C++ (although C++ is inherently faster and lighterweight). But it can't optimize virtual calls away like the JVM can.
It's not virtual calls that make Eclipse randomly freeze for ten seconds or more. And I've wasted more time having to hit CTRL+C a dozen times to get it to copy than I ever have in virtual function calls in C++. Or restarting it when it runs out of RAM despite having a ton free on the machine, or runs out of handles because they're not closing something properly because, hey, garbage collection will take care of that, right?
Eclipse is a decent IDE when it works, but I'm sure it would work a lot better if it wasn't written in Java.
Re: Giant contribution (Score:4, Interesting)
Depending on the kind of development you are on, maybe. There's plenty of people moving to shinier things though, mainly due to Java's excessive verbosity and lack of support of functional features. For insance, you see Fortune 500 companies placing ads for Scala developers. And people don't move to Scala because they have nostalgia for the C++ era's compile times. There's plenty of growth out there by other second tier languages who people choose to increase speed of development. And there's of course C#, which actually attempts to evolve at a decent rate.
So while Java is still a very used language in industry, you won't see any language getting any uptake today if they replicate Java's love for boilerplate.
We could also talk about the tools that are often used with Java that just promote the mindless verbosity. We all remember how terrible EJB 1.0 and 2.0 were. But then we got Spring and Hibernate, which are only slightly better than the disease. You can choose between monstruous XML formats with no real type checking, leading to a whole lot of runtime errors, or annotations that are slightly less verbose, and yet are just as prone to runtime errors. You end up needing such high test coverage to double check for those 'helpful' technologies that you might as well have been using a purely dynamic language in the first place: It's not as if the compiler protects you from careless mistakes in annotations or XML files. To offset this, we need an IDE and some complex configuration, raising the bar for building even the simplest application. No wonder people found Rails so refreshing when it first came out.
Re:Java (Score:2, Interesting)
I'd give up the modest goal "work a lot better" and trade it in for "it happens to work a lot".
Exits quickly, but Eclipse is spectacularly slow preventing actual work from getting done. It is like a low FPS video game where the problem is supposed to be your setup, but I've never found a computer powerful enough to run Eclipse at a tolerable speed considering my impatience.
The modern emacs (Score:3, Interesting)
Re:Still using it (Score:2, Interesting)
It's still a sluggish bloated memory hog ...
I wonder how much of that is due to the fact that it's written in Java. Seriously, I don't know but I'm curious. This is not meant as flamebait (though I'm still glad I wore my Nomex undies today).
Whenever I see Java benchmarks, they let a program "warm up" (like it was made of vacuum tubes) before taking benchmark numbers. For things that run many times over, like server side stuff, that makes sense. But what about client side stuff like Eclipse? Does anybody have benchmarks for non-JIT'ed code? I understand you have to run the same piece of code many times (I've heard 10k) before it will be JIT'ed. A similar issue with memory. It seems that in order to run efficiently you need a heap that's much bigger than what you'd need if things were "manually" allocated/deallocated.